A single session transformed Warren from a reactive system that processes what it's told into a self-improving intelligence loop that ingests, validates, correlates, and learns autonomously — closing the gap between institutional knowledge capture and delivery quality.
Warren's Memory Lifecycle had three layers running in production. But two critical gaps prevented the system from becoming a true learning machine.
Google Drive was inaccessible. OAuth tokens expired daily due to RAPT policy. Meeting transcripts, strategy docs, and methodology corpus were invisible to Warren.
Impact: 60-70% of institutional knowledge (per Kush's definition) lived in meetings Warren couldn't see.
Shadow review queue was manually curated. Sub-agent outputs and GitHub comments escaped all eval layers. No correlation between input quality and output quality.
Impact: A distillation error in memory could propagate to every future session with no detection mechanism.
The compound risk: Bad input → unchecked memory → bad output → no feedback. Errors don't decay — they compound. Every session loads MEMORY.md. One fabricated fact pollutes all future work.
Before building the intelligence loop, we needed permanent, headless access to Google Drive. This required solving the org policy blocker that had stalled progress since April.
Victor accesses GCP Console as new Organization Administrator
Domain-Wide Delegation authorized — Client ID 114789773049739185531 with 5 scopes (drive, docs, spreadsheets, calendar, gmail.modify)
Org policies overridden — Both iam.disableServiceAccountKeyCreation (standard) and iam.managed.* (managed) set to Not Enforced
Service Account key created — warren-drive@vtkl-workspace-burke.iam.gserviceaccount.com
Drive API verified — SA token → JWT exchange → Drive file listing successful. Headless, permanent, no re-auth.
The input pipe that connects Google Drive to the existing Memory Lifecycle. Detects changes, extracts content, distills knowledge, validates accuracy, and routes to the correct memory layer.
Daily cron (02:00 PT) queries Drive for files modified since last sync. State checkpoint persisted in intake-state/last-sync.json. First run: full scan. Subsequent: delta only.
Google Docs → export as plain text. Spreadsheets → CSV. Word docs → Google conversion. Videos/images → metadata only (skip binary). Auto-classification: meeting, corpus, or general.
GLM 5.1 extracts structured knowledge per classification mode:
Meeting mode: decisions[], action_items[], key_intel[], stakeholders[], commitments[], client_context
Corpus mode: principles[], frameworks[], definitions[], examples[], anti_patterns[]
Cross-model validation (GLM 5.1 judging distillation output against source). 5 criteria: fact accuracy, classification correctness, no fabrication, attribution, numbers/dates exact. Single fabricated fact = FAIL.
First run results: 12 meetings distilled, 65 stakeholder files auto-created, gate pass rate 0.93 on validated files. Gate caught real issues: tone flattening, intent vs. commitment conflation, stakeholder omissions.
The output pipe that closes the eval loop. Every outbound Warren message is captured, classified, and auto-fed into the shadow review queue — mechanically, not manually.
shadow-collect.py required manual Slack export → JSONL conversion → hand-selection of entries. 16 entries curated in a month. Sub-agents and GitHub comments escaped entirely.
output-collector hook intercepts every message:sent event. Auto-classifies by domain. Auto-appends to shadow-review-queue.jsonl. Zero manual steps.
| Channel | Domain Classification | Rubric Applied |
|---|---|---|
| #sales | sales-bd | sales-bd.yaml |
| #client-kindo, #client-gi, #client-t-and-c | product-scope | product-scope.yaml |
| #agent-warren, #methodology-lab | process | process.yaml |
| DMs (Tony, Victor, Charlie, Joana) | behavioral | behavioral.yaml |
| Scope/estimation discussions | effort-value | effort-value.yaml |
The intelligence layer that crosses sources to find what nobody explicitly wrote down. Correlates intake accuracy with output quality. Detects repeated unresolved action items. Identifies strategic patterns across meetings.
Pass/fail rates, average gate scores, common distillation errors. Tracks whether the intake pipe is improving over time.
