Competitive Intelligence
decisions/2026-06-26-ci-command-center-and-dedicated-bot.md
Decision: CI Command Center and dedicated bot
Context
Competitive Intelligence is a critical standalone project inside Chowmes, not a minor Athena reporting workflow. The current Algolia public-source CI pilot runs daily and weekly jobs, but the output quality is not yet good enough for the ambition: a best-in-industry competitive intelligence system that produces source-backed, action-oriented intelligence for Arijit and eventually for Algolia stakeholders.
Athena is the CEO/orchestrator for My OS. Athena should monitor CI quality, review strategy, assign fixes, and escalate important decisions, but Athena should not be the daily CI delivery agent in Telegram.
Decision
Competitive Intelligence will become its own dedicated operating product inside Chowmes, with a separate CI bot, CI dashboard, signal ledger, report archive, quality gates, and workspace-owned execution system.
Rationale
- CI has its own cadence: daily pulse, weekly digest, source monitoring, source repair, signal scoring, and pattern synthesis.
- CI has its own users: Arijit first, and potentially Algolia CMO, Product Marketing, Sales Enablement, Product, Partnerships, and executive stakeholders later.
- CI needs a distinct delivery voice and interface. Athena should supervise the division, not deliver every daily report herself.
- CI quality depends on evidence discipline, source health, delta tracking, confidence scoring, and action routing. Those need their own product surface and operational metrics.
- A dashboard is required because Telegram delivery alone cannot support history, pattern review, source coverage, weekly digests, action tracking, or stakeholder-facing executive use.
Scope
The CI system should include:
- Dedicated Telegram bot for CI daily and weekly delivery.
- Persistent signal ledger for sources, snapshots, deltas, signals, confidence, evidence, owner, and recommended action.
- Daily pulse that is concise, evidence-backed, and honest about quiet days.
- Weekly digest that identifies patterns, shifts, risks, opportunities, and recommended actions.
- Dashboard for report archive, signal search, competitor timelines, source health, source gaps, action recommendations, and owner routing.
- Quality gates for source coverage, claim support, confidence scoring, broken-source detection, hallucination resistance, and forced-narrative prevention.
- Athena oversight loop for quality review, strategic critique, and escalation.
Alternatives Considered
| Option | Why rejected |
|---|---|
| Keep CI as an Athena-delivered daily report | Blurs Athena's CEO role, overloads the main chat, and prevents CI from developing its own interface, memory, and operating metrics. |
| Treat CI as only a skill under Chowmes | A skill is enough for collectors and report generation, but not enough for a product-grade dashboard, bot, action workflow, and source-quality operating loop. |
| Build My OS Command Center first and include CI inside it | Risks burying CI inside a broader cockpit before the CI data model and quality bar are strong. CI needs its own product foundation first. |
| Create many CI agents immediately | Premature. The first priority is the ledger, dashboard, quality gates, delivery bot, and source strategy. Specialized agents come only when the workflow proves they need separate memory or permissions. |
Consequences
- Competitive Intelligence becomes a first-class workspace/product, not an Athena side report.
- Athena remains accountable for oversight, but CI gets its own delivery surface.
- The next CI work should prioritize operational truth, weekly cron repair, quality audit, dashboard research, dashboard IA, and dedicated bot design.
- The My OS Command Center remains a separate higher-level system. It should observe CI status and quality, but should not replace the CI dashboard.
- Future work must define access boundaries before giving Algolia stakeholders dashboard access.
Open Questions
- What should the dedicated CI bot be named and what voice should it use?
- Should the CI dashboard be hosted only on localhost/private tunnel first, or exposed through authenticated Caddy later?
- What is the first dashboard audience: Arijit only, Algolia internal stakeholders, or both with permission tiers?
- Which Algolia teams should action routing support first: Product Marketing, Sales Enablement, Product, Partnerships, or Exec?
- What metrics define "best in industry" for this CI system?