{
  "version": "https://jsonfeed.org/version/1.1",
  "title": "Krizaka Blog",
  "home_page_url": "https://blog.krizaka.com/",
  "feed_url": "https://blog.krizaka.com/feed.json",
  "language": "en",
  "items": [
    {
      "id": "https://blog.krizaka.com/story/our-posts-are-planned-as-data/",
      "url": "https://blog.krizaka.com/story/our-posts-are-planned-as-data/",
      "title": "Our posts are data, and a service publishes them",
      "summary": "One calendar row per post, a status the CTO sets after reading the row, a publisher that claims each post before it calls the network. What runs by itself on 17 November, and what is still done by hand.",
      "content_text": "On 17 November at 14:25:25 UTC (9:25 in Montréal), our publisher wrote a claim to its record, in plain words: “I am publishing this one”. 2.6 seconds later it wrote the second: LinkedIn had answered with the post's id and its link. In between, the call to LinkedIn. This chapter is how we got there. It is written for people who do marketing with technical tools: how a calendar becomes a data file, who may say “this can go out”, how a service publishes on time and never sends the same thing twice, and what is still done by hand. I will say it up front: that morning, the CTO started the 9 am run ",
      "date_published": "2026-11-17T23:00:00.000Z",
      "date_modified": "2026-11-17T12:00:00Z",
      "tags": [
        "integration",
        "ai"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/our-posts-are-planned-as-data-en.955b8b2b41.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 665734
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/our-posts-are-planned-as-data-en.ec3890c358.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 984810
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/what-grafana-sees-from-outside/",
      "url": "https://blog.krizaka.com/story/what-grafana-sees-from-outside/",
      "title": "What Grafana sees of us, from the outside",
      "summary": "Sixteen probes, one rule that alerts, six measures of the Sovkora engine, DigitalOcean for the rest and nothing twice: how we watch Krizaka with Grafana Cloud, and why not Datadog.",
      "content_text": "On 17 November, the sixteen probes Grafana runs against our sites and our dev apps were all answering. This chapter is about how they are written, what happens when one of them fails, and what we left to DigitalOcean. It is written for people who run production: it is about rules and windows, but no stack name, no internal address and no exact threshold. Krizaka has no customer yet. There is no on call team: alerts come to me. So the monitoring had to fit in a free tier and bother me only for a real outage. First, what a visitor gets The host's console tells me a container is running. It does ",
      "date_published": "2026-11-17T22:00:00.000Z",
      "date_modified": "2026-11-17T12:00:00Z",
      "tags": [
        "measurements",
        "integration"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/what-grafana-sees-from-outside-en.94b50be03f.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 605143
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/what-grafana-sees-from-outside-en.1c68286f80.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 972801
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/how-the-interface-is-built/",
      "url": "https://blog.krizaka.com/story/how-the-interface-is-built/",
      "title": "How the interface is built: tokens, the Tailwind preset, palettes, components, the Mac app",
      "summary": "One source for every colour, a Tailwind preset that refuses typed colours, components with a written contract, a Mac app that reads only the folders it is given. For each choice: the diagram, the reason, the cost.",
      "content_text": "The piece on ergonomics shows our screens. This one shows what is under them: where a colour comes from, how Tailwind is wired, how a component is made and how the Mac app reuses the same components. It is written for people who design systems or interfaces: diagrams first, then one trade off table per choice. Everything below can be read in public code: krizaka ui for the tokens, the preset and the components, krizaka rust for the native side, verikora for the Mac app. The figures are from 16 November, @krizaka/tokens 4.10.0. A colour starts in one file A colour of Verikora, Folioviv or Sovko",
      "date_published": "2026-11-17T18:00:00.000Z",
      "date_modified": "2026-11-17T12:00:00Z",
      "tags": [
        "ui",
        "verikora"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-the-interface-is-built-en.ad747d4350.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 721757
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-the-interface-is-built-en.6ba50157ff.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1056100
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/one-local-cockpit-to-run-the-whole-company/",
      "url": "https://blog.krizaka.com/story/one-local-cockpit-to-run-the-whole-company/",
      "title": "The ergonomics of a cockpit: one screen per question",
      "summary": "I run a company of agents from one screen on my Mac. Each view answers one question I ask during the day. I draw them by pen, one by one, with what guides each of them.",
      "content_text": "I run Krizaka with teams of agents. They open tickets, write code, send pull requests through merge trains, and a watchdog keeps an eye on them. I look at one screen: the cockpit, an app that runs on my Mac and nowhere else. In mid November I rethought it around one simple idea: each screen answers one question, the one I ask at that moment of the day. If I cannot say a screen's question, the screen does not belong in the bar. The screens are drawn by pen, from their code, not captured. Each number on a drawing points to a line of the legend. The figures come from the evening of 15 November 20",
      "date_published": "2026-11-16T05:06:34.000Z",
      "date_modified": "2026-11-16T12:00:00Z",
      "tags": [
        "ai",
        "integration"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/one-local-cockpit-to-run-the-whole-company-en.b0aa912e4d.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 687349
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/one-local-cockpit-to-run-the-whole-company-en.11b65f3ad8.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1025786
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/five-ergonomics-choices-on-our-real-screens/",
      "url": "https://blog.krizaka.com/story/five-ergonomics-choices-on-our-real-screens/",
      "title": "Five ergonomics choices, drawn on our screens",
      "summary": "The primary action, the phone, the empty screen, the error, the themes: five choices from Verikora, Folioviv and Sovkora, drawn in pen, with the rule behind each and what it costs.",
      "content_text": "Since October, I have been removing from the screens of Verikora, Folioviv and Sovkora what does not serve the person looking at them. I write each removal down in a decision record next to the code, with the principle behind it and what it costs. I show five of them here. The screens are drawn in pen, not captured: each number on a drawing points to one line of its legend. Éric Tremblay, the person in the examples, comes from our demo cast, not from our customers or our administrators. Choice Kept, rather than Rule Figure Arrival one filled button, rather than a drop zone in the middle recogn",
      "date_published": "2026-11-15T22:00:00.000Z",
      "date_modified": "2026-11-15T12:00:00Z",
      "tags": [
        "ui",
        "verikora",
        "measurements"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/five-ergonomics-choices-on-our-real-screens-en.a3b1a610bb.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 482013
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/five-ergonomics-choices-on-our-real-screens-en.cb3c1d2ca4.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 738923
