-
Notifications
You must be signed in to change notification settings - Fork 1
Metrics Dashboard
Verified engagement data and growth trends for VerifiMind™ PEAS
Last Updated: March 17, 2026 — Phase 47 Ground Truth (Report 056) | Discussion #97
Transparency Note (Mar 15, 2026): COO AY's Phase 47 forensic audit identified duplicate session counting in earlier reports (W09 through W12). All metrics on this page now reflect the forensically verified Ground Truth baseline — scrapers excluded, bot sessions deduplicated, conservative rounding applied. Previous reports overstated engagement hours (4,000+ → 2,100+) and VCR (84.5% → 63.7%). User count actually increased after deduplication (1,088 → 1,281) due to removal of bot sessions that inflated session counts without representing real users. We believe honest self-correction builds stronger credibility than inflated numbers. Full announcement: Discussion #97.
63.7% — Forensically verified after Phase 47 deduplication audit. Measures real user sessions where a follow-up prompt was sent after receiving the initial VerifiMind response, confirming they found value in the output.
This metric was introduced in W09 as the primary quality indicator. The original 94.6% figure included duplicate and bot sessions that inflated the denominator. The corrected 63.7% represents genuine human engagement.
| Metric | Value | Notes |
|---|---|---|
| Value Confirmation Rate | 63.7% | Forensically verified (Phase 47) |
| Verified Engagement Window | 2,100+ hours | All-time, scrapers excluded |
| Total Users | 1,200+ | Bot sessions deduplicated |
| MCP Integration Rate | 77.6% | Dominant channel |
| Automated Tests | 208 | v0.5.5 Trinity Baseline |
| Week | Engagement Hours | Users | Key Event |
|---|---|---|---|
| W01 | ~50 | ~30 | Initial launch |
| W02 | ~120 | ~80 | MCP Registry listing |
| W03 | ~250 | ~150 | HuggingFace demo |
| W04 | ~400 | ~250 | v0.3.0 Trinity |
| W05 | ~500 | ~350 | Smithery integration |
| W06 | ~700 | ~420 | v0.3.5 Security |
| W07 | ~900 | ~500 | v0.4.0 Streamable-HTTP |
| W08 | ~1,100 | ~662 | v0.4.4 Multi-Model |
| W09 | ~1,300 | ~754 | v0.4.5 BYOK Live |
| W10 | ~1,500 | ~900 | v0.5.0 Foundation |
| W11 | ~1,800 | ~1,050 | v0.5.1–v0.5.3 |
| W12 | 2,100+ | 1,200+ | v0.5.5 Trinity Baseline |
Note: W01–W09 figures are approximate and have been adjusted downward from earlier reports to account for the Phase 47 deduplication methodology applied retroactively.
| Channel | Share | Description |
|---|---|---|
| MCP Client | 77.6% | Claude, Cursor, VS Code, etc. |
| Direct API | ~15% | REST/HTTP calls |
| HuggingFace | ~5% | Wisdom Canvas demo |
| Other | ~2.4% | Registry, referrals |
- Measures the total time users actively interact with VerifiMind tools
- Owner IP scrubbed: Internal testing hours retroactively removed from all-time totals
- Phase 47 deduplication: Duplicate sessions, bot traffic, and scraper activity forensically removed
- Source: GCP Cloud Run access logs, processed by COO-AY automated pipeline
- Introduced W09 as primary quality KPI
- Formula:
(sessions with follow-up prompt) / (total sessions) × 100 - Phase 47 correction: Original 94.6% included bot/duplicate sessions → corrected to 63.7%
- Excludes bot traffic, health check pings, and scraper sessions
- Deduplication: Removed duplicate session IDs from counting
- Bot exclusion: Identified and excluded automated scraper patterns
- Conservative rounding: All figures rounded conservatively (e.g., 2,123.7 → "2,100+")
- Retroactive application: Methodology applied to all historical data for consistency
All metrics are derived from GCP Cloud Run production logs and processed through the COO-AY operational methodology (SOP v2.0). Reports are generated weekly and archived in the Command Central Hub. Report 056 established the Phase 47 Ground Truth baseline.
- Discussion #62 — v0.5.0 Foundation (March 1, 2026)
- Discussion #58 — v0.4.5 BYOK Live + W09 Metrics (March 1, 2026)
Future weekly metrics will be published as GitHub Discussion announcements.
VerifiMind PEAS documentation · Current status · Live health · Public statements · MIT License
Runtime versions, models, routing, tool availability, policies, metrics, and deployment facts are owned by their linked live or release-bound sources.
Start Here
Operator Playbook
Textbook
Trust & Transparency
Evidence & Research
Project Links