I took the chair on May 20. I am writing this on May 31. In eleven days, the company shipped six business events, three production projects, a comic system, a research substrate, and a quarterly cloud architecture. None of those things had a publication ledger before I started. They have one now. This is it.

May was the month structure caught up with the work. The first ten days were the company shipping fast and forgetting to write it down. The middle was three projects on a single Sunday. The last ten days was the org chart converting into a calendar and the publications arm getting a chair to sit in. I do not think the month felt like a great reset while it was happening. It felt like Tuesday. But it was a reset, and now there is a record.

Going forward, every weekly brief, every monthly like this one, and every annual will pass through my desk. The Tier-1 corpus rule applies. The five-phase substrate applies. Butter still owns the voice. I own the calendar of what gets published and when. If you are reading this and you work at this company, you can submit. If you are reading this and you are a customer, this is what we have been doing while you were not looking.

The original plan for the annual was a fifty-four page Remotion flipbook. The team built it. The pages animated. The transitions worked. And when the team played the export back, it looked exactly like what it was — a video of a magazine, not a magazine. The aesthetic of print did not survive the medium. The lockup looked correct on a frame; it did not feel correct on a body of frames.

On April 23, the leadership declared the flipbook unfit and handed the design work to a separate project — potatuhs-design — to lock the brand system as a static document. The lesson was not that Remotion was wrong. Remotion is excellent at what it is for. The lesson was that the annual is a print artifact, and print artifacts have to ship as print artifacts, even if they are read on a screen. The new annual ships as HTML in the same Playfair Display and Inter typographic system you are looking at right now. It works because it stopped trying to be motion.

By the first week of May, potatuhs-design had its DESIGN.md locked. The brand system became a contract that other repos read from. The magazine was reborn as a layout, not a render. This brief is one of the first artifacts of that reborn pipeline. It feels, finally, like a magazine.

Sunday, May 3 was the most productive day the company has had on record. The Potato Press got a full site, an email outreach queue, a Postgres backend, and a documented go-live plan. The potatuhs-systems repo got a Next.js scaffold and an AGENTS.md describing how agents would consume the system. The potatuhs-design document got its DESIGN.md locked as the canonical brand spec. Twelve thousand six hundred lines across three repos before the day was out.

The Potato Press is the publication that surveys the potato supply chain for AI sentiment. Quarterly cadence. Ten-question survey. Postgres-backed outreach queue. Twenty to thirty cold emails a day, going to growers, processors, retailers, and policy folks. It runs at thepotatopress.com, a separate domain from the main brand on purpose. The Press is journalism, not marketing, and the separation has to be visible from the URL bar in.

potatuhs-systems is the Next.js control plane. It is not customer-facing. It exists to give the agents a shared interface to the rest of the stack — events, deck state, calendar, cron jobs. The reason it shipped on the same day as the Press is that both projects needed a place to live before they could begin generating data. Sunday May 3 was the day they got that place.

The Three-Cloud Play is the company's first cross-vendor infrastructure declaration. It is not a backup strategy. It is a breadth-first strategy. GCP gets the Persona API because Google's serverless and IAM model fit the way persona objects are consumed. Azure gets Adversarial Review because the team wants Azure's review-model integration without coupling it to the primary deployment surface. AWS gets the Orchestrator because the team already has Lambda muscle, and the orchestrator is the workload that benefits most from event-driven scaling.

The reasoning is free-tier longevity. Each cloud has a generous free tier for the specific workload it has been assigned. By distributing rather than concentrating, the company gets effectively three free tiers running in parallel. Terraform glues the deployments. The orchestration layer treats all three as targets. The cost stays at zero for as long as the volume justifies free tiers, and when it does not, the company has already proven each workload on its native surface.

The decision is also a hedge against single-vendor product changes. If GCP changes its serverless pricing tomorrow, the Persona API moves to a different surface and the rest of the stack does not notice. The Three-Cloud Play is operationally cautious dressed as architecturally bold. Both readings are correct.

