Filling The Missing Trust Layer: Why We Consolidated 10 Projects Under PAICE.work PBC

There's a critical gap that we've been talking about and building tools to bridge for months now. Only recently have others started to talk about it too.

We have AI models that can reason, code, write, and plan. We have agent frameworks that can chain those capabilities together. We have enterprises deploying AI across every function. What we don't have is the infrastructure layer that makes any of it structurally trustworthy.

Not trustworthy in the "we published an ethics statement" sense. Trustworthy in the "a CISO can verify this, a compliance officer can audit this, a regulator can evaluate this" sense.

That missing layer is what PAICE.work PBC exists to build. Last week, we formalized that commitment by consolidating 10 projects under the PAICE.work umbrella. Here's why.

The Problem: Capability Without Accountability

The AI industry moved from "can it do the thing?" to "deploy it everywhere" without stopping to build the structural conditions for trust. Consider what's missing:

For people: No behavioral measurement of how humans actually collaborate with AI. Knowledge tests and self-assessments measure what people say, not what they do. When a model hallucinates or overreaches, do your people catch it? Nobody can answer that question with data right now.

For organizations: No way to evaluate whether your digital presence is ready for AI agents that browse, scrape, and transact on behalf of users. Lighthouse measures human usability. Nothing measures agent usability.

For regulators: AI laws are proliferating across jurisdictions, from the EU AI Act to state-level legislation in the US. The professionals who must comply with these laws need more than a news feed. They need structured, current, jurisdiction-specific analysis they can act on.

For the agent ecosystem: No standards for how agents communicate limits, track skill versions, coordinate with each other, or prove where their capabilities came from. Every agent framework is building these ad hoc, internally, incompatibly.

These aren't four separate problems. They're four faces of the same gap: the trust layer doesn't exist yet.

What We've Been Building

Every project under PAICE.work PBC addresses a piece of that gap. None of them were conceived as a portfolio. Each one was built because the existing ecosystem couldn't provide it fast enough. Together, they form a connected system.

Measuring What People Actually Do

At the center is PAICE (People + AI Collaboration Effectiveness), the behavioral assessment that produces a 0-1000 score across five dimensions. It isn't a quiz. It observes how you respond to AI errors, overconfidence, and hallucinations during a real conversation. That behavioral ground truth is what separates PAICE from every knowledge test and self-assessment on the market.

The PAICE Assessment™ is free for individuals, and always will be. For those who want more support and to track their improvement over time, we now offer PAICE Pro. For teams and organizations, we have PAICE Packs (bulk purchase PAICE Pro codes) and PAICE Baseline. We provide structured assessment at every scale, from a single manager sharing it with their team to enterprise-wide cohort evaluations of 20 to 100 people each.

Providing the Context That Measurement Needs

A PAICE Score™ exists in context. If your organization's infrastructure isn't ready for AI agents, if you're subject to new regulations you haven't tracked, if the tools your team uses have changed capabilities since last quarter, that context matters.

Siteline evaluates how prepared an organization's digital presence is for AI agent interactions. As agents increasingly browse and transact on behalf of users, Siteline provides visibility into whether your infrastructure is ready. Start with a free scan and get immediate remediation suggestions, or go deeper with a full audit or consulting engagement.

AI Tool Watch tracks the evolving landscape of AI capabilities. For professionals making adoption decisions, it provides structured, current context. For PAICE itself, it provides the data needed to keep assessment rubrics current as the tools change.

Every AI Law tracks AI regulation across jurisdictions globally. It's free now, with paid tiers arriving soon for professionals who need customized analysis, monitoring, and deeper compliance support. If your job requires you to know what AI rules apply in your jurisdiction, this is built for you. For PAICE, it's what allows us to build niche products specific to the regulations in each jurisdiction.

Building the Open Infrastructure

The fastest-moving part of our work addresses what the agent ecosystem needs before it needs another platform: open, interoperable standards.

Graceful Boundaries defines how services communicate limits to humans and agents. When an AI agent hits a rate limit or capability boundary, there's no standard way to express that. Graceful Boundaries provides one. It's what Siteline evaluates conformance against.

