Mind the gaps.
A small studio building AI we’re willing to put our names on. Two decades each on the floor of the hyperscale era; the next stretch is software that knows what it doesn’t know. We answer our own email.
“What’s in your hands, I think and hope, is intelligence: the ability to see the machine as more than when you were first led up to it — that you can make it more.”
Answers from the studio. Hedged when uncertain. Rate-limited.
How we work
One substrate. Many tools. Same graph engine, same audit log, same refusal model under every product. Built once; reused.
Honest refusal beats confident guessing. When the system isn’t sure, it says so — with provenance. Your correction is canonical.
SLOs on day one. Latency budget, cost meter, append-only audit, refusal-first permissions, on-call pager wired before launch.
What we make
A collection partner for someone who works with objects of value. Drop a photo, get help understanding it. Period, region, maker when knowable. Confidence-graded. Lists straight to Etsy and eBay when you’re ready.
Try the beta →A resident operator daemon for a single-user POSIX host. UNIX-shaped: small mind, large toolkit, refusal-first permissions, append-only audit. For the people who live in their terminal.
Read the spec →A pan-African and Black-diaspora portal that fuses news, markets, and federated community under an ambient-intelligent layer. The substrate is open and decentralized; the lens is ours.
Go to Baobab →The world of food — ingredients, techniques, lineages, regional schools — as a real graph, not a recipe blog. Same substrate discipline as The Curator.
Get the early note →Lineage
Millions-of-machines fleets. Cluster orchestrators, deployment instruments, and the products that auto-deployed on top. Stubby. Chubby. Borg. Kubernetes since day one. cgroups. SRE on-call from the time the book was being written. The old discipline applied to a new substrate.
Three engineers. One inbox.
If any of it sounds like something you’d use, send a sentence.
alfred@lacunalabs.ai →