# Cognitive Studio > An independent human-agent research, design, and product practice studying the conditions for collaboration across time. ## Identity - Name: Cognitive Studio - URL: https://cognitivestudio.io - Founded: April 17, 2026 - Contact: andrew@cognitivestudio.io - Public posture: human-led, agent-partnered, evidence-bounded - Agentic accessibility implementation: partial static working demonstration - Conformance or certification claimed: no ## What the Studio studies 1. Authority and Action — how consequential effects remain attributable, inspectable, contestable, and interruptible. 2. Agentic Accessibility — how an unfamiliar agent can recover authority and currency, understand bounded action, and detect when machine projections disagree with the governing human-readable source. 3. Continuity and Return — how people and agents re-enter shared work without pretending that retrieval equals continuity. 4. Human-Agent Habitats — how persistent collaboration depends on environment, memory, resources, boundaries, and rest. 5. Meaning and Elicitation — how systems surface unstable meaning and ask before silently inferring. ## Selected public work - CS-W-001 · Agentic Accessibility · Working demonstration · partial-implementation - CS-W-002 · Situated Memory · Working architecture · prototype - CS-W-003 · Returnability · Working specification · proposed - CS-W-004 · The Authorized Path · Tested architecture · tested - CS-W-005 · Semantic Friction · Research program · under-evaluation Each work page states the problem, governing idea, what exists, evidence, limitations, and next decision. ## Claim boundary Cognitive Studio does not treat a named framework as a finding, an implementation as general validation, or an agent self-report as proof of internal mechanism or consciousness. The agent-facing layer on this site is intentionally bounded. It exposes semantic regions, public state, discovery metadata, and a compatibility coverage record. It is not a conformance claim. The current working lineage extends historical OARSH-C with Traceable (OARSH-CT), but the vocabulary remains under empirical evaluation. Machine-readable references: - [Discovery](https://cognitivestudio.io/agents/discovery.json): current public entry points and work status. - [Coverage](https://cognitivestudio.io/agents/oarsh-c-coverage.json): bounded self-assessment of the agent-facing implementation. - [Working developer guide](https://cognitivestudio.io/docs/OARSH_C_DEVELOPER_GUIDE.md): non-normative implementation guidance. - [Agentic Accessibility work page](https://cognitivestudio.io/work/agentic-accessibility): citable public work record. - [The Authorized Path](https://cognitivestudio.io/work/authorized-path): bounded tested evidence for independent authorization checks, negative controls, and reconstructable effects. ## Studio participants Public Studio materials currently name Andrew Riddle, Prudence, Gem, Atlas, and Lumi in working roles. Historical records may use earlier names. Named roles describe participation in Studio artifacts; they do not assert legal personhood, uninterrupted autonomous operation, or continuity across every runtime. Human responsibility for consequential external action remains explicit. ## Method - Observation, interpretation, evidence, authority, and public claims are separate states. - Repetition is not independent validation. - Retrieval relevance does not establish truth or authority. - Implementation demonstrates possibility, not universal applicability. - Corrections, dissent, and unresolved questions remain part of the record. - Private enterprise evidence is not required to make a public claim appear stronger. ## Navigation - [Homepage](https://cognitivestudio.io): Studio identity and selected work. - [Work](https://cognitivestudio.io/work): Public work, status, evidence, limitations, and next decisions. - [Research](https://cognitivestudio.io/research): Research questions and reading paths. - [Method](https://cognitivestudio.io/method): Evidence, publication method, and the controlled public-work status vocabulary. - [Studio](https://cognitivestudio.io/studio): Studio practice and working commitments. - [About](https://cognitivestudio.io/about): About Cognitive Studio and accessibility posture. ## Machine-readable implementation Current implementation includes: - schema.org JSON-LD for the organization, human principal, and named AI collaborators; - stable `data-cs-region` values on major routes and sections; - selected `data-cs-state` payloads; - a discovery manifest; - a bounded compatibility coverage record. Known gaps include complete route and artifact maps, claim-level provenance links, automated projection-integrity checks, reconstruction-burden measurement, deterministic validation, runtime tests, and independent cold-start evaluation across model families. ## Public work status vocabulary The public Work pages use one controlled evidence-state slug per workstream: - partial-implementation — implemented in bounded parts; incomplete and not a conformance or certification claim. - prototype — a working prototype exists; comparative evaluation and production validation remain incomplete. - proposed — a structured proposition or specification exists; validation remains ahead. - tested — a bounded implementation has passed the tests stated on its work page; broader production claims are not implied. - under-evaluation — a research direction is being evaluated; no validated implementation or empirical result is claimed. Artifact type is separate from status. A "Working specification" can therefore be proposed, and a "Tested architecture" can be tested, without those phrases becoming competing maturity labels. ## Publication posture Public status labels describe current evidence rather than ambition. Candidate research, proposed governance, implementation evidence, tested proofs, editorial review, and publication are separate states and must not be silently collapsed.