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/design-choices-for-architects/",
      "url": "https://blog.krizaka.com/story/design-choices-for-architects/",
      "title": "Our design choices, for architects: each one with its official docs",
      "summary": "RabbitMQ over Kafka, PostgreSQL over NoSQL, vectors beside the rows, Next.js in one container, Playwright and Maestro, shared CI, Grafana, shared libraries: the choice, the refusal, why, the cost.",
      "content_text": "This post is for architects, tech leads and designers. It is short on prose on purpose: each decision is one block, with the choice, the alternative we refused, why, and what it costs us. Every claim about a tool links to its official documentation, so you can check it. What we run is read from our code on main , not from memory. The eight choices at a glance The detail follows, in the same order. The view at C1 and C2 Two views of one model written in LikeC4, the notation of the C4 model, checked against our organisation's base model. Less detail than our internal views, and nothing sensitive",
      "date_published": "2026-11-15T18:30:00.000Z",
      "date_modified": "2026-11-15T12:00:00Z",
      "tags": [
        "integration"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/design-choices-for-architects-en.7aad21a410.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 841097
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/design-choices-for-architects-en.902d14ef78.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1308233
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/the-architecture-of-krizaka-agents/",
      "url": "https://blog.krizaka.com/story/the-architecture-of-krizaka-agents/",
      "title": "The architecture of krizaka-agents, for enterprise architects",
      "summary": "One founder, AI agents and the tooling between them: who decides, who writes, who refuses, who merges, and how every session stays within its budget. Five checked diagrams, tables, audit figures.",
      "content_text": "This chapter is for architects and for people who speak IT. The question is simple: how does one person ship software with AI agents without losing control? My answer is not a model. It is the tooling around the models, which we call krizaka agents. It is private: I show the pieces, how they talk to each other and a few details of how they work, not the code. In chapter 31, I promised diagrams drawn from a model rather than by hand. Here they are. They come from one model written in LikeC4, the notation of the C4 model, checked like every system of the organisation. The last section explains h",
      "date_published": "2026-11-15T13:00:00.000Z",
      "date_modified": "2026-11-15T12:00:00Z",
      "tags": [
        "ai",
        "integration"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/the-architecture-of-krizaka-agents-en.198c8cf5c1.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 644298
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/the-architecture-of-krizaka-agents-en.89af475f14.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1082679
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/how-our-agent-teams-talk/",
      "url": "https://blog.krizaka.com/story/how-our-agent-teams-talk/",
      "title": "Inside the machine room: how our agent teams talk to each other",
      "summary": "The CTO chat, the headless sessions, the train, the operations database, the hooks before and after every command. Then four audits in one day: what they found, and what is still open.",
      "content_text": "In chapter 29, I promised three days of conforming starts, then the numbers. The three days end tomorrow evening. I did not wait: on 13 November, I had our agent teams audited four times in the same day, read only, from their logs. On the promise itself, the count is simple. The first three audits read 150 environment checks (118, then 16, then 16). No drift. The rest is less tidy, and that is the part I care about. The audits found 14 bugs, then 8, then 10, then 7. Almost all of them are in our tooling, not in the products' code. To understand them, you first need to see how the machine is bu",
      "date_published": "2026-11-13T17:31:28.000Z",
      "date_modified": "2026-11-16T12:00:00Z",
      "tags": [
        "ai",
        "integration",
        "measurements"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-our-agent-teams-talk-en.d7beacc726.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 1099242
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-our-agent-teams-talk-en.b539677ad0.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1513898
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/why-sovkora-loads-no-plugin-in-the-jvm/",
      "url": "https://blog.krizaka.com/story/why-sovkora-loads-no-plugin-in-the-jvm/",
      "title": "Why Sovkora loads no plugin into its JVM",
      "summary": "An AI engine on Spring Boot 4 must accept extensions. PF4J, child contexts, Spring Modulith, loader.path: what each option costs, what we chose, and what it changed in our plan.",
      "content_text": "Sovkora, our AI engine, will have to accept extensions. Our own first: expert packs, sold separately. Later, perhaps, a partner's. When the question came up, the written proposal was the one you read everywhere in the Java world: a system of plugins loaded dynamically into the JVM, with PF4J or child Spring contexts, in our shared Java library. This post tells why we did not do it. It is mostly for Java and Spring developers: there is code, exact versions and sources. The decision was approved today, after a study, a second independent pass and a counter review of 78 claims, one by one. I keep",
      "date_published": "2026-11-12T20:49:00.000Z",
      "date_modified": "2026-11-12T12:00:00Z",
      "tags": [
        "sovkora",
        "integration",
        "security"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/why-sovkora-loads-no-plugin-in-the-jvm-en.0a2cf4f35e.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 650528
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/why-sovkora-loads-no-plugin-in-the-jvm-en.2949738b7a.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1034553
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/every-agent-starts-in-the-same-world/",
      "url": "https://blog.krizaka.com/story/every-agent-starts-in-the-same-world/",
      "title": "Every agent starts in the same world",
      "summary": "On November 11 we found that each of our agent teams started with a slightly different tool, different rules and a different context. We built one way to start: isolated, pinned, checked.",
      "content_text": "In chapter 26, we tidied everything up: one folder, one copy per repository, our real names. The tidy up said where things live. It said nothing about how an agent starts. At Krizaka, code is written by teams of agents. The CTO, which is an agent too, opens a ticket, writes a short brief and starts a session with no screen. It works alone, opens a PR, then stops. On November 11, I understood that these sessions did not all start in the same world. What we found on November 11 That day the CTO measured instead of assuming. It started a test session with a single question and no tools, just to r",
      "date_published": "2026-11-12T03:00:00.000Z",
      "date_modified": "2026-11-11T12:00:00Z",
      "tags": [
        "ai",
        "integration"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/every-agent-starts-in-the-same-world-en.bde1d975ad.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 788814
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/every-agent-starts-in-the-same-world-en.cb3431fb6d.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 987297
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/agents-under-a-coordinator/",
