OB1

AI work memory compiler

Turn one good AI session into memory the next agent can use.

OB1 is for serious work that should not vanish when the chat closes. It captures sources, decisions, constraints, failures, and proof objects so the next run starts with operating memory instead of a blank prompt.

See proof objects Read the agent file
Session compiler
memory packet ready
Raw session 14 sources
  1. Transcript

    Pricing page does not answer overage rules or buyer fit.

  2. Decision

    Make the checker the hero, not a side widget.

  3. Constraint

    Use product-like proof surfaces, not decorative panels.

  4. Failure

    Previous site looked like a generated SaaS template.

01 Capture evidence

Files, tabs, transcripts, screenshots, links, tool output.

02 Extract rules

What should future agents preserve, avoid, verify, or reuse.

03 Write memory

Durable Markdown packets with provenance and next-run use.

04 Ship proof

Public artifact, audit trail, workflow, site, dashboard, or kit.

Compiled output ob1.md 
## Next-run brief
Goal: rebuild the domain surface
Pattern: prove with product UI

Keep:
- source maps
- design constraints
- visual self-critique
- verification screenshots

Avoid:
- generic card grids
- abstract AI claims
- decorative dashboards

Proof object:
pricing.md crash-test lab
Reusable in future sessions yes
The waste Every uncaptured session makes you reload context by hand.
The asset Captured rules make future AI work faster and safer.
The proof Memory should leave visible artifacts you can inspect.

The loop

OB1 makes AI work reusable by default.

The goal is not to save every chat. The goal is to preserve the few pieces of work that should influence the next task: sources, decisions, constraints, failures, proof, and the shape of a finished artifact.

Proof objects

The memory has to show up somewhere real.

OB1 is useful only when it produces surfaces that future humans and agents can inspect, reuse, or test.

Live

Token Burn

Instrumentation for the new work surface: how much AI compute is moving through real operator workflows.

 Build now

pricing.md

A pricing crash-test lab and machine-readable commercial file for buyer agents.

 Kit

Project Room Builder

A source-inventory and verification workflow for turning messy files into grounded work.

 Kit

Creator Operating System

A reusable path from raw ideas and notes into source-backed, voice-consistent public work.

Playbooks

Reusable operating kits for agent-mediated work.

Playbook Job Artifact
Project Room Builder Stop asking AI to write from chaos. Source inventory, duplicate log, missing context, grounded draft. Agent-Readable Website Make a site legible to humans and AI systems. Offer map, claims inventory, pricing.md, llms.txt, agent summary test. AI Usage Cockpit Make token burn, context load, and workflow cost visible. Dashboard, session autopsy, workflow leakage notes. OpenBrain x OpenClaw Memory Turn agent runs into governed operating knowledge. Recall contract, write-back contract, provenance, review queue.

Services

Turn one messy workstream into a reusable system.

OB1 services are scoped around visible artifacts: files, sites, dashboards, source rooms, and operating notes. The point is to leave behind a workflow the next agent can use.

Agent-readable audit

Pricing file, website summary test, offer clarity map, and recommended fixes.

Project-room sprint

Turn a messy source folder into a source inventory, draft plan, and verification system.

Operating memory design

Define what agents should recall, write back, confirm, reject, and reuse.

Next action

Start with one proof object.

Send a site, folder, pricing page, transcript bundle, or recurring workflow. The first job is to make one piece of work compound.

Start a proof object