Chirality

Getting started

Three ways in. You own the result in all three.

Everything is free and open source under the MIT licence. What differs is how much of the structuring you do yourself.

Free · open source

Take it and run

Download the application and point it at your own folders, with your own models. No reduced edition, no trial period, nothing that reports back to us.

  • The whole application, exactly as we use it ourselves
  • A local model, your own API key, or an OpenAI login
  • Write your own agents and workflows from nothing
  • Source your team can audit, and fork if it ever needs to

Two things to know. macOS 15 or newer today, with Windows to follow, and this is a release candidate: it works, and it is still being finished in the open. It has a real learning curve; the assistant teaches you as you go.

Paid engagement

Work at scale, with us

When many workflows need to run at once, across many people and agents, we build the coordination system that lets them run concurrently without conflicting. Consulting, done properly, and not cheap.

  • A coordination plane stood up for your organisation, so concurrent work does not collide
  • Your workflows, references and checklists, shaped to your work
  • Agents written for your disciplines and your standards
  • Your people taught by doing the work
  • All of it left behind as files in your own repositories

Alberta and British Columbia for now. The work is done in the room, so we take clients we can sit with. Priced per programme, a few organisations at a time.

What the engagement builds

Coordination is the thing that gets expensive.

Not the work. What scales against you is everybody, person or agent, needing to know what is currently true while working at the same time. A coordination system is what lets many workflows run at once without conflicting, and standing one up is the substance of onboarding.

Below a certain size you do not need it

  • One person and an assistant, or a small team on one package at a time, are already served. The files are the record and the agents keep them current.
  • Take the application and the catalogue, and come back if you outgrow them.
  • What you outgrow is knowing what is true. Every run rebuilds the current state from status files and ledgers before work begins, and concurrent sessions cannot see each other.

What we set up when you have

  • Orientation becomes a query answered in under a second, with a citation for every claim.
  • Staleness is detected by checksum, not by somebody noticing.
  • Sessions declare where they are working, so a collision surfaces before it reaches your version control.
  • One live view of what is in flight and which decisions are waiting on you.

It is open source too, and it is the part you can delete. Removing the coordination plane costs you throughput and never costs you correctness; the files remain the only authority. It is in early development, so do not plan around it yet.

The return

The point is not doing today’s work cheaper.

Work you could not take on before

Every practice turns down work where the checking would eat the fee. Bring more of the checking in front of the same people and that line moves.

Complexity you can now hold

The harder a job is, the more of it is coordination, traceability and verification. That is the portion this addresses, so the advantage grows with difficulty.

An asset that outlives the engagement

What we build with you remains yours when we leave, and keeps improving as your people use it.

Why free and open source

Because owned capability is the whole proposition.

It would be incoherent to tell a firm to own its own knowledge and then keep the instrument on a lease. So all of it is free, and most firms will use it without ever speaking to us. We would rather pollinate the field than farm it.

The software has no geography. The engagement work is where we can sit with your people, which for now means Alberta and British Columbia. The problems are the same in Houston, Rotterdam and Perth.