Skip to main content
RECEIPTS PACK — STEPS 12 + 13

Phase 2 Plan + Beta Process

Phase 2 Stage 1: Packaging + External Proof. 3-dev beta onboarding. Definition of done.

Phase 2 Stage 1 — Deliverables

What is MemoryWell (1-pager)
Terminology Glossary
Architecture Diagram Set (A/B/C)
Audit Ledger (Stage 11 + 12)
Reliability Contract
Security & Privacy Contract
Operations Pack (SLO/alerts/playbook/cost)
Golden Demo Script (15 min, deterministic)

Definition of Done

One external dev completes integration using only the BetaKit docs
TTFS (time to first store) recorded — target < 30 min
TTFV (time to first value) recorded — target < 2 hours
Friction log collected — at least 5 points of confusion identified
Docs patched based on friction log before Phase 2 build starts
Rule: Phase 2 build does not start until the Definition of Done is met. Patch docs, not code — unless code is truly broken.

3-Dev Beta — What You Track

MetricFull nameTargetHow to measure
TTFSTime to First Store< 30 minMetricEvent entity — first mwStoreMemory call per tenant_id
TTFVTime to First Value< 2 hoursManually recorded — when dev reports first useful result from their integration
Points of confusionFriction log entriesCollect allFriction form / direct message — log to FrictionLog (or notes doc)
Feature vs doc requestsCategorised feedbackSeparate clearlyTag: [DOC_FIX] vs [FEATURE_REQUEST] vs [BUG]
Error rateAPI errors per tenant< 5%BetaMetricsDashboard → error rate column per tenant

Support Channel Rules

1.Each dev gets a direct contact channel (email or DM) — not a public forum
2.Response time commitment: within 24 hours
3.Bug reports: reproduce locally first, then share the exact curl/payload that failed
4.Feature requests: welcome but flagged — Phase 2 build waits until first integration succeeds
5.Doc feedback: highest priority — patch docs same day if possible
6.If integration is blocked for > 2 hours: escalate to priority support

What Drives Phase 2 Build Direction

Only after the first external integration succeeds. The data tells you what to build — not gut feeling.

INPUT SIGNAL

Who adopts MemoryWell first

BUILD DECISION

Build deeper tooling for their vertical (chatbot, knowledge base, agent memory)

INPUT SIGNAL

What vertical screams loudest

BUILD DECISION

Prioritise the use case with fastest time-to-value and clearest pain point

INPUT SIGNAL

Fastest conversion path

BUILD DECISION

Pick the layer (Vault / Bottle / Verse) that needