[Epic] Product & tech track — buildable items from the dimensional diagnosis

**Epic C · Product & tech track** — buildable items from the dimensional diagnosis.
_Derived from the fact-checking cycle analysis (v2, 2026-07-03). Market/organization/capital strategy is intentionally out of scope._

# 🎯 Strategic Context

## Why Now?
> _Owner: CTO_

The instrumentation (Epic A) and process (Epic B) tracks fix how we *measure* and *run* the pipeline. This track builds the product and infrastructure that let the operation **scale beyond the founders** and start converting institutional interest into sustainable revenue. Several items are hard blockers on growth we're already pursuing: onboarding a hub costs 2–4 weeks of founder time, there is **no lead-source attribution at all**, there is **no payment path**, and MongoDB cost **doubled to ~USD 180/mo in Mar 2026** with no monitoring. These don't wait for the metrics work — they gate the next stage of the company.

## Problem Statement
> _Owner: CTO, Design Lead, PE_

#### For Fact-Checking Teams (front-line reviewers):
Onboarding is founder-dependent and the usage documentation gets explicit "confusing" feedback, so new reviewers and orgs stall before they're productive. Stuck tasks have no automatic owner, so work silently orphans when a volunteer disappears. Live collaborative editing (TipTap/Yjs) is unstable, interrupting the actual review work.

#### For Platform Operators & Institutional Partners (hubs):
Operators can't attribute where partners come from (no UTM/landing pages), can't safely let a new hub publish publicly before it's proven (no trust graduation), and can't transact (no PIX/boleto, invoice, contract, pricing page, or trial). Governance features that Epic B's process work depends on — SLA + auto-owner, orphan alerts — don't exist yet. Silent test-coverage regressions don't break the build, and infra cost is unmonitored.

#### Impact:
- **Lead attribution: 0%** today — every acquisition decision is blind.
- **Onboarding: 2–4 weeks of founder time** per organization — the primary scaling ceiling.
- **Monetization: no path to a single paid conversion** — revenue is structurally blocked.
- **Infra: ~USD 180/mo MongoDB (2× since analysis window)** with no alerting; silent CI coverage regressions.

## Solution Hypothesis
> _Owner: CTO, Design Lead, PE_

Build the missing product surfaces and infra in priority order, reusing what already exists: Unleash feature flags enable **namespace trust graduation**; `cycleMetrics` (Epic A) hosts **computed telemetry fields**; assistant-ui powers a **self-served onboarding copilot**; ClaimReview structured data revives dormant **organic SEO**. **Evidence:** the four 🔴 items (UTM, SLA+auto-owner, docs redesign, payment) each unblock a concrete, already-active growth motion and none require new instrumentation to begin. **Assumption:** self-service can absorb enough onboarding to remove the founder from the critical path.

## What We're Claiming (Narrative)
> _Owner: CTO_

A new fact-checking organization can now sign itself up, learn the platform through a guided copilot instead of a two-week founder walkthrough, start in a safe internal-publishing mode, and graduate to public publishing once its quality proves out — then pay us with a Brazilian invoice and contract, all without a founder in the loop. Behind the scenes, stuck work reassigns itself, we know which channel every partner came from, and our infra cost and test coverage are watched instead of discovered. This is the track that turns Aletheia from a founder-run project into a product a hundred hubs can adopt on their own.

## Success Outcomes
> _Owner: CTO, Product Ops, PE_

Targets are benchmarked to current baselines (mostly zero — these are net-new capabilities).

| Outcome | Type | Target | Why Realistic | Measurement | Comments |
| --- | --- | --- | --- | --- | --- |
| Lead-source attribution | Quant | 100% of new sign-ups carry a UTM/source | Baseline 0%; standard UTM plumbing | Umami + DB source field (TECH-3) | Unblocks all channel decisions |
| Founder-free onboarding | Quant | An org reaches first published check with 0 founder-hours | Copilot + docs redesign remove the walkthrough | Onboarding funnel (TECH-2 / INST-7) | Baseline 2–4 founder-weeks |
| Backlog auto-ownership | Quant | 100% of tasks idle beyond SLA get an auto-owner/alert | Backs PROC-2b; INST-4 provides the state | Job logs; `lifecycle` transitions | TECH-4/5, needs INST-4 |
| First paid conversion | Quant | Payment path live; ≥ 1 institution transacts | Baseline: no path exists | Payment/invoice records (TECH-11) | Revenue unblocked |
| Infra cost monitored | Quant | Alert on MongoDB spend; hold/reduce from ~USD 180/mo | Baseline doubled, unmonitored | Cost dashboard (TECH-19) | Prevent silent cost creep |
| CI protects coverage | Quant | 0 silent coverage regressions merged | `coverageThreshold` gate is deterministic | CI status (TECH-15) | Baseline: regressions pass today |

---

# 📋 Scope and Implementation Plan
> _Owner: CTO, Product Lead, Design Lead, Engineering Lead_

### Epic Release Timeline
Mostly parallel to Epics A/B. The 🔴 items can start immediately; TECH-1/4/5 gain full value once their instrumentation dependency lands. Grouped by theme; priority inherited from the dimensional diagnosis.

**Telemetry & product data**
- [ ] #2511 — TECH-1 · `review_time_minutes` + `accuracy_binary` computed fields — 🟡 (needs INST-2 #2492)
- [ ] #2512 — TECH-2 · Committee activation funnel as measured stages — 🟡
- [ ] #2513 — TECH-3 · UTM tracking + per-channel landing pages — 🔴

**Throughput & governance (features behind PROC-2)**
- [ ] #2514 — TECH-4 · SLA + automatic task owner — 🔴 (needs INST-4 #2494; backs PROC-2b)
- [ ] #2515 — TECH-5 · Orphan-task alerts + automatic redistribution — 🟡 (needs INST-4 #2494)
- [ ] #2516 — TECH-6 · Trust graduation per namespace (Unleash flags) — 🟡

**Self-served onboarding**
- [ ] #2517 — TECH-7 · Self-served onboarding copilot (assistant-ui) — 🟡
- [ ] #2518 — TECH-8 · Redesign the usage documentation — 🔴
- [ ] #2519 — TECH-9 · Reproducible setup/self-hosting guide — 🟡

**Discovery & monetization**
- [ ] #2520 — TECH-10 · Reactivate SEO via ClaimReview structured data — 🟡
- [ ] #2521 — TECH-11 · Payment integration (PIX/boleto + invoice + SaaS contract) — 🔴
- [ ] #2522 — TECH-12 · Public pricing page — 🟡
- [ ] #2523 — TECH-13 · Trial without credit card — 🟡

**Stability & infra**
- [ ] #2487 — TECH-14 · Frontend RBAC unit tests (`permissions.ts`) — 🟢
- [ ] #2524 — TECH-15 · `coverageThreshold` as a CI gate — 🟢
- [ ] #2525 — TECH-16 · Stabilize live collaborative editing (TipTap/Yjs) — 🟢
- [ ] #2526 — TECH-17 · Reassess AWS SDK v3 migration — 🟢
- [ ] #2527 — TECH-18 · `replicas ≥ 2` + readiness probe — 🟢
- [ ] #2528 — TECH-19 · Monitor/optimize MongoDB cost — 🟡

**Sequencing & dependencies:** TECH-1 → INST-2. TECH-4/5 → INST-4. TECH-4 is the mechanism behind PROC-2b. TECH-6 (namespace trust) enables safe hub activation for TECH-11's paid rollout.

> 💡 For ideas outside this scope, open a feature request so we can track and revisit them in a future iteration.

---

# ❓ Outstanding Questions
> _Owner: CTO, PE, Lead Engineer, Design Lead, Product Ops, Data Lead_

**Q:** Which payment provider(s) cover PIX + boleto + nota fiscal with the least integration burden, and who owns the SaaS contract template?
**Q:** Is self-hosting (TECH-9) a supported product tier or a partner accommodation? It changes scope significantly.
**Q:** What is the pricing model (per-seat, per-hub, per-check)? TECH-12/13 depend on it.
**Q:** Does the onboarding copilot (TECH-7) reuse the existing Copilot module, and what is its content source of truth?
**Q:** What is the acceptable MongoDB cost ceiling, and which collections/queries drive the 2× increase?

---

# ⚠️ Risks
> _Owner: PE, Lead Engineer_

| Risk | Likelihood | Impact | Mitigation |
| --- | --- | --- | --- |
| Payment/tax integration (PIX, nota fiscal, contract) is heavier than scoped | High | High | Spike a provider evaluation first; treat TECH-11 as its own mini-project with legal input |
| Trust graduation misconfigured → a hub publishes publicly before it's ready | Medium | High | Default new namespaces to internal-only; require explicit quality gate to graduate (TECH-6) |
| Self-served onboarding degrades review quality vs founder walkthrough | Medium | Medium | Pair copilot with namespace trust graduation; monitor first-check quality |
| Auto-redistribution creates notification noise / reassignment churn | Medium | Medium | Threshold + grace period; make alerts actionable, not chatty (TECH-5) |
| MongoDB cost optimization risks query regressions | Low | Medium | Profile before changing; pair with the new CI coverage/perf gates |