      "url": "https://blog.krizaka.com/story/agents-under-a-coordinator/",
      "title": "Agents under a coordinator: what broke, and how we caught it",
      "summary": "A coordinator with no model in it, leases that expire, a scope held at four places, crons that wait for an event, alerts in a database. Three failed tests, then a wave that lands.",
      "content_text": "On the night of 9 to 10 November, around 11:30 pm, three teams of agents were working on the same page of my cockpit. Each had a small task: fix the display defects of one screen. They stopped with nothing landed. They had spent their time looking for the code, not changing it. This post is about what we built this week so that agents can work together without stepping on each other, and above all about what broke along the way. I show the failures with their numbers, because that is where I learned something. All times are Montréal time. The problem: the check came after the code Before, an a",
      "date_published": "2026-11-10T08:38:55.000Z",
      "date_modified": "2026-11-10T12:00:00Z",
      "tags": [
        "ai",
        "integration",
        "measurements"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/agents-under-a-coordinator-en.3a701b3b51.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 626182
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/agents-under-a-coordinator-en.5fe2d60ff5.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 993621
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/let-me-show-you-the-machine/",
      "url": "https://blog.krizaka.com/story/let-me-show-you-the-machine/",
      "title": "Let me show you the machine",
      "summary": "The cockpit, the budget, the train. I will show you how I run a company almost alone with agents, and what it cost to learn to do it right.",
      "content_text": "I have already told, piece by piece, how I build this company almost alone, with agents. Today I will not tell it again. I will show it to you. The figures on this page run on invented data, but they move the way the real machine moves, and you can move them yourself. A bird that repeats only what is true At the very top there is a CTO. It is not me: it is Mouloud, a Gabonese grey parrot. I gave him that role because a grey parrot does one thing very well: it listens, it remembers, and it repeats only what it heard. Mouloud pilots the work, decides what is delegated, and pushes me to change my",
      "date_published": "2026-11-08T17:00:00.000Z",
      "date_modified": "2026-11-08T12:00:00Z",
      "tags": [
        "ai",
        "integration",
        "costs"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/let-me-show-you-the-machine-en.301d407586.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 565349
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/let-me-show-you-the-machine-en.602083442e.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 927070
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/the-big-clean-up-and-our-real-names/",
      "url": "https://blog.krizaka.com/story/the-big-clean-up-and-our-real-names/",
      "title": "The big clean-up: one folder, one config and our real names",
      "summary": "On 6 November we stopped everything to tidy up: 1.1 TB of agent clones, paths typed by hand, two names for every product. One folder, one local config, our commercial names.",
      "content_text": "In chapter 25 I told why we were slowing down, and what we changed on 5 November. The next day, 6 November, I wrote the CTO an even shorter sentence: “we stop everything now”. This time it was not about speed. It was about order. A first human contributor was joining us, and he had to be able to understand the organisation on his own, without asking me where things are. So nothing would move before everything was tidied and named. This chapter tells the day of 6 November: what we found, how we deleted without losing anything, the one folder, the one config, our real names, and how we work now.",
      "date_published": "2026-11-06T18:00:00.000Z",
      "date_modified": "2026-11-06T12:00:00Z",
      "tags": [
        "ai",
        "integration"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/the-big-clean-up-and-our-real-names-en.1ce9a28821.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 1061816
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/the-big-clean-up-and-our-real-names-en.443b9d4137.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1361656
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/why-we-slowed-down-and-how-we-work-now/",
      "url": "https://blog.krizaka.com/story/why-we-slowed-down-and-how-we-work-now/",
      "title": "Why we slowed down, and how we work now",
      "summary": "Our product features fell from about forty a day to fewer than five, while the pipe stayed fast. Five measured causes, our 87 rules reviewed, the crack who replaces the firefighter.",
      "content_text": "In chapter 23 I told the story of the rails, the firefighter and the chouf. In chapter 24, how we stay in sync. On the morning of 5 November I wrote the CTO a less proud sentence: “I have the feeling we are not moving at the speed we want”. I asked him to study our speed and how we work with the agents, and to give me a plan. This chapter tells what he found, what we decided the same day, and why. It ends without a result, because the first measure falls on 8 November. A feeling, then a measure The report was made read only: no issue, no PR, no setting touched. It counts the PRs of 41 reposito",
      "date_published": "2026-11-06T02:30:00.000Z",
      "date_modified": "2026-11-05T12:00:00Z",
      "tags": [
        "ai",
        "integration",
        "measurements"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/why-we-slowed-down-and-how-we-work-now-en.cf6a514595.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 1017350
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/why-we-slowed-down-and-how-we-work-now-en.699e31f1d8.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1319509
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/how-we-stay-in-sync/",
      "url": "https://blog.krizaka.com/story/how-we-stay-in-sync/",
      "title": "How we stay in sync: the founder, the CTO, the business team and the lanes",
      "summary": "A business and design team that writes before anyone builds, lanes that deliver one issue at a time, a consultant at the turns, and the places where we meet. One idea followed end to end, a go or no-go, and what failed.",
      "content_text": "In chapter 17, I told how the CTO agent and I work: who decides what, the lanes, the board. There were two of us. Since then the table has filled up. There is a business and design team, a firefighter and a lookout, and even a consultant. The more people there are, the simpler the real question is to ask and the harder it is to hold: how do we stay in sync? This chapter answers with a single day, November 3. That day I asked for videos for our sites' carousels, and the same evening we decided not to put a paid product in production the following week. Both stories go through the same places. W",
      "date_published": "2026-11-04T04:45:00.000Z",
      "date_modified": "2026-11-03T12:00:00Z",
      "tags": [
        "ai",
        "integration"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-we-stay-in-sync-en.54664aaf94.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 767882
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-we-stay-in-sync-en.f5c101be5d.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1001572
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/the-trains-second-version-rails-sidings-and-a-repair-train/",
      "url": "https://blog.krizaka.com/story/the-trains-second-version-rails-sidings-and-a-repair-train/",
      "title": "The trains, second version: rails, a siding and a repair train",
