Human Atlas

What Was Built

Human Atlas is a public research workspace for exploring human contexts across life stages, occupations, and society — while keeping evidence, geography, uncertainty, and ethical boundaries visible. The repository description positions it as a human atlas for brainstorming sessions to create a product and for solo founder work.

The first public snapshot landed on 2026-08-17 with a standalone browser interface, generated occupation and society data packs, build scripts, and offline validation tests.

The Problem

Product and venture research often jumps from a persona sticky note to feature ideas without structured context: Which life stage drives the pain? Which occupation workflow matches the buyer? What civic or cultural constraints apply in Indonesia versus globally? Generic persona templates hide what is known, inferred, or still an evidence gap.

Human Atlas exists to give founders a dated, inspectable snapshot they can open offline, cite field by field, and rebuild when sources change — without pretending coverage is complete.

Why This Problem Is Difficult

Research spanning religion, ethnicity, migration, and public actors carries ethical risk if presented as exhaustive truth. Human Atlas explicitly documents:

• Named people restricted to dated public-record entries.
• Prohibited inferred religion or politics for private individuals.
• Directory coverage claims of dated open registry, not exhaustive.
• Recommendation boundaries that keep ai_hypothesis and evidence_gap fields visible in research but excluded from product recommendations.

Building this as a serverless static snapshot adds engineering constraints documented in the system architecture article.

Beginner Mental Model

Human Atlas is a research map book, not a CRM or survey product:

• Life answers "where in a person's journey might this need appear?"
• Work answers "which job family experiences this workflow?"
• Society answers "which cultural, civic, or household context shapes the need?"
• Colored evidence tags tell you whether to cite, validate, or keep researching.

Requirements and Constraints

Requirement	How the repo satisfies it
Open without setup	Single HTML file + sibling data directory
Reproducible snapshot	Dated manifests (build_date: 2026-08-17) with checksums
Honest coverage limits	directory_coverage.gaps strings per domain
Source traceability	sources registry with URLs, licences, and scopes
Validation before trust	Python validators + Node offline smoke test
GitHub Pages entry	index.html redirects to the atlas HTML

Architecture Summary

See the full technical breakdown in Human Atlas System Architecture.

Snapshot Contents (2026-08-17)

Layer	Scale	Notes
Life map	15 stages, 190 observations	Linked life-stage research
Work map	436 unit profiles	ISCO-08 backbone + ESCO/O*NET/KBJI supplements
Society map	312 nodes, 343 edges	Eight domains with domain-leader opportunity hooks
Sources	17+ society sources	Wikidata CC0 identifiers, Pew, Kemenag, IDEA, KPU, UNESCO, ISO, UN
Validation	All packs release_ready: true	Checksum + evidence gate scripts pass on checked-in snapshot

Evolution and Milestones

Date	Milestone
2026-08-17	Repository created; first commit with atlas HTML, occupation and society packs
2026-08-17	GitHub Pages index.html added for public hosting
2026-08-17	Validation scripts and offline smoke test checked in

Future work implied by the repository (not yet evidenced as shipped):

• Expanded public-actor directories beyond initial Indonesia examples.
• Additional society domains or deeper country packs.
• Rebuilt snapshots when source spreadsheets or PDFs update.

Key Decisions

Decision	Rationale
Static snapshot over live API	Reproducible research baseline for brainstorming sessions
ISCO-08 as Work backbone	Global occupation interoperability
Society hypotheses separated from facts	Prevents over-confident product recommendations
Offline file:// support	Founders can archive and share an exact dated bundle
Domain-level opportunity hooks	Links research nodes to validation experiments

Relationship to Other Projects

Human Atlas complements venture-building tools like Delivery Foundry — Foundry governs how software gets delivered with evidence gates, while Human Atlas structures who and in what context a product idea might matter. It does not share code with Foundry; the link is conceptual (research intake → mission brief → delivery loop).

Lessons Learned

1. Research tools need explicit uncertainty UI — four evidence statuses beat one "confidence score."
2. Dated snapshots enable diffable research — rebuild scripts make "what changed since August?" answerable.
3. Privacy rules belong in data, not README footnotes — manifest-level enforcement scales as packs grow.
4. Start with validation scripts on day one — checksum + smoke tests protect a 12 MB HTML file from silent drift.

Related

• Human Atlas System Architecture
• LLM Guardrails — analogous separation of advisory vs authoritative layers

Sources

• Repository: okfriansyah-moh/human-atlas
• Commits: 97f78f5 (first commit), 7d61cc0 (GitHub Pages index)
• README and manifests dated 2026-08-17