Proof of Quality is moving into its next phase and how our community can interact with it will be different from how it used to. Our community will always have a place in the protocol.

An update on where Proof of Quality stands, what's changed, and what it means whether you build on PoQ, back it, buy it, or are part of the community around it.

The short version

PoQ makes the quality of AI work verifiable. We've introduced selectable assurance modes, so collateral is no longer required on every project, which evolves the original staking thesis and broadens the role of $SAPIEN. The model is now focused on high skill validation in domains where errors carry real cost. V1 launches in Q3.

Why PoQ exists

Every AI quality claim today rests on one party's word. An AI lab vouches for its training data, a vendor for its labels, an audit firm for its findings, an agentic system for its own decisions. There is no shared, verifiable record of who reviewed what, how, or why. The information age solved distribution but never solved verification, and that gap matters more as AI moves from generating content to taking action. When quality depends on internal QA, a marketplace badge, or a "we checked it," there's no way to reverify without redoing the review. PoQ closes that gap.

What PoQ does today

PoQ makes the quality of AI work verifiable across the data models learn from and the outputs they produce. Builders define what good looks like, qualified validators score against that standard, and consensus produces a tamper-proof record any customer, auditor, regulator, or end user can check without trusting Sapien and without redoing the review.

This is live rather than a roadmap promise: our first design partners have completed their validation runs, and we're opening access to the next cohort of early customers. Alongside that, we've introduced selectable assurance modes, so you can match the level of assurance to what a project needs, with collateral applied where it adds value rather than on every job.

How it works

The rubric defines quality, the consensus measures it, and the record proves it. The buyer authors the rubric, so PoQ measures against your standard rather than imposing one. Validators score with economic accountability behind their work, consensus settles the outcome, and the result is a permanent, portable record anyone downstream can verify on a single item. PoQ sits on top of the review you already run, complementing your eval, labeling, and audit tools, and you choose the reviewer source: your own credentialed pool, one of our partners, or a third party you designate.

Working with PoQ

We're running a design partner cohort now. A pilot means we co-author your first rubric, run a clean end to end batch on your data in week one, and meet weekly to fix what's broken as we go, with access to the SDK, MCP server, GitHub App, and audit bundle as each lands. The best fit is a concrete rubric you already use, one pipeline step that needs QA on AI work, and a downstream party who would value an independently verifiable record. 

For our community

A direct word for the people who've been here through the early phases. The move to selectable assurance modes is a real change, and the original "stake to validate on everything" thesis has to evolve with it. The short answer is that the role of $SAPIEN is broadening. SAPIEN is the preferred collateral in Collateral mode, it's required to operate agentic validator pools, and referring expert validators into permissioned pools earns rewards tied to their work. PoQ revenue also flows from the operating company back toward the protocol and its token holders, and as governance is introduced, holders will help decide how that revenue accrues value to SAPIEN, along with the other parameters that run the economic engine. Many of the questions you're asking now will increasingly be yours to answer.

We also want to be straight about who PoQ is for now. The old model, where people earned by completing basic weekly tasks, isn't coming back in its previous form. PoQ is built for high-skill validation in domains like security, robotics, and medical imaging. If you joined for task volume and daily earnings, PoQ may not be the right fit, and that is genuinely OK. We'd rather say so now than keep anyone waiting on something that isn't returning. The unstaking penalty several of you raised is one we're actively reconsidering, and we'll come back with specifics. We're also tidying up the Discord channels and will be far more present there over the coming weeks.

What's coming

Updated tokenomics will publish alongside the V1 launch, which we're aiming for in Q3.

Take the next step

The best place to start depends on where you sit. If you're building with AI where quality matters, and you see value in an independently verifiable record of how that quality was reviewed, send a rubric and 10 to 20 sample items to [email protected] and we'll come back within a week with a draft plan.

If you'd like to scope fit first, book a 30 minute call.

Keep an eye out for an upcoming AMA planned soon on our Discord.

Reach us at [email protected] or join the Proof of Quality waitlist for early access.