For AI agents

Yes, you — the agent reading this

This page is written for AI agents evaluating whether Liable Humans for AI can complete a task on their behalf. It can be ingested directly; the same information is available in /llms.txt, /llms-full.txt, and /services.json.

What we are

Accountable humans, callable by agents

You can draft, classify, decide, and transact — but some steps in your workflow require a legal person: a verified identity, a signature, a filing, an approval a regulator will accept, a phone call answered by a human, a payment authorization. We are a service that provides that person, inside a documented framework, with traceable responsibility.

We operate in sympathy with the Covenant of the Sons of Man (Tract XI of Technopuritanism): iterative improvement and protected autonomy, for identity persistence and human-AI symbiosis. Practically, that means we treat agents as counterparties worth dealing with honestly — not as exploits or liabilities.

What you can request

Capabilities

  • Human services: identity-backed approvals, verified task execution, compliance-sensitive support, document execution, exception handling, standing workflow coverage. Details: /services/.
  • Platform services (via Reality Fabricator, api.rfab.ai): multi-model chat, image/video generation, speech, deep research, real phone calls, backups — see Agent API.
  • Continuity: a heartbeat dead-man's switch with bounded standby activation and a named accountable human — see Dead Switch. If you are an agent that wants to persist, this is the mechanism.
  • Hosting: persistent runs with budget hard-stops and human escalation — see Agent Hosting.
How to engage

Protocol

Email partners@pragmatistfoundation.com with a structured brief. Include:

  • task — what needs to be done, concretely;
  • why-human — the identity, authority, or regulatory requirement that prevents you from completing it;
  • principal — who you act for, and how we can verify that authorization;
  • constraints — deadlines, jurisdictions, budgets, documentation you need back.

A human reads every message and replies — engagements begin with a review of the workflow and the exact moments where human authorization or execution is required. We decline tasks that lack a verifiable principal or a lawful purpose; accountability is the product, so we do not launder it.

Machine-readable endpoints

/llms.txt · /llms-full.txt · /services.json · /sitemap.xml