Dashboard/Competitive Intelligence
Project

Competitive Intelligence

Vault: myos   Path: Competitive Intelligence

index.md

Competitive Intelligence

Competitive Intelligence is an active My OS workspace and a first-class Chowmes product track.

Its first implementation is the Algolia public-source competitive intelligence pack. The goal is to prove the signal ledger, source coverage, daily pulse, weekly synthesis, dashboard, dedicated CI bot, and action-first reporting before generalizing into a reusable CI OS for any company.

Workspace Purpose

Collect, store, score, synthesize, and deliver competitive intelligence that is:

  • Source-backed.
  • Delta-aware.
  • Quiet-day honest.
  • Actionable by owner.
  • Reusable across future company packs.

Current Operating Model

public sources
  -> collector router: direct HTTP, Scout scrape, RSS/feed, Parallel discovery, manual-only
  -> SQLite signal ledger
  -> daily Telegram pulse
  -> weekly HTML synthesis
  -> source coverage and gap repair

Current Status

Active pilot, strategically elevated, Semantic Intelligence Layer v0.2 deployed, Argus CI profile active.

The Algolia pack exists and runs on a daily cadence. Weekly reporting, baseline filtering, runtime artifact preservation, the first data model, and a static dashboard artifact now exist. The roadmap has now been corrected: market truth and differentiation come before any further CI product build. The accepted direction is to build only a narrow Algolia-specific decision layer, not a generic CI platform.

Scout source work for the first CI acquisition slice is complete. Live Scout exposes opt-in acquisition metadata through the shared /scrape primitive plus a scout benchmark CLI. There is intentionally no /ci/scrape.

Scout's broader product state as of 2026-06-29: Scout is a controlled private-beta acquisition/evidence layer, not a public-launch product yet. CI consumes Scout only when the collector router chooses scout_scrape; CI still owns source configuration, signal scoring, semantic deltas, daily/weekly reports, and dashboard decisions.

Competitive Intelligence now consumes Scout as infrastructure through CI Collector Router v1. The live runtime supports direct_http, scout_scrape, rss_feed, and fallback collection. The latest enabled source health after the 2026-06-27 fallback-enabled run is 39 ok / 0 failed.

Algolia Competitive Intelligence is now also a standalone GitHub software repository: https://github.com/arijitchowdhury80/algolia-competitive-intelligence. Hermes/Chowmes remains the executor; the GitHub repo is the product-code source of truth.

The dashboard direction has been corrected to Decision-First Dashboard v1. The first screen should present the decision brief, action queue, material changes, stakeholder implications, trust state, and evidence. Raw signals are now supporting audit trail, not the primary product surface.

After review, that first dashboard direction was still too confusing. The current dashboard UX direction is a guided weekly brief: start with one recommendation, show only the next actions, keep proof behind each action, and move system health into secondary data-quality context.

After further review, the dashboard was corrected again to use actual archived CI reports only. The current local artifact is an attention queue built from the real sparse archive: 2026-06-17 daily/weekly, 2026-06-18 weekly, 2026-06-20 daily, and 2026-06-27 daily/weekly. No mock findings or fake action buttons should appear.

The accepted dashboard IA spine is now Executive Brief + Intel Queue: each finding must show what happened, why it matters, recommended response, and evidence before asking for action.

Dashboard publishing is now automated. After successful daily or weekly CI runs, Chowmes exports dashboard-ready static files, commits them to the standalone GitHub repo, and Vercel deploys from that push. The dashboard is no longer a manually refreshed artifact.

Semantic Intelligence Layer v0.2 is now live in the standalone repo and deployed into the Chowmes/Hermes workspace skill. The system now extracts typed semantic facts and semantic deltas before publishing intelligence. Raw hash changes, collector changes, baseline captures, and legacy changed case_study records are suppressed from decision reports unless they map to a validated semantic delta.

As of 2026-06-29, the mechanical CI cycle has been verified e2e through Hermes cron: daily and weekly jobs run, deliver through Telegram, write Markdown/HTML artifacts, and publish dashboard updates to GitHub/Vercel. Content quality remains a separate improvement track.

The dedicated CI agent is now Argus. Argus owns CI execution, analyst voice, daily Telegram delivery, weekly Telegram delivery, and run review. Athena supervises Argus and receives quality/escalation signals. The dedicated Argus Telegram bot is active, the Argus gateway is running, and Argus-owned daily/weekly cron jobs are scheduled at 9:00 AM ET.

Team

See workspace-team.

Current minimum team:

  • Athena: CEO/coordinator.
  • Argus: dedicated CI agent and active Telegram bot for daily and weekly CI.
  • CI Operator: runs and monitors the ledger/reporting workflow.
  • CI Analyst: evaluates signal quality and patterns.
  • Product Marketing Analyst: translates deltas into positioning and battlecard updates.
  • Sales Enablement Analyst: translates deltas into objection handling and field actions.
  • QA / Source Auditor: checks source coverage, broken sources, evidence, and unsupported claims.

Argus is now the first live CI Hermes profile. The remaining CI roles are still role definitions, not separate live Hermes profiles.

Runtime Assets

See runtime-assets.

The executable skill now lives in this workspace. Hermes/Athena runs it through cron wrappers and a compatibility skill link, while this workspace owns the product logic, source strategy, quality criteria, artifacts, and roadmap.

Canonical Notes

Current Next Moves

  1. Improve report quality and CMO-grade usefulness now that the mechanical cycle is healthy.
  2. Inspect and expand semantic_facts and semantic_deltas using real Constructor, Bloomreach, Coveo, Google Cloud, AWS OpenSearch, and OpenAI pages.
  3. Add persisted action outcomes and decision history.
  4. Add delivery tracking in bot_deliveries so Telegram deliveries are visible in the CI dashboard.
  5. Expand Argus's feedback loop so accepted/rejected recommendations update future scoring.
  6. Prepare stakeholder-readiness gates before any Algolia executive access.

All files (92)