Atlas OS

"What does the Institute need from me today?" — generated from the real data in the panels below.

Loading live observation records from Atlas Community Service…

Look up a real turkey (Meleagris gallopavo) gene by symbol (Ensembl) — real Ensembl id, location, and external database cross-references. Returns nothing rather than guessing when the gene isn't found.

Resolve a real DOI to its real bibliographic record (CrossRef) — title, authors, journal, year. Returns nothing rather than guessing when the DOI isn't found.

Drag to pan. Scroll to zoom. Hover highlights neighboring nodes. Click node opens Context Panel tabs.

Real computed Mendelian probabilities from MFG-Atlas's genetics simulation engine, using Atlas-Ontology's live inheritance rule data — not a heuristic score. Recorded as work by the Institute's Data Scientist.

Real network of the platform's genotype/allele/phenotype ontology, computed fresh from Atlas-Ontology on every run — not a heuristic layout. Recorded as work by the Institute's Data Scientist.

Real Wright's inbreeding coefficient, kinship, and effective population size computed from the Institute's real bird pedigree (sire/dam records) — not a heuristic. Recorded as work by the Institute's Population Geneticist.

Real allele-frequency diversity across the Institute's registered varieties, computed from real baseline genotype records — not per-bird diversity (no per-bird genotype data exists yet). Recorded as work by the Institute's Population Geneticist.

Photos submitted from a breeder's corrected identification in the mobile app — confirm before they can be used for identification, or reject to discard.

Loading live validation results from Atlas Validation Service…

Real, append-only evidence of every capability that moved backward — never a guess about why, always a recorded reason.

Real spend and a real monthly USD cap on every AI-assisted capability call the Institute makes through the AI Gateway. A call that would push spend past the cap is blocked outright, never silently allowed through.

Every open decision across the Institute, in one place — strategic recommendations, Knowledge Evolution proposals, Institute Intelligence recommendations, and open escalations. Acting here is identical to acting in that item's own section below; nothing here decides anything itself.

Real cross-record analysis of the Institute's own accumulated data — which investigations/publications have real child records yet, and how real Institutional Memory decisions and disagreements group — computed live, never a fabricated correlation. Recorded as the Data Scientist's real work.

The recommended path through a working day, top to bottom: Morning Review (read the Executive Briefing and Institute Health below) → Priority Approval (Pending Reviews — acknowledge, dismiss, or route each strategic recommendation) → Research Execution (Run Daily Cycle to refresh briefing, priorities, and evolution detection; monitor Active Investigations and the Live Activity Feed) → Publication Review (Publication Queue) → Knowledge Review (Knowledge Evolution Queue — approve or reject each proposal) → End-of-Day Summary (re-check Institute Health and Recent Decisions before signing off). Every step below stays independently auditable — nothing here changes what an action does, only the order it's suggested in.

Real per-researcher activity — assigned/active work, elapsed time, and blocked state — from Atlas Presence. Click a researcher for their real-time context, operational summary, and to ask them a question.

Real researcher escalations (GET /escalations) — raised automatically by the autonomous execution cycle for stalled work, or manually by a researcher. Open escalations require executive attention; escalation itself is never auto-resolved.

The Institute's permanent record — every archived decision, disagreement, and outcome the Institute has recorded, queried live from Atlas Institutional Memory. Nothing here is ever edited or deleted; corrections are recorded as new entries that reference what they correct.

The Institute evaluating its own operational effectiveness — real, evidence-backed metrics computed from current Institute data, deterministic insights detected from real trends, and advisory-only improvement recommendations. Never invents new science; never automatically implements an organizational change. Also runs automatically every 6 hours via Institute Intelligence's own Cloudflare Cron Trigger.

Every active researcher, grouped by department. Assignment and work-product counts are real counts from Atlas-Research-Staff — not a computed "busy score." Click a researcher to open their workspace.

What this AI staff member has actually done — real assignments, work products, review decisions (theirs and ones they made on others' work), escalations they raised, and staff-to-staff requests. Every node and edge here is a real Atlas-Research-Staff record; nothing is scored or invented.

A short tour of what Atlas OS is and how it's organized — the same one shown automatically the first time this browser opened it.

The same real operator session used for Institute and Executive Experience write actions — signing in or out here applies everywhere.

Provider credentials, capability routing, and usage are configured in the AI Control Center — the only part of Atlas OS that may hold or display AI provider connection state. No Institute service talks to an AI provider directly.

Core connection and review credentials are managed in the Evidence Center Connection panel to preserve current review workflows.

The five real operator roles enforced by every Institute service. This is a reference table, not a live entitlements check — see Account & Session above for your current signed-in role.

Saved only in this browser's local storage — not sent to Atlas. Preferred and Avoided Traits affect Breeding Center scoring; the other fields are recorded here but not currently used elsewhere.

Adding or configuring a provider needs a real administrator or director account signed in — this is a separate thing from a provider's own API key below. The same session used everywhere else in Atlas OS; signing in here applies everywhere.

Leave the API key blank to add the provider unconfigured — you can set or rotate its credential later from its card above.

Every AI-assisted capability platform-wide, and which provider (if any) it's currently routed to. A capability with no route stays honestly "not configured" everywhere it's used — nothing is fabricated to fill the gap.