Kilva — open-source AI systems

AI systems that do the work.

A personal collection of open-source AI systems — workflow automation, data pipelines, and integrations built for real operations. Every machine here is free, MIT-licensed, and yours to fork.

Fig. 01 — the friction the machines remove

Every company has the same 3 leaks.

Repetitive work

The same emails written, the same reports rebuilt, the same data re-entered — every week. That work doesn’t need a person.

Scattered tools

Ten apps, none of them talking. Answers buried in tabs, docs, and inboxes. A system can hold all of it together.

Slow processes

Decisions waiting on manual steps. Follow-ups that slip. Speed is a systems problem, and systems are what the machines here solve.

Fig. 02 — the machines here

4 kinds of machines, built in public.

A

Custom AI systems

Purpose-built tools for your specific workflow. Not a chatbot, not a wrapper — a system shaped around your operation.

Try the demo →

B

Workflow automation

The repetitive daily work your team dreads, handled end to end by machines that don’t get bored or sick.

Try the demo →

C

Data pipelines

Messy, scattered data turned into structure — cleaned, connected, and ready the moment you need an answer.

Try the demo →

D

Integrations

We connect the tools you already use so they share one brain. Your stack stays yours — it just starts working together.

Try the demo →

Every machine here is open source — code, docs, and all. Free to run, free to fork, free to make yours.

Fig. 03 — how a machine gets built

4 steps, no surprises.

  1. 01

    Understand

    Start with the problem — where the time actually leaks — before a line of code is written.

  2. 02

    Design

    Architecture before code. How the system works is decided in the open, then built.

  3. 03

    Build

    Real systems tested against real data — not screenshots, not sandboxes.

  4. 04

    Publish

    Released with code, docs, and a license. Anyone can run it, fork it, and make it theirs.

Fig. 04 — time leak calculator

Calculate your team’s time leak.

10 people
6 hrs / week
$65 / hr

Annual time wasted

3,000 hrs / yr

Annual financial leak

$195,000 / yr

What a machine could recover (~75%)

2,250 hrs ($146,250 saved)

See the machines →

Fig. 05 — demo · custom AI system

The Brief Machine. Describe the problem — it designs the system.

The Brief Machine is listening. Pick a scenario below or describe your own operation.

Fig. 06 — demo · workflow automation

The Inbox Machine. 6 emails, zero humans.

00:00 0.0 hrs saved

    Fig. 07 — demo · data pipeline

    The Messy CSV Machine. Paste chaos, get structure.

    Fig. 08 — demo · integrations

    The Connector Board. One event, everywhere.

    Fig. 09 — straight answers

    Fair questions, honest answers.

    Can I use these in my own operation?

    Yes — that’s what open source is for. Every machine is MIT-licensed: run it, adapt it, integrate it. No permission needed.

    Do you build custom systems?

    Not as a service. Everything here is built in public and released free. If you need changes, fork the code — it’s yours.

    Can the machines work with tools I already use?

    The ones here connect to common tools — Slack, Gmail, Sheets, Stripe, Notion. Fork one and wire it to your stack.

    Do these cost anything?

    No. Free forever — no accounts, no API keys, no pricing page.

    Fig. 10 — the code

    Everything here is open source.

    Every machine on this page is free, MIT-licensed, and on GitHub. Fork it, run it, shape it to your operation — and if you build something good, contribute it back.