Seneca — Negotiation Coach

Summary

Seneca — Negotiation Coach — regenerated from canonical/guides/tools/seneca.md. Do not hand-edit this page — update the canonical file and re-project. Status: live. Primary domain: none (internal only). Pipeline role: negotiation coaching bot — agent-facing.

Content

Purpose

Agent-facing negotiation coaching chatbot. 10x capability vs. field agents working from memory (per tool-stack-inventory). Coaching theme consistently emphasizes: "Persistently underused. Nuke in your pocket vs. fighting with a slingshot."

Current state

  • Status: Live
  • Auto-reads Olympus (since 2026-04-17) for live team context — 10x capability leap
  • Challenge: Underused by agents despite high capability

Where it lives

  • Code: Tools/Seneca/
  • Deploy target: Flask + Anthropic API (verify specific deploy URL — no .stoicagent.com subdomain listed)
  • System prompt: v3 (per parent CLAUDE.md)
  • Knowledge base: includes Fred's writing voice (pulls from Write Like Fred)
  • Multiple deploy versions exist (per parent CLAUDE.md) — clarify active version during refresh

Integrations

  • Anthropic API — core LLM
  • Olympus — auto-reads for team context (live since 2026-04-17)
  • Write Like Fred v4 — voice calibration baked into system prompt
  • Potentially: Athena/Iris ARGUS — negotiation context from deal risk

Related

  • Inventory row: canonical/business-state/tool-stack-inventory.md → "Seneca"
  • Brand voice: canonical/brand/write-like-fred-v4.md
  • Olympus briefing: canonical/guides/tools/olympus.md

Typical development touchpoints

  • System prompt iteration (currently v3)
  • Olympus integration — ensuring Seneca reads fresh published content, not cached
  • Agent adoption push — how to make Seneca the default negotiation aid (not an afterthought)
  • Knowledge base curation — what beyond Write Like Fred belongs here?
  • Deploy version consolidation (multiple versions exist)

Active WIP location

working/drafts/tools/seneca/ (create as needed during active development)

Open questions / gaps

  • Primary deploy URL / access pattern — is there a single canonical URL agents use?
  • Multiple deploy versions — which is the current production version? Are others deprecated?
  • Adoption metric — how do we measure whether an agent used Seneca before/during a negotiation?
  • Olympus sync freshness — if Olympus rebuild isn't wired yet (per Olympus briefing), is Seneca reading stale snapshots?
  • Agent-facing UX — is Seneca a web chat, Slack bot, something else? (Clarify — parent CLAUDE.md says "Flask + Anthropic API")