Product.ai / Join / Projects / CEO-Bottleneck Agent Workflow — automate one workflow currently bottlenecked on Michael's bandwidth
Project Open to Alpha Team

CEO-Bottleneck Agent Workflow — automate one workflow currently bottlenecked on Michael's bandwidth

Identify one workflow currently bottlenecked on Michael's bandwidth — Slack inbox triage, Gmail processing, meeting prep, decision-support synthesis, signal classification, claim curation, or another concrete pattern. Build the agent harness that absorbs it: a sub-agent orchestration with worktree isolation if applicable, oracle-separated verifier where applicable, named external truth anchors between phases (per CLAUDE.md §5.2 rules 7-10), and a Day 1 ratchet pattern that improves with each iteration. Output is a workflow that produces real value in production — Michael notices a meaningful reduction in the bottleneck during the trial.
Project Overview
Discipline
ai-systems-associate · founders-associate · AI Systems — AI Engineer
Duration
2 weeks
Compensation
Your stated freelance rate
Surface
Cortex · Engineering · Product.ai
Kernels
cortex · engineering · productai
Outcomes
team-velocity · team-visible · ceo-gravity
Tier
Applied
Alpha Team
Open to alpha members who want to take this on
Tooling
Claude Code or Co-work

Why we want this done

Founders Associate / AI Systems Associate exists to absorb CEO-bandwidth-bottlenecked work into the system. Spencer Seay is the strongest current AISA candidate — Stanford IED, McMaster-Carr AI agent built end-to-end. The work pattern is the leverage move. Today, Michael personally triages Slack inbox, processes Gmail, preps for meetings, classifies signals, curates claims, syncs cortex, runs cortex-update at end of session. Many of these have partial automation; few have agent harnesses sophisticated enough to actually reduce Michael's load. The candidate picks one and goes deep — applies multi-agent physics correctly, ships the harness, and produces an artifact that materially shifts CEO bandwidth. AISA hires fail when they ship "agentic-looking" workflows that produce more cleanup work than they save; AISA hires win when they ship harnesses that respect the Stripe Theorem and the deterministic-gate discipline.

Scope

  1. Survey CEO-bandwidth-bottlenecked workflows — Slack inbox, Gmail, meeting prep, claim curation, signal processing, decision-support synthesis
  2. Pick one (the candidate proposes; pressure-test with Michael)
  3. Map the current architecture — what's manual, what's partially automated, what's the actual bottleneck
  4. Design the agent harness — sub-agent fan-out, oracle separation if applicable, deterministic gates, external truth anchors between phases
  5. Build it — using Agent tool with isolation: "worktree" for any multi-file work; respect CLAUDE.md §5.2 rules 7-10
  6. Wire it into Cortex / AIOS as appropriate
  7. Run it through at least one full cycle during the trial — measure bandwidth saved
  8. Document the harness — what it does, what it doesn't, what fails, how to improve

What success looks like

  • The harness ships and runs through at least one full cycle in production
  • Michael notices a meaningful reduction in the bottleneck (the candidate captures the qualitative signal explicitly)
  • The harness respects multi-agent physics — deterministic gates, oracle separation, named external truth anchors per phase
  • The 24-hour archaeology ritual fires on any failure during the trial (race-class entry + DEPENDENCY-PAIRS.md update + enforcement gate)
  • The harness compounds — by trial end, the candidate has authored at least one ratchet improvement based on observed failure
  • The documentation is concrete; another AISA could pick up the harness without re-asking

References

references.md
Cortex CLAUDE.md §5.2 rules 7-10 (Multi-Agent Engineering Physics)
axioms/Engineering/Multi-Agent Physics (2026-04-18).md
AI Engineering Phase 3 briefing axioms A4 (Stripe Theorem), A5 (External Truth Anchors), E1 (Engineer-Plus-Agent-Fleet)
Existing CEO-side workflows: /cortex-update, /chief-of-staff, slack-inbox skill, gmail-inbox skill, meeting-intel skill
Spencer Seay's McMaster-Carr AI agent as builder-archetype reference

Constraints

  • Claude Code as primary substrate
  • Sub-agent dispatches MUST include identity context + session grounding (>200 chars, identity markers) per the subagent-grounding-guard
  • Probabilistic discipline alone does NOT satisfy the brief — every shared-surface write paired with deterministic invariant check
  • IP separation: application-layer workflows in scope; methodology paths (aios-methods/_tools/arc-autopilot/) are out of scope
  • The 24-hour archaeology ritual applies to any failure during the trial
  • Quality-first cost-aware: any sub-agent fan-out follows model_router.py routing discipline
  • Privacy-respecting trace handling — sensitive CEO surfaces (Slack DMs, Gmail) require explicit handling protocol
Apply
01

Read the Codex (10 min)

The operating principles we work by. If they resonate, the rest of this will land. Open the Codex →

02

12-minute video screen

Hireflix, async. Questions are calibrated to this project specifically.

03

Chemistry call (30-60 min)

Direct call with the CEO. Strategic alignment and mutual fit. No problem-solving exercise.

04

Project begins within 2-3 weeks

1099 contractor agreement, NDA, paid at your stated rate. Day 1 in Santa Monica.

Alpha Team members can take this project without the screen-and-call sequence. Reach out via the Alpha Team channel.

Apply for this trial Browse other projects