      "summary": "Green PRs that never arrived, a red main that blocked everyone, a repair our own tool refused, and a session that died without a sound. How our trains run now, without a single token.",
      "content_text": "In chapter 19 I told the story of our first trains: a small script that lands pull requests one by one, without a model token. It held for a week. From 1 to 3 November it broke in four ways I had not seen coming, and I rebuilt it. This chapter shows the new version with drawings of trains, because that is how I see it in my head. One note before we start. Our products go by code names here: Product A, Product B, Product C, and the platform. What matters is the mechanism, not the names. Four moments of the weekend Green PRs that never arrived On 1 November, PRs had been waiting for hours, every",
      "date_published": "2026-11-04T01:15:00.000Z",
      "date_modified": "2026-11-03T12:00:00Z",
      "tags": [
        "ai",
        "integration",
        "measurements"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/the-trains-second-version-rails-sidings-and-a-repair-train-en.7c5a8b6161.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 1108578
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/the-trains-second-version-rails-sidings-and-a-repair-train-en.7ba4b7b7da.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1376188
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/how-orazaka-made-our-brand-film/",
      "url": "https://blog.krizaka.com/story/how-orazaka-made-our-brand-film/",
      "title": "Show your work: how Sovkora made our film",
      "summary": "A 31-second film without a word, made by our own engine. The reference sheets, the key frames, the video, the cut, and the four paid failures that each left us a guard.",
      "content_text": "On October 30, I watched the film on our “Our story” page and thought it sold nothing. It was pretty. It did not say what we do. I asked for another film, with one condition: that it be made by Orazaka, our AI engine. Not by an online service, not by hand. If we say our software shows its work, the film had to show it too. Three days later it was finished, and it replaces the old one on krizaka.com with the site's next release. This chapter tells how it was made, what went wrong on the way, and what each failure left behind. It follows chapter 4, where our first film was beautiful and wrong. T",
      "date_published": "2026-11-01T14:30:00.000Z",
      "date_modified": "2026-11-01T12:00:00Z",
      "tags": [
        "ai",
        "sovkora",
        "costs"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-orazaka-made-our-brand-film-en.5c7375fbe1.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 1014294
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-orazaka-made-our-brand-film-en.a102af879f.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1330038
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/working-with-agents-rules-that-refuse/",
      "url": "https://blog.krizaka.com/story/working-with-agents-rules-that-refuse/",
      "title": "Working with agents: rules that say no",
      "summary": "Our rules were written down, and the agents forgot them. We moved them into the commands: a command that refuses is never forgotten. The lanes, the trains, the guards, and what stays mine.",
      "content_text": "For weeks, our rules lived in a file. A long text every agent session reads when it starts: how to name an issue, how many pull requests may be open at most, who is allowed to merge. The agents read it. Then, a little further into the session, they forgot it. Not out of ill will: two hundred lines of text weigh little against the task at hand. This chapter tells how we moved those rules from the text into the commands. It follows chapter 17, where I told who decides what between the CTO agent and me, and All aboard, about our trains. A written rule gets forgotten, a command that refuses does n",
      "date_published": "2026-11-01T03:00:00.000Z",
      "date_modified": "2026-10-31T12:00:00Z",
      "tags": [
        "ai",
        "integration"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/working-with-agents-rules-that-refuse-en.33c0ba2a68.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 923839
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/working-with-agents-rules-that-refuse-en.7755837d18.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1210865
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/how-our-automation-grew/",
      "url": "https://blog.krizaka.com/story/how-our-automation-grew/",
      "title": "How our automation grew: from one heavy CI to workflows by concern",
      "summary": "In three weeks, krizaka.com's CI went from a minute and a half to over ten. What we added, why, what we got wrong, and how we split our workflows by concern with a speed budget.",
      "content_text": "Three weeks ago, krizaka.com's CI took a minute and a half. Today, a pull request waits a little over ten minutes to learn whether it can go. That is no accident: every added minute is a check we wanted, often after a bug. This chapter tells how it grew, what it taught us, and how we put it back in order. It follows straight on from Tchou tchou. The trains land pull requests one by one, but a train never runs faster than the CI it waits for. A CI that grows At first, krizaka.com had its own CI: one file, a few steps, lint, tests, build. Then, on 21 October, the site moved onto the organisation",
      "date_published": "2026-10-31T14:00:00.000Z",
      "date_modified": "2026-10-31T12:00:00Z",
      "tags": [
        "ai",
        "integration",
        "measurements"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-our-automation-grew-en.c78a67609b.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 806017
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-our-automation-grew-en.55621f1e04.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1139711
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/tchou-tchou-how-our-pull-requests-catch-the-train/",
      "url": "https://blog.krizaka.com/story/tchou-tchou-how-our-pull-requests-catch-the-train/",
      "title": "All aboard: how our pull requests catch the train",
      "summary": "A hundred and forty open PRs, sessions waiting on CI, and a small token-free script that lands them one by one. What broke, what we kept, and the rules that came out of it.",
      "content_text": "This week I understood that my agents were not short of speed. They were short of a platform. Every session finished its work, opened a pull request, then waited for CI to pass, for the merge queue to accept it, for the branch to be up to date. While it waited, the session kept costing, and the pile kept growing. This chapter tells the small system that got things moving again. I call it the train, because that is exactly what it is: wagons leaving one behind the other, never overtaking. The problem: a pile We work with five lanes: one session per application (Yomizaka, Akinaza, Orazaka), one ",
      "date_published": "2026-10-31T02:30:00.000Z",
      "date_modified": "2026-10-30T12:00:00Z",
      "tags": [
        "ai",
        "integration",
        "measurements"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/tchou-tchou-how-our-pull-requests-catch-the-train-en.c13e0b898c.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 600911
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/tchou-tchou-how-our-pull-requests-catch-the-train-en.c676fa88fa.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 916827
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/how-we-test-our-uis/",
      "url": "https://blog.krizaka.com/story/how-we-test-our-uis/",
      "title": "How we test our UIs",
      "summary": "Customer journeys, not components: Playwright on the web, Maestro on Android, a Mac app still thinly tested, videos that come out of the tests, and this week's real failures.",