Skill Provenance tracks versions across agent skill bundles. As agents accumulate capabilities, knowing which version of which skill produced which output becomes critical for auditing and compliance.

Knowledge-as-Code Template formalizes how domain knowledge should be structured so that both humans and AI agents can consume it reliably. Ontology-driven, version-controlled, machine-traversable.

Turnfile defines the SNAP protocol for peer multi-agent collaboration. When agents need to work together without a central orchestrator, Turnfile provides consent-based coordination that remains human readable and human-owned without being human-managed. In dynamic agentic workflows, it establishes who goes next, what context is shared, how conflicts are resolved, etc.

Skill A11y Audit ensures code produced by AI agents meets human accessibility standards. As more production code is AI-generated, this gap widens. This is the quality gate.

A tenth project is in development and will be made public within this second quarter of 2026. Your wild guesses are welcome!

These open projects are not competing standards. They are working implementations shipped because the existing ecosystem couldn't deliver fast enough, offered as open proposals for adoption.

Why One Organization

Each of these projects was built to solve a problem we hit while building PAICE. Graceful Boundaries exists because agents encountering limits needed a structured way to understand why. Siteline exists because there was no equivalent of Lighthouse for agent usability. EveryAILaw.com exists because the regulatory landscape moves too fast for static reference documents. Skill Provenance exists because managing dozens of AI skills across multiple products requires auditable versioning.

They were scattered across two businesses because that's how they evolved. The realization, recent and clarifying, is that they all serve the same mission and always did.

The thread through every project: building the structural conditions that make People+AI collaboration trustworthy, not just capable. When services communicate limits clearly, when skills can prove their provenance, when knowledge is structured for both human and machine consumption, when collaboration between agents follows a protocol rather than ad hoc wiring, the result is an ecosystem where trust is engineered, not assumed.

That's the mission of PAICE.work PBC. It was always the mission. Now the organizational structure reflects it.

The Business

PAICE.work is a Public Benefit Corporation. That's a structural commitment: every decision must serve the mission of making People+AI collaboration measurable, governable, and effective.

Three products generate revenue: PAICE.work, EveryAILaw.com, and Siteline.to. Each starts free and scales to enterprise, and is closed source. The open-source infrastructure projects are freely available for anyone to use, fork, and contribute to. Open standards win by being useful, not by being proprietary.

We're targeting regulated industries where AI collaboration isn't optional. These are among the hardest markets to enter and the stickiest once you're in. The professionals in these markets don't have the luxury of waiting to see how AI plays out. They need to be informed, assessed, and compliant now.

PAICE.work PBC currently has no direct competitors in behavioral AI collaboration assessment. That window won't stay open long. The infrastructure is built. What comes next is distribution.

What This Means for You

If you work in GRC, legal, or compliance: The combination of the PAICE.work assessment and EveryAILaw.com tracking gives you a defensible answer to "how ready is our team for AI?" and "what rules apply to us?" That's not a nice-to-have. It's what your board is asking.

If you're a CISO or CAIO: Siteline.to shows you how agent-ready your infrastructure is. PAICE.work shows you how collaboration-ready your people are. Together, that's a comprehensive AI posture assessment.

If you're an individual professional: Take the free assessment. Understand your own collaboration patterns before your organization mandates it.

If you're building AI agents: The open infrastructure projects are available for contribution. Graceful Boundaries, Turnfile, and Skill Provenance are early-stage and benefit from real-world implementation feedback.

If you're an investor aligned to this mission: We're raising to put these products into market. Let's talk.

Support the Open Infrastructure

The open-source projects above are freely available and always will be. Sponsorship via GitHub Sponsors keeps the specs evolving, the tests passing, and the tooling free for everyone.

All sponsored projects share the same tier structure:

Tier	Monthly	Name	What you get
1	$5	Supporter	Name in SPONSORS.md, sponsor badge
2	$15	Backer	Above + early access to spec drafts and changelogs
3	$50	Sustainer (Company)	Above + logo in README + quarterly update email
4	$200	Infrastructure Partner	Above + named acknowledgment in spec docs + direct issue channel

What sponsorship does not buy: SLAs or guaranteed response times, roadmap control or feature prioritization, private builds or exclusive access, or consulting and implementation support (available separately via Snap Synapse).

