10 to the 23 AI logo

Domain Atlas / Benefits navigation & public-facing chat

Case fileUnited States (pilot deployments)small deployment

Nava assistive benefits chatbot

What happened

Nava's assistive chatbot work pairs retrieval-grounded answers about public benefits with explicit design constraints: grounding responses in vetted policy sources, positioning the tool as assistance to navigators rather than an authority to applicants, and publishing its design rationale. It stands in deliberate contrast to public-facing chatbots that answered authoritatively and wrongly.

The sociotechnical reading

As a stylized model, Nava-class deployments show what the cautious corner of the design space looks like: bounded scope, curated retrieval, a professional operator between the system and the affected person, and published reasoning that makes external scrutiny possible. The contrast case is the public-facing chatbot that tells a business to break the law with full confidence. Same underlying technology; radically different sociotechnical exposure — which is the Atlas's recurring point.

The concepts used in this reading are defined in the Field Guide; the governance responses live in the Practice Library.

Grounding sources for this case

The same sources that ground this model organization in the PAN library — evaluations, government documents, investigative reporting, and advocacy documentation, each labeled by tier.

Seeing your organization in this case file?

The histories here are documented after the harm. Mapping a live deployment's pathways and pressures — before the incident report — is engagement work: intake, diagnosis, prescription, and monitoring, with every limitation stated.