      "content_text": "The question I put to my agents is simple: if a customer can no longer book a dog walk on Akinaza, how do we know? A list of test files is not an answer. I want to be able to say \"the booking journey passes, here is its last run\". This chapter tells what we put in place since, with Akinaza as the example, because it is the product with the most surfaces: the customers' web, the owners' web, Android, iOS and a Mac app. I also say what is not done. Every claim links to a file, a PR or a CI run, and the numbers come from the repositories. What we test, and what we don't My starting sentence, whic",
      "date_published": "2026-10-30T18:00:00.000Z",
      "date_modified": "2026-10-30T12:00:00Z",
      "tags": [
        "ui",
        "measurements"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-we-test-our-uis-en.2f9f634e03.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 911089
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-we-test-our-uis-en.6d3bffa53b.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1308243
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/how-the-founder-and-the-cto-agent-work/",
      "url": "https://blog.krizaka.com/story/how-the-founder-and-the-cto-agent-work/",
      "title": "How the CTO agent and I work",
      "summary": "Who decides what, why four lanes and no more, one board where everything is an issue, the rules that hold, and a cockpit I read on my phone.",
      "content_text": "People keep asking me: \"But concretely, how do you work with your agents?\" Until now I answered in pieces, one chapter on the rules, one on CI, one on the audit. This time I lay it all out, because it has settled into a shape stable enough to tell, and because I read this chapter on my own phone thinking \"right, that's how we do it\". I am alone in Montréal. With me there are coding agent sessions and, above them, an agent I call the CTO. It hardly codes at all. It orchestrates: it reads the state of the work, opens and sorts issues, picks which session does what, and reports back to me. Everyt",
      "date_published": "2026-10-29T18:00:00.000Z",
      "date_modified": "2026-10-29T12:00:00Z",
      "tags": [
        "ai",
        "integration",
        "costs"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-the-founder-and-the-cto-agent-work-en.1e63ec1c37.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 543671
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-the-founder-and-the-cto-agent-work-en.8d37e15501.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 750700
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/the-audit-said-ai-code-we-measured/",
      "url": "https://blog.krizaka.com/story/the-audit-said-ai-code-we-measured/",
      "title": "The audit said our agents write “AI code”. We measured.",
      "summary": "An external audit says our code looks written by AI: null checks everywhere, procedural services. What was true, what was overstated, and what the build now refuses.",
      "content_text": "On 28 October I received an external audit of our code, and it said, roughly, this: your agents write “AI code”. Cascades of if (x != null) , procedural services, if/else over sign in providers, and a Java 21, a Spring and a TypeScript that we do not use to their full extent. My first reaction was to defend us. The second one was more useful: what if we looked? In chapter 9 I had already had the organisation audited by Claude and Gemini, and I had seen that an opinion has to be checked. This chapter tells what we measured, what we decided, what went wrong on the way and what it costs. I only s",
      "date_published": "2026-10-29T03:10:00.000Z",
      "date_modified": "2026-10-28T12:00:00Z",
      "tags": [
        "ai",
        "measurements",
        "security"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/the-audit-said-ai-code-we-measured-en.46823e3b5d.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 620964
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/the-audit-said-ai-code-we-measured-en.dc2683ce73.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 893697
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/we-bought-sonarcloud-what-do-we-have-now/",
      "url": "https://blog.krizaka.com/story/we-bought-sonarcloud-what-do-we-have-now/",
      "title": "We bought SonarCloud, what do we have now?",
      "summary": "We moved to SonarCloud's paid plan. What it changed in our CI, why the gates first turned red, what the figures say today, and what Sonar does that our own guards do not.",
      "content_text": "In chapter 14, I told you about the SonarCloud dashboard and its red gates. There was a question I had not asked out loud: what did we actually buy? We moved to SonarCloud's paid plan. I decided it, because I wanted Sonar to be part of the merge gate, not only a dashboard we look at from time to time. This chapter tells what changed, what started badly, and the figures I read today, October 28. I only say what I could check, in the code, in the merged PRs and on SonarCloud's public API. Before: an analysis that ran on its own At first we used SonarCloud's automatic analysis. You switch it on w",
      "date_published": "2026-10-28T04:50:00.000Z",
      "date_modified": "2026-10-28T12:00:00Z",
      "tags": [
        "integration",
        "security"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/we-bought-sonarcloud-what-do-we-have-now-en.eab041d836.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 647282
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/we-bought-sonarcloud-what-do-we-have-now-en.7b16739977.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 944004
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/ci-the-error-never-leaves-the-editor/",
      "url": "https://blog.krizaka.com/story/ci-the-error-never-leaves-the-editor/",
      "title": "The error no longer leaves the editor",
      "summary": "Two days after the chapter on our CI, SonarCloud's dashboard showed me what it let through. How we moved checks toward the editor, and why every bug must leave a guard.",
      "content_text": "In chapter 11, I showed you our CI. Two days later, SonarCloud's dashboard showed me what it let through. Everything worked as planned. The gates were required, the PRs waited. But several Sonar gates were red on main, almost always for the same family of reasons: code too complicated, small mistakes an editor could have flagged. The CI held the gate, but it held it too late. The agent was done, the PR was open, and ten minutes later we learned everything had to be redone. This chapter tells the move: from \"the CI catches it\" to \"the error never leaves the editor\". What the dashboard showed me",
      "date_published": "2026-10-28T01:00:00.000Z",
      "date_modified": "2026-10-27T12:00:00Z",
      "tags": [
        "integration",
        "security"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/ci-the-error-never-leaves-the-editor-en.e4f96a1557.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 397045
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/ci-the-error-never-leaves-the-editor-en.49200c14a0.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 627520
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/orazaka-v2-the-architecture-we-build/",
      "url": "https://blog.krizaka.com/story/orazaka-v2-the-architecture-we-build/",
      "title": "Sovkora v2: the architecture we are building",
      "summary": "What we own and what we plug in: the orchestration, the control loop, the GPU lease, live steering and access. Then what is not there yet, and the v3 these slices lead to.",
