Why Synapti is a collective

The economics inverted

Work that used to require fifty people with the wrong architecture now requires five with the right one. Coordination collapsed before the work did. The bottleneck moved from execution to specification, from headcount to context, from hours to judgment.

The standard org structures were not designed for this. Pyramidal consultancies optimize for selling the labor of people they pay below market. Solo consulting optimizes for autonomy at the cost of capability and continuity. Both leave most of the value created on someone else's table.

The shape of the org has to change before the work changes.

What a collective actually is

Synapti is a small group of senior practitioners who share infrastructure, context, and ownership. Closer to a research lab than a firm: principal investigators running their own projects, drawing on shared tools and peer review, contributing to a common reputation. Closer to a guild than to a payroll: master craftspeople with their own commissions, who happen to share a workshop and standards. Closer to a jazz ensemble than to a hierarchy: members trade lead based on the project, not the title.

One word for that combination is collective. The mechanics matter more than the metaphor.

The mechanics

Everyone here has equity. Contribution flows into ownership through Slicing Pie, which tracks who contributed what (cash, time, deferred billing, IP) and converts the ledger into shares. Nobody guesses the cap table on day one and lives with the wrong guess for years. This works because it does not require us to know in advance whose work will matter.

Engagements take one of two shapes. Consulting at market rates. Or hybrid, reduced cash plus equity, when the project is one we want to be part of. Either is valid. We optimize for the outcome shipping, not for the hours billed.

We split capacity across four output streams: products (Decipon and what comes after), open source (UIM, ACS, ITP, and the rest of the synaptiai org), writing (field reports about what we built and what we learned), and client work. Most members touch two of those four in any given quarter.

What this is not

Not a freelancer marketplace. Not a co-working space. Not a YouTube channel with a Discord server attached. A collective is a working unit with shared standards, shared context, and shared upside. It needs care and curation, and it stays small for that reason.

We grow by endorsement, not by application volume. New members are sponsored by existing ones who have worked with them and stake their judgment on it.

What we believe

Models are easy. Operations are the product. Most AI projects fail for boring reasons: the workflow was never specified, inputs were never versioned, evaluation was an afterthought, costs went unmonitored, ownership lapsed at handover. The work that compounds is the work upstream of the model: spec, evals, governance, instrumentation. That is the work we do.

Integrity is a moat. The technology compounds for good or for ill depending on who builds the guardrails. We build the guardrails. The Influence Tactics Protocol that powers Decipon, the constitutional architecture work, the BDSK governance kit, the lucid epistemic audit: those are not side projects. They are the position.

Capable people working together with shared context outperform lone geniuses on almost every dimension that matters. The collective exists to make that multiplier concrete.