Competitive Intelligence OS Notes

This note consolidates the competitive-intelligence material into the Competitive Intelligence workspace.

Current State

The Algolia competitive research v2 baseline has been rebuilt as a public-source competitive intelligence engine. The workspace owns the executable source, logic, artifacts, and quality model; Chowmes/Hermes owns runtime execution and delivery.

Pattern:

collect -> normalize -> score -> store -> synthesize -> deliver

Implementation Homes

• Local Chowmes project: /Volumes/Data/Dropbox/AI-Development/Personal/ChowMes
• Hermes skill: /opt/data/knowledge/obsidian/MyOS/Projects/Competitive Intelligence/skills/competitive-research
• Hermes ledger: /opt/data/knowledge/obsidian/MyOS/Projects/Competitive Intelligence/artifacts/competitive-research/ci.sqlite
• Hermes compatibility link: /opt/data/skills/competitive-research
• Daily wrapper: /opt/data/scripts/competitive-research-daily.sh
• Weekly wrapper: /opt/data/scripts/competitive-research-weekly.sh
• Worklog: 2026-06-18-competitive-research-v2-worklog

What Exists

• SQLite signal ledger.
• Daily Telegram pulse.
• Weekly HTML synthesis.
• Direct-source collectors for changelog, docs, press, blog, product, case study, analyst, forum, protocol, and roadmap surfaces.
• Source-backed action-first output model.
• Quiet-day handling so reports do not manufacture narratives from baseline/no-change data.

Content Quality Rule

The data must dictate the text.

If a weekly run has baseline captures and no meaningful changed signals, the report should say that plainly. It should not force an AI visibility, positioning, or market-shift story from zero supporting signals.

Known Reliability Gaps

Some public sources block or fail server-side fetching and need repair or replacement before quiet days can be interpreted as real market quiet.

Known examples from the 2026-06-18 rebuild:

• Gartner search/product MQ: 403.
• Lucidworks blog/product: 403.
• Yext product: 503.
• Model Context Protocol blog: 404.
• CMSWire: 500.
• OpenAI blog: 403.
• Perplexity blog: 403.
• Algonomy blog: 404.
• Elastic release notes: 404.
• Reddit forums: 403.

These should be treated as source coverage gaps, not competitive silence.

Generic CI OS Direction

Do not generalize too early.

Run the Algolia pack for 1-2 real daily cycles first. Then decide:

• Which source types produce useful signals.
• Which sources are mandatory in every company pack.
• What belongs in Hermes skill logic versus a backend app.
• Whether generic CI OS packaging should include shared engine, company profile packs, report templates, Telegram delivery rules, and optional paid/internal integrations.