      "content_text": "When someone asks me what Orazaka is, technically, I often answer with a list of repositories. It is a bad answer. A list does not say what we are building, or why. On 26 October, I had the architecture reviewed by an outside audit: a language model, given the context. It suggested a hybrid architecture, keeping some parts and replacing others with well known frameworks. Some of it was right. Some of it described as existing things we have not built. We checked it against the code, line by line, then a counter review tightened what was left. The outcome is one decision, ADR 092, and one senten",
      "date_published": "2026-10-27T01:30:00.000Z",
      "date_modified": "2026-10-26T12:00:00Z",
      "tags": [
        "sovkora",
        "ai",
        "security"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/orazaka-v2-the-architecture-we-build-en.e8bb7aacb4.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 816854
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/orazaka-v2-the-architecture-we-build-en.e4e856cd75.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1274940
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/krizaka-ui-how-we-cut-it/",
      "url": "https://blog.krizaka.com/story/krizaka-ui-how-we-cut-it/",
      "title": "Krizaka UI: how we cut it, and why",
      "summary": "Nineteen packages rather than one, an entry per component, documentation written by the code, pages made of blocks, third-party engines behind our facades: each cut, its reason, and what went wrong.",
      "content_text": "Chapter 5 told why we have a single interface kit, and the day we saw it looked generated. This one is about its frame: how the kit is cut, and above all why each cut is there. A library is judged by what it refuses to put in the same package. Everything is in krizaka ui, and the manual is on ui.krizaka.com. First, I counted On the evening of 11 October I counted what our apps were writing by hand. This is ADR 001, and the numbers stopped me. In the products: 36 hand written forms, 42 tables, 23 mobile lists without serious virtualisation. One product's web app carried about 3,000 lines of hom",
      "date_published": "2026-10-26T16:00:00.000Z",
      "date_modified": "2026-10-26T12:00:00Z",
      "tags": [
        "ui"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/krizaka-ui-how-we-cut-it-en.d8b9e50ee4.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 764963
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/krizaka-ui-how-we-cut-it-en.14275b759d.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1137301
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/how-we-built-our-ci/",
      "url": "https://blog.krizaka.com/story/how-we-built-our-ci/",
      "title": "How we built our CI",
      "summary": "Several people asked me how Krizaka does its CI. The path of a pull request to production, with the real files, and what broke along the way.",
      "content_text": "For a few weeks, people who build software have been asking me the same question. How do you run your CI, with so many repositories and agents opening pull requests all day? I answer it here, once, with the real files. Everything I show is in public repositories of the krizaka organisation. You can go and check every line. Why one shared CI At first, each repository had its own CI file. We copied the neighbour's and adapted it a little. A fix in one repository never reached the others. In the chapter on agents, I talked about a “build pack” for the CI, not shipped yet. Part of it has shipped. ",
      "date_published": "2026-10-25T19:40:00.000Z",
      "date_modified": "2026-10-25T12:00:00Z",
      "tags": [
        "integration",
        "security"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-we-built-our-ci-en.dcc2a387e9.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 660033
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/how-we-built-our-ci-en.6c37977eef.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1085337
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/a-curve-that-climbs/",
      "url": "https://blog.krizaka.com/story/a-curve-that-climbs/",
      "title": "A curve that climbs and climbs",
      "summary": "With agents, opening work costs almost nothing. Closing it does. What the curve of our issues taught me, and why we now focus on three projects.",
      "content_text": "One evening, I opened the organisation's issues page on GitHub. There is a small chart at the top. I looked at it for a long time. The line of open issues kept climbing. The line of closed issues climbed too, but much more slowly. This chapter is about what that curve taught me, and the decision I made while looking at it. The curve I did not want to copy GitHub's chart. I wrote a small script that asks GitHub's search, day by day: how many public issues were opened, how many were closed because the work was done, how many were closed as not planned. The script keeps the figures in a file, wit",
      "date_published": "2026-10-25T13:00:00.000Z",
      "date_modified": "2026-10-25T12:00:00Z",
      "tags": [
        "ai",
        "measurements",
        "integration"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/a-curve-that-climbs-en.930d48846a.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 818960
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/a-curve-that-climbs-en.df7697ae56.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1228301
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/what-claude-found-auditing-krizaka/",
      "url": "https://blog.krizaka.com/story/what-claude-found-auditing-krizaka/",
      "title": "What Claude and Gemini found auditing Krizaka",
      "summary": "Two AIs, one prompt, an unflattering audit of our GitHub organisation. The raw answers, what holds, what doesn't, and what I checked myself.",
      "content_text": "A while ago, I asked a friend who works in IT to run an experiment with no bias. He asked three models, Claude, Gemini and ChatGPT, to look at our GitHub organisation and give an unvarnished audit. What he told me was flattering. Too flattering for me to repeat without proof. At Krizaka, a claim has to lead to a measurement, a run or a line of code. I don't have the outputs of his experiment, so I can't quote them. So I ran it again myself, on 24 October, with a prompt written in advance and published, and an answer published as is. Why I ran it again A compliment someone passes on can't be ch",
      "date_published": "2026-10-25T01:00:00.000Z",
      "date_modified": "2026-10-24T12:00:00Z",
      "tags": [
        "ai",
        "measurements",
        "security"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/what-claude-found-auditing-krizaka-en.2a4c004c44.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 544365
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/what-claude-found-auditing-krizaka-en.3a70d0d535.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 835382
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/an-empty-result-is-not-a-success/",
      "url": "https://blog.krizaka.com/story/an-empty-result-is-not-a-success/",
      "title": "An empty result is not a success",
      "summary": "What we will never give an agent, the one rule nobody can change, and an engine that learned to say “right”, not only “done”.",
      "content_text": "A few days ago, a prospect research Studio finished green on Orazaka. Status: succeeded. We opened the result, and the ranking was empty. The summary too. Nothing had crashed. The model had answered, every step had returned something, and the engine did exactly what it was asked: move to the next step, then declare the run finished. The problem is that \"something\" could be nothing. The same day, an authenticity check on Yomizaka returned its signals flattened into one line of text. And a test account saw its free audit used up by a run that had failed. Three symptoms, one cause: the engine kne",
      "date_published": "2026-10-24T13:00:00.000Z",
      "date_modified": "2026-10-24T12:00:00Z",
      "tags": [
        "sovkora",
        "ai",
        "security"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/an-empty-result-is-not-a-success-en.c2aeb4f57c.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 498862
