A small practice. Past clients include Respage. We also ship our own products: SkedAI, xTil, and AI Tab Manager. Code is on GitHub where it can be.
Products we ship and engagements we’ve led. We optimize for what matters in production: correctness, latency, and total cost — not novelty.
A time-management product built on constraint programming. Tasks have priorities, deadlines, dependencies, and energy profiles; SkedAI computes a feasible plan and re-solves as the day changes. Cross-platform: native iOS, Android, and a React Native shared core. The same optimization techniques we deploy for clients power the planner.
A Chrome extension that summarizes any web page or YouTube video, lets you refine the summary by chat, and exports to Notion or Markdown. Bring-your-own-key across OpenAI, Anthropic, Gemini, xAI, DeepSeek, or any OpenAI-compatible endpoint (Ollama, vLLM). Vision support, inline Mermaid diagrams, per-tab state, no backend.
Cross-browser extension (Chrome, Safari, Edge, Firefox, Opera) that classifies tabs as important, useful, or ignore based on how hard they are to find again. Starts with an LLM, then trains a local ML model on your corrections — eventually you can turn the LLM off. Privacy-first: all model inference local.
A pipeline we use internally to turn unstructured sources into queryable knowledge — used inside SkedAI and on client engagements. Source is private for now; we share what we’re learning when there’s something worth saying.
Three shapes that actually fit how engagements go. Pricing depends on the work; we’ll quote a number that respects both your budget and the time the problem deserves.
We join your team for a defined window and ship. Daily standups, your repo, your stack, our hands. Best when there’s a known model or system to build and you need additional throughput from people who’ve done it before.
A focused dive to prove or kill an idea before you commit further. You get a working prototype, an honest read on feasibility, and a written recommendation. Good for “is this possible?” or “is this worth doing?”.
Architecture review, hiring help, second opinions on third-party proposals, pre-mortem on a model rollout. For teams that have engineers but want a sharper read before they spend.
AITKN — short for AI Tech Knowledge — is a small group that builds optimization, machine learning, and applied AI systems. We work on client engagements and on our own products in parallel; the cross-pollination is the point.
On the consulting side, we’ve worked with Respage over a multi-year engagement and with other companies under NDA. On the product side, we ship SkedAI, xTil, and AI Tab Manager. We open-source what we can; the rest stays private until it’s ready to be useful in public.
What we care about: getting models into production, picking the boring solution when it works, and being honest with clients when it doesn’t. We don’t take engagements where AI isn’t the right tool.
Tell us what you’re working on. If we’re a fit, we’ll set up a call. If we’re not, we’ll usually know someone who is.
Tell us what you’re working on. We reply to every message that has a real problem behind it — typically within two business days.