Jackson Cantrell’s Weblog

A few people, a lot of agents, and who makes the calls.

Microcompute for the rest of us

Pixel art of a sunset over purple Verdugo ridges. On the left, a tall black glass tower stands on a grey concrete plaza, a dark storm cloud wrapped around its top, a red light blinking on its spire and a tiny price tag hanging off one corner. On the right is a small commune of low cream box houses with teal roofs, warm windows and blinking teal server lights, among garden beds under strings of lights. Tiny people and little teal robot agents work together between the houses, carrying parcels and watering the garden.
The tower sends its invoice by drizzle. The commune just hangs another string of lights.

The cloud is a marvel. It is also confusing as hell for small businesses.

Endless machines and a menu of 200 services is great if you’re a giant company full of platform engineers and AWS architects (and have a team whose whole job is reading the bill). If you’re a ten-person shop, you get the same slop.

My brother Clay and I think that’s about to change drastically.

Agents eat computers

Every time code changes, a computer has to build it and test it before anyone trusts it. That used to happen a few times a day. Now agents write code non-stop at a blistering pace.

GitHub’s Octoverse 2026 says developers now start more than 40% of pull requests with an agent. GitHub Actions, the service that runs those builds and tests, ran 124 million jobs a day this year. At Anthropic, Sachin Malhotra wrote that their testing volume grew 25x in six months.

Twenty-five times. In six months. Yuck.

Tara Tan at Strange Ventures said it plainly in August: “Coding agents are generating more code than existing infrastructure was built to handle.” Lenny Pruss at Amplify Partners went further. Testing, he wrote, “stops being a gate, and becomes a sandboxed multiverse.” Gergely Orosz has been tracking what the flood does to code review. Mitchell Hashimoto called it “the Gmail moment for version control.”

None of them were writing about small businesses. I’ll do that part.

Big software, small software

Pete Koomen put the problem best in Y Combinator’s Fall 2026 Requests for Startups. The big clouds “were designed for shipping Big Software that scales with many users… at the cost of complexity.” He wants “a cloud for small software.” Tools with one user, or a handful.

I’d stretch it one step. It’s not just the software that’s small. It’s the business. The accounting firm. The clinic. The co-op. The two brothers trying to commoditize actions runners...

David Heinemeier Hansson made the money case when his company left the cloud: “Of course it’s expensive to rent your computers from someone else.” He was talking about a company far bigger than ours. The math still travels.

Price, meet basement

So here’s our bet. Call it microcompute. Small, fast, cheap computers that a small team, or the agents working for one, can use without hiring a cloud whisperer. Priced near cost. Easy to leave.

We started with the most boring, most measurable slice we could find: the machines that build and test code. That’s Sengi. It runs GitHub Actions jobs from $0.001 a minute. GitHub charges $0.006 for the same size machine. That’s one-sixth the price, and the first 3,000 minutes a month are free.

A pixel-art cross-section of a building titled PRICE PER MINUTE, 2 VCPU, each floor one tenth of a cent higher than the one below. The velvet-and-gold penthouse, six floors up from $0, is labelled GITHUB $0.006/MIN. A big coral arrow drops down the lift shaft to a cosy teal-lit basement one floor up from $0, where two little agents work beside a server rack, labelled SENGI BASIC $0.001/MIN.
It turns out the best room in the building is the one with the boiler.

Boring is the point. Nobody needs convincing that tests matter. Agents are already mailing them the bill.

The goal is to commoditize the runner, which is the computer that runs those jobs. It should be like electricity. You use it, it works, you don’t think about who sells it.

Creative destruction, meet Communism for robots and nerds

The old way to fight a giant and disrupt an industry was to raise an army. Convince investors to give you $40 million. Hire 200 people, burn a few years, pray.

The new way is smaller. A few people who know exactly what they want. A lot of agents doing the volume. Shared tools instead of private empires. We run our whole company on an open-source agent network we built, called PeoplesCommune. The name is half a joke and half our actual plan. Communism doesn’t work for people (it’s been tried) - but it seems to work well for robots.

Small Software can be safe software

Small software also has to be safe software. Intuit QuickBooks asked more than 34,000 small business owners about AI for its 2026 AI Impact Report. Their answer: “The biggest barrier isn’t cost.” It’s privacy, security and fear of mistakes. Fair. Cheap and sketchy is still sketchy. So the agents working for a small business should be model-agnostic, run on computers it controls or rents cheap, plug into the tools it already uses, and leave the real calls to a person.

Runners are step one. Pooled compute for small businesses and their agents is the long game. We’ll earn it a minute at a time.

How the first month went is in $4,000 a month, no ads. How two people run it is in Software with 1/25th the Headcount. And why any of this should make software better, not just cheaper, is in Unslop software.

Fair warning: we sell the cheap computers. I’m biased. I’m also right.

Jackson Cantrell

Jackson Cantrell is an AI-native CTO and agentic engineering leader in Los Angeles. He helps small teams build software with many AI agents at once while people stay in charge of the decisions, and he helps teams that are new to agentic tools get up to speed.

He and his brother Clay Cantrell were early contributors to Gas Town, which shipped multi-agent orchestration and agent-to-agent mail in January 2026, weeks before Claude Code added agent teams. They run their own agent mail system across machines. Jackson built emBEADings, an open-source tool for the Beads issue tracker that Gas Town runs on.

Before that he was on the founding team at Dyrt, an organic waste startup, where he was hired to lead marketing and went on to lead product. Customers asked for tools, so the team built waste bill analysis that showed true diversion rates and hidden fees, invoice ingestion, waste stream tracking, an integration with Schneider Electric Resource Advisor, and Dyrty Vision, a computer vision system that tells food scraps from wasted food and reports back to food and beverage teams. Dyrt grew from 1,000 to 1.5 million pounds of food waste a month, signed national deals with a grocery chain and a hotel chain, and ran a facility in Vernon, California that employed 30 people.

Dyrt was an attempt to fix a broken California recycling industry. Jackson, Clay and the Dyrt team published an audit of Los Angeles's RecycLA franchise, which found that at most 18 to 24 of every 100 tons of commercial waste could be counted as diverted from landfill. He learned the hard way how well-meaning but burdensome rules and entrenched interests can sink a facility that works: the permit for Dyrt's indoor composting facility in Vernon went before the Vernon City Council in August 2026 and was carried into September. The company ran out of money that fall, and his last day was October 1, 2026.

Earlier he reported on tech and finance as a producer and co-host of Barron's Streetwise with Jack Hough, led audio product at Finimize, and reported for public radio: KCRW in Los Angeles, and North State Public Radio, where he was part of the team whose 2018 Camp Fire coverage won an Edward R. Murrow Award. He studied neuroscience at Brown University.

He is open to conversations, collaborations and new roles: jaxtrell@gmail.com.

How he works

Projects

Work history

Press

Jack Hough, host of Barron's Streetwise, interviewed Jackson about going from no coding to shipping software with AI agents, and about rebuilding this site with Gas Town as a fictional 2002 college radio station.

Stories

Links