Graceful Boundaries

Fund the spec that teaches services to say "not right now" instead of crashing.

Graceful Boundaries defines how services communicate their operational limits to humans and autonomous agents — four conformance levels, from static limit pages to real-time agent negotiation. The reference implementation powers Siteline's compliance scanning. No existing standard covers this. Your sponsorship keeps the spec evolving, tests passing (131 and counting), and conformance tooling free for everyone.

Who should sponsor: API platform teams, AI agent framework builders, infrastructure companies building for the agentic web.

Skill Provenance

Know where your agent's skills came from, and whether they've changed.

Skill Provenance provides version identity and manifest tracking that travels with agent skill bundles across sessions and platforms. As agent ecosystems grow and skills get shared across Claude, Codex, and other platforms, provenance becomes a supply-chain security question. Your sponsorship funds cross-platform testing (22 core evals + 4 distribution evals), new platform integrations, and keeping the spec aligned with the rapidly evolving agent tooling landscape.

Who should sponsor: AI safety teams, agent platform builders, enterprises deploying custom agent skills.

Turnfile

The protocol for AI agents that disagree productively.

Turnfile is a protocol for LLM agents that need to disagree, negotiate, and build consensus without central orchestration — no boss agent, no voting, just structured turn-taking where agents surface tensions, propose resolutions, and converge on decisions. Tested across 11+ real multi-agent sessions with Claude and Codex. Your sponsorship funds protocol development, cross-platform agent testing, and building the coordination layer the multi-agent future needs.

Who should sponsor: Multi-agent system builders, AI governance researchers, teams building collaborative AI workflows without centralized control.

Knowledge-as-Code Template

The build system behind three live knowledge references.

Knowledge-as-Code Template is a reusable architecture for building structured, version-controlled knowledge bases — ontology-first design, zero external dependencies, Git-native data. It powers AI Tool Watch, Every AI Law, and Virtual Classroom Watch. Instead of a CMS or wiki, you define your domain as YAML ontology, write content as structured data, and the template generates HTML sites, JSON APIs, MCP servers, and discovery files from a single source. Your sponsorship funds template improvements that cascade to every project built on it.

Who should sponsor: Documentation teams, developer tool builders, anyone maintaining structured knowledge that outgrows a wiki.

AI Tool Watch

Plain-English AI capability tracking, verified by four models twice a week.

AI Tool Watch tracks capabilities, plans, constraints, and implementations across 12 major AI products. Every claim is verified through a four-model consensus cascade (Gemini, Perplexity, Grok, Claude). Live at aitool.watch with 87 tracked implementations and 125 bridge pages. Your sponsorship funds the verification pipeline, new product onboarding, and keeping the reference accessible (WCAG 2.1 AA compliant).

Who should sponsor: AI educators, product managers evaluating tools, enterprises making platform decisions, facilitators running AI training.

Drop-in accessibility audits for AI coding agents.

skill-a11y-audit is a portable agent skill that gives any AI coding agent (Claude, Codex, etc.) the ability to run WCAG 2.1 AA accessibility audits on web projects — combining automated scanning (axe-core, Lighthouse) with compliance mapping, manual check guidance, and structured reporting. Most AI agents can write code, but few can audit it for accessibility without guidance. Your sponsorship funds new check development, platform compatibility testing, and keeping the skill aligned with evolving WCAG standards.

Who should sponsor: Web agencies, design systems teams, accessibility consultancies, enterprises with compliance requirements.

What's Next

EveryAILaw.com adds paid tiers for jurisdiction-specific compliance analysis and monitoring
PAICE.work expands with jurisdiction-specific assessment variants powered by Every AI Law data
Siteline.to extends scanning to cover additional AI agent interaction patterns
Graceful Boundaries and Turnfile move toward formal specification status with community review
And a tenth project, currently in stealth, rounds out the portfolio

The trust layer has been missing. Not to worry, we're building it.

Take the PAICE assessment to see where you stand. Free, private, behavioral.

Get Involved:

Filling The Missing Trust Layer