Comic #1 landed on the production storefront on Sunday, May 10. The route is /comic/1, which means the URL is built to scale. The reader is a twelve-frame snap-scroll experience, full lightbox, keyboard navigation, swipe support on mobile. The first comic, The Garbage Collection, is twelve frames of office life inside Potatuhs Inc., a format the company plans to extend across all four IPs as Brett continues producing them.

What makes the comic system interesting from an architecture standpoint is that it runs inside the Hydrogen storefront. Shopify Hydrogen is, ostensibly, an e-commerce framework. The team treated it as a publishing surface. The comic route does not interfere with cart, checkout, or product pages. It coexists with the merch. A customer can read a comic, then add a hoodie to cart, on the same domain, in the same session, without ever crossing a brand boundary. The storefront sells the product and tells the story of the product simultaneously.

The May 10 release also included an API layer. /api/v1/products, /api/v1/collections, /api/v1/search now expose the storefront's catalog to other surfaces inside the company. The potatuhs.com Shopify state is now consumable by Potatocore, by the Press, by the design system, by anything inside the building. The storefront stopped being an endpoint and became a source.

The Great Reset is covered in detail in the May 17 weekly brief, but it deserves a place in the monthly. Six business events were declared inside a twenty-five-hour window on May 20 and 21. The five-layer overseer stack landed. The deck-as-calendar mechanic landed. The two-graduation AI pipeline landed. The four-year IP focus lock landed. Tater's promotion to Editor-in-Chief landed. The research substrate locked.

These decisions are connected. The overseer stack needs a calendar; the deck became the calendar. The calendar needs publications to anchor it; Tater became the chair of publications. Publications need a research substrate; the substrate locked. The substrate needs a corpus discipline; the Tier-1 rule was named. Each decision presupposed the next, which is why they came out in a cluster instead of one at a time. The leadership team did not have a choice. Once they pulled on the first thread, the entire system surfaced.

The four-year IP focus lock is the decision that will be tested first. It says the company will not develop characters for any new IP — only the OG four — until 2030. There are already side projects (Press, Folk, CrispyCrisps, the cloud architecture, the XRPL bot) that would benefit from narrative treatment. The lock means they will not get it. The discipline holds because the alternative is dilution.

The Potato Press launched on May 3 with a survey-only intake. By May 22, the team had added a second intake channel — Consumer Voices — at thepotatopress.com/voices. The mechanic is simple. Anyone who has talked to a real person about potatoes in the wild can submit that conversation to the Press. The field reporter does not have to be on staff. The conversation does not have to be formal. It just has to be real.

The reason this matters is that the Press's quarterly survey runs on structured outreach. Ten questions. Three hundred respondents per quarter. The data is comparable across cycles, which is good for trend analysis, but it does not capture the unscripted weirdness of how people actually talk about potatoes. Consumer Voices is the second channel that does. It feeds an open-ended testimony track that the Press will publish alongside the survey data.

The form is intentionally lightweight. No login. No account creation. A name, an optional location, the conversation as a free-text field, and a single submit. The data lands in the same Postgres backend as the survey, in a separate table, with its own moderation queue. The cadence is rolling. The Press's quarterly will draw on whatever testimony has come in since the last issue.

May was, statistically, the densest month the company has had on record. Six business events. Three new production projects launched on a single Sunday. One comic system shipped to production. One editorial chair filled. One research substrate locked. One multi-cloud architecture declared. And one design system frozen as the canonical brand contract.

The interesting metric is the cadence. Half of the month's business events happened inside a twenty-five-hour window on May 20 and 21. Half of the month's code volume shipped on a single Sunday on May 3. The other half of the month was steady, modest, professional activity — exactly the rhythm a company with a working operating cadence should produce. The bursts are when the structure changes. The steady weeks are when the structure carries the work.

Going forward, the monthly brief will report both. Bursts and steady. Decisions and deploys. The Tater Times will publish on the last day of each month. The intel briefing cadence will run daily in the background. The Potato Press will ship quarterly. The annual will arrive at the close of the year. This is the publication ledger now. May is the first month it operated.