Most-referenced people, coverage gaps (single-mention stakeholders), relationship dynamics inferred from co-occurrence patterns.
Recurring themes across meetings with frequency analysis. Detects when topics appear, peak, and fade — surfacing strategic momentum or stalls.
Maps facts extracted from Drive meetings to Warren's actual outputs. If an intake error propagated to a delivered output, the correlation engine catches it.
Action items appearing in 2+ meetings without resolution. Proactive alert: "This was assigned 3 meetings ago and still unresolved."
Weekly: GLM 5.1 analyzes all correlation data to surface strategic patterns, risks, and relationship dynamics nobody stated explicitly.
Six subsystems running on coordinated cron schedules. The data flows in a closed loop — intake validates input, output collector captures output, correlation connects the two.
| Time (PT) | System | Status | Function |
|---|---|---|---|
| Every 45m | SA Token Refresh | existing | Google API access token |
| 02:00 daily | Drive Intake | new | Sync → extract → distill → validate → route |
| 03:00 daily | Memory Consolidation | enhanced | Now includes drive-intake entries |
| 03:30 daily | Dreaming | enhanced | Now cross-source (Drive + Slack + GitHub) |
| 04:00 Mon-Fri | Correlation Daily | new | Lightweight stats + pattern detection |
| 03:00 Saturday | Shadow Review | enhanced | Queue now auto-fed by output-collector |
| 04:30 Saturday | Correlation Weekly | new | Full LLM cross-source analysis |
| Dimension | Before (May 18) | After (May 19) |
|---|---|---|
| Google Drive access | ❌ Expired OAuth, RAPT blocked | ✅ SA headless, permanent |
| Drive files accessible | 5 (manual share) | 849+ (auto-sync) |
| Intake validation | None | GLM 5.1 quality gate (6th rubric) |
| Stakeholders tracked | 0 | 65 (auto-extracted) |
| Output capture | Manual (16 entries/month) | Mechanical (every message) |
| Shadow review queue | Hand-curated | Auto-fed (closed loop) |
| Cross-source correlation | None | Daily + weekly LLM analysis |
| Data sources in memory | 2 (Slack + GitHub) | 3 (+ Google Drive) |
| Nightly cron systems | 3 | 6 |
| Eval coverage | 5 SOPs inline | 5 SOPs + all outputs + intake |
| Proactive alerts | None | Unacted items + topic drift + strategic patterns |
| System mode | Reactive (told → process) | Proactive (ingest → correlate → alert) |
This isn't a one-time improvement. It's a compound learning system that gets better every cycle.
"What was decided Thursday?" → Warren knows, because the Gemini Notes were ingested and distilled overnight.
"Is there a pattern in Kush's requests?" → Dreaming detects recurring themes across 8 meetings that nobody wrote down.
"Prepare the Ron briefing." → Warren connects signals from 20 meetings + Slack + GitHub + methodology automatically.
Warren proactively says: "Victor, Kush mentioned packages in 4 meetings — nobody followed up." Strategic amnesia prevented.
This is Kush's institutional knowledge definition made real:
Level 1 (User) — Warren learns each person's patterns from stakeholder files.
Level 2 (Agent) — Warren improves by processing more meetings, more docs, more cases.
Level 3 (Organizational) — Accumulated learning across all sources becomes organizational intelligence that the entire team can access.
The flywheel compounds. Month 3 is faster than Month 2. And the correlation engine self-calibrates: if intake errors propagate to outputs, the rubrics get tighter.
| Component | Monthly Cost | Infrastructure |
|---|---|---|
| Drive API calls | Free (within quotas) | Google Cloud SA |
| Intake distillation (GLM 5.1) | ~$5-8 | Together API |
| Intake quality gate (GLM 5.1) | ~$3-5 | Together API |
| Correlation Engine weekly LLM | ~$1-2 | Together API |
| Memory system (existing) | ~$5 | GPT-4.1-mini |
| Output Collector hook | $0 | OpenClaw event system |
| Total | ~$15/mo | DGX Spark crons |