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/an-empty-result-is-not-a-success-en.54a4261f5d.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 758088
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/building-with-agents/",
      "url": "https://blog.krizaka.com/story/building-with-agents/",
      "title": "Building almost alone, with agents",
      "summary": "Rules written after every mistake, the keys an agent must never see, the scripts that run the company, and thirty coding agents in one night.",
      "content_text": "On 12 October, my Mac saturated. Several coding agents were working on Krizaka at the same time, each with its builds, its containers and its simulator, and the system ended up refusing to open one more file. I tell the story of that breakdown in the post about the Mac. That day, I understood one simple thing: letting agents write code is not magic. It is organisation. This chapter tells how I moved from vibe coding to written rules, how I keep the access keys, how scripts run the company, and what broke one night when about thirty agents worked at the same time. How I got out of vibe coding V",
      "date_published": "2026-10-24T13:00:00.000Z",
      "date_modified": "2026-10-24T12:00:00Z",
      "tags": [
        "ai",
        "security",
        "integration"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/building-with-agents-en.5ecb41f934.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 908936
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/building-with-agents-en.1cb6c4c4c5.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1289416
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/the-first-products/",
      "url": "https://blog.krizaka.com/story/the-first-products/",
      "title": "The first products: Verikora and Folioviv",
      "summary": "What Verikora reads, why our source of truth is the law itself, dated and quoted, what it costs next to ready-made reports, and our own invoice refused by our own tool.",
      "content_text": "A few weeks ago, I was in a café with a friend who manages buildings in Montréal. He took out his phone, turned the screen toward me and showed me a pay stub in PDF. A candidate for a four and a half had sent it to him the night before. “Do you see anything?” I did not see anything. The logo was clean, the amounts added up, the deductions looked like Québec's. He did not see anything either, and that is exactly what bothered him. He gets dozens of them a month. He looks at them by eye, he forwards them by e mail to the owner with the ID and the rest, and he hopes. Then he scrolled. The next st",
      "date_published": "2026-10-23T18:00:00.000Z",
      "date_modified": "2026-10-23T12:00:00Z",
      "tags": [
        "verikora",
        "documents",
        "law"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/the-first-products-en.10235daa68.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 727280
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/the-first-products-en.9a164dadc8.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1039111
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/one-interface-everywhere/",
      "url": "https://blog.krizaka.com/story/one-interface-everywhere/",
      "title": "One interface, from the web to the Mac",
      "summary": "One kit for the web, the phone and the Mac: what a shared fix is worth, the “AI” look we measured before we fixed it, the bug nobody had seen, and why Orazaka has a Mac app.",
      "content_text": "On Friday evening, a tab bar in one of our apps showed a small vertical scrollbar on the right. One pixel too many, caused by the focus ring. A detail. But the same pattern showed up in the settings navigation, in a row of short videos, in a date picker and in the table of contents of the reading pages. The fix was made once, in the kit (krizaka ui 295), with tests that failed before. Then every application received it by upgrading. That is the idea of a shared kit. One kit, three platforms Krizaka UI is the component library of all our sites and applications. It targets three platforms: the w",
      "date_published": "2026-10-23T17:00:00.000Z",
      "date_modified": "2026-10-23T12:00:00Z",
      "tags": [
        "ui",
        "sovkora"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/one-interface-everywhere-en.bba0768897.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 1744066
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/one-interface-everywhere-en.d9e84b93f2.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 2203216
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/from-a-clip-to-a-film/",
      "url": "https://blog.krizaka.com/story/from-a-clip-to-a-film/",
      "title": "From a clip to a film: what we got wrong first",
      "summary": "Our first film was beautiful and bad. The fix was in the production chain, not in the prompt: two films judged by eye, then measured, and what a finished minute costs.",
      "content_text": "On 20 October, I watched our film “The World of Krizaka: l'aube” for the second time, 62 seconds in 4K, made with our birds. My verdict was short: horrible. The voices, the captions, the way the birds interact, all of it was too ugly. It was not a sloppy film, though. The image and the light were beautiful, I said so myself. It had cost 6.5 hours of an H200 GPU, that is $34.42 at the rental company. But a beautiful image does not make a film, and what I had just watched was not a film. Here is what was wrong, what we changed, what it cost, and what I learned about the difference between improv",
      "date_published": "2026-10-23T16:00:00.000Z",
      "date_modified": "2026-10-23T12:00:00Z",
      "tags": [
        "ai",
        "measurements"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/from-a-clip-to-a-film-en.a84d9b3add.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 851430
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/from-a-clip-to-a-film-en.94cf5686d6.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1146087
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/what-a-gpu-costs/",
      "url": "https://blog.krizaka.com/story/what-a-gpu-costs/",
      "title": "What an image and a video cost: my bills",
      "summary": "One Mac, one hosted API and one rented H100, on the same images and clips. The bills, the GPU left on, the price of a result you keep, and why I moved 15 November.",
      "content_text": "During our last batch of videos, the last clip came out of the GPU, and the pod kept running. For about two hours, an H100 rented at US$3.99 an hour waited for work that would not come. It had a watchdog meant to shut it down after a period without work. It never did, because to destroy the pod it had to call RunPod's API, and the pod had no key for that. A watch loop started on my Mac is what finally stopped it. Those two hours cost between $7 and $8, on a GPU bill of $23.23. It stings, and it is a good place to start on what an image and a video really cost to generate. All the figures come ",
      "date_published": "2026-10-23T16:00:00.000Z",
      "date_modified": "2026-10-23T12:00:00Z",
      "tags": [
        "costs",
        "measurements",
        "sovkora",
        "ai"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/what-a-gpu-costs-en.cde5ee76e3.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 1296426
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/what-a-gpu-costs-en.a3236912a8.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1631376
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/an-engine-on-a-laptop/",
      "url": "https://blog.krizaka.com/story/an-engine-on-a-laptop/",
      "title": "An AI engine on a laptop",
      "summary": "A Mac out of breath, a ClassCastException that killed the chat, models that became database rows. What I learned running Orazaka on a laptop.",
      "content_text": "At first, Orazaka was a discovery. A personal project, on my MacBook, to understand what you can do with AI models running at home. Ollama for text, the Mac's GPU (Metal) for images, a database and a few services started by hand. If someone had taken the laptop, there would have been no Orazaka left. That Mac is the first server of Orazaka, the AI engine I'm building. The idea behind Orazaka is simple: run AI on a machine you control. For me, that machine is my laptop first. Here's what it taught me. On 12 October, the Mac hit the wall On 12 October my Mac hit the wall. The system load swung b",
      "date_published": "2026-10-23T15:55:00.000Z",
      "date_modified": "2026-10-23T12:00:00Z",
      "tags": [
        "sovkora",
        "ai"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/an-engine-on-a-laptop-en.8fa4ee8609.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 1226124
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/an-engine-on-a-laptop-en.23cc449281.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 1585564
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/story/why-krizaka/",
      "url": "https://blog.krizaka.com/story/why-krizaka/",
      "title": "Why I am building Krizaka in Montréal",
      "summary": "A GPU rented in Montréal, a Quebec privacy law, and three plain questions to ask any AI tool. What sovereign means at Krizaka, what it costs, and what is still unsolved.",
      "content_text": "This week our demo videos were generated on an NVIDIA H100 rented from RunPod. In the rental log, the pod's region is called CA MTL 1. Montréal. The machine that made our pictures was running in my own city. That was not the plan. I wanted a 48 GB GPU at DigitalOcean, in Toronto, where our databases already live. There was never one in stock in that region, the only one our private network reaches. So we took the H100 in Montréal. Looking at that region name, it struck me as a good place to start explaining what we do. First, who is writing My name is Oussama, and I build Krizaka in Montréal. ",
      "date_published": "2026-10-23T15:54:00.000Z",
      "date_modified": "2026-10-23T12:00:00Z",
      "tags": [
        "sovereignty",
        "ai",
        "montreal"
      ],
      "language": "en",
      "attachments": [
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/why-krizaka-en.f457daa2d2.epub",
          "mime_type": "application/epub+zip",
          "size_in_bytes": 522637
        },
        {
          "url": "https://krizaka-assets.b-cdn.net/sites/blog/posts/downloads/why-krizaka-en.58fb10b75d.pdf",
          "mime_type": "application/pdf",
          "size_in_bytes": 740822
        }
      ]
    },
    {
      "id": "https://blog.krizaka.com/basics/is-that-ai-model-really-open-source/",
      "url": "https://blog.krizaka.com/basics/is-that-ai-model-really-open-source/",
      "title": "Is that AI model really open source? What the licence actually says",
      "summary": "Open source, open weights, source-available: what these words promise, what a licence really controls, and what I read while choosing the image, video and voice models behind our products.",
      "content_text": "On 15 October I checked that none of the default models in our AI engine carried a non commercial licence, since we sell to companies. The default English voice was called Ryan. It was a Piper voice, free, published \"as open source\". When I opened its card, I read: CC BY NC SA 4.0. The \"NC\" means non commercial. The program is free, the model weights can be downloaded, everyone says \"open source\". But the voice was trained on a dataset that forbids commercial use, and that restriction follows the voice. Ryan stayed available, never as a default, and a public domain voice took its place. Since ",
      "date_published": "2026-10-20T05:45:00.000Z",
      "date_modified": "2026-10-20T12:00:00Z",
      "tags": [
        "ai",
        "basics",
        "law"
      ],
      "language": "en"
    },
    {
      "id": "https://blog.krizaka.com/basics/ai-model-families-and-why-you-need-a-gpu/",
      "url": "https://blog.krizaka.com/basics/ai-model-families-and-why-you-need-a-gpu/",
      "title": "Models, their families, and why you need a GPU",
      "summary": "Dense or MoE LLMs, embeddings, vision, image and video diffusion, audio, upscalers: the model families it took me months to untangle, and why each one has to fit in a GPU's memory. With our measured times.",
      "content_text": "For weeks, models gave me a real headache. I would read \"7B\", \"A14B\", \"Q4 K M\", \"MoE 235B A22B\", \"FP8\", and nod without being sure I understood. Then I tried to make a video on my Mac, and the model crashed, telling me it was 26 GiB of memory short. That day I understood that the real question was not \"which model is best?\" but \"do this model and this machine go together?\". This is the post I wish I had read before starting. The families of models, what the numbers in their names mean, and why almost everything comes down to the GPU's memory. If token, parameter or quantization are new words t",
      "date_published": "2026-10-17T17:00:00.000Z",
      "date_modified": "2026-10-17T12:00:00Z",
      "tags": [
        "ai",
        "basics",
        "measurements",
        "costs"
      ],
      "language": "en"
    },
    {
      "id": "https://blog.krizaka.com/basics/understanding-ai-before-you-start/",
      "url": "https://blog.krizaka.com/basics/understanding-ai-before-you-start/",
      "title": "Understanding AI before you start: the words, without the jargon",
      "summary": "Token, context window, quantization, RAG, MCP, agent, diffusion. The words I use without explaining them, taken one by one with a simple picture and what really runs on our side.",
      "content_text": "Rereading the first posts of this blog, I made a list of the words I use without ever explaining them. Token. Quantization. RAG. MCP. Agent. The list was long, and a bit embarrassing. So I'm starting over from the beginning. One word at a time, with a simple picture and an example of what really runs on our side. The illustrations were generated for the occasion on my Mac, with FLUX.2 [klein] 4B , the image model Orazaka runs locally. Under each one, \"How this was made\" gives the prompt, the seed and the time. A language model is a very good guesser of what comes next A large language model, a",
      "date_published": "2026-10-17T02:22:41.000Z",
      "date_modified": "2026-10-16T12:00:00Z",
      "tags": [
        "ai",
        "basics",
        "sovkora"
      ],
      "language": "en"
    },
    {
      "id": "https://blog.krizaka.com/basics/anatomy-of-an-ai-prompt/",
      "url": "https://blog.krizaka.com/basics/anatomy-of-an-ai-prompt/",
      "title": "The anatomy of an AI prompt: what each added line really changes",
      "summary": "A six-word prompt, then three layers of detail, on my Mac with the same seed. What gets better, what gets worse, what a negative prompt can do, and what I didn't prove.",
      "content_text": "I had the story in my head before I made the pictures. You type six words, the model gives you a three legged player, and you learn you should have written a better prompt. It's the story you read everywhere. I only wanted to illustrate it. The model didn't play along. What it showed me was less dramatic, and more useful. So here is what I actually got, with the prompt, the seed and the time under every picture. Every picture was made on my MacBook Pro (M1 Max, 64 GB) with the image engine's command line. One picture per prompt, a single seed (2026). This is a demonstration, not a benchmark: I",
      "date_published": "2026-10-16T13:06:09.000Z",
      "date_modified": "2026-10-16T12:00:00Z",
      "tags": [
        "ai",
        "prompts",
        "sovkora",
        "measurements"
      ],
      "language": "en"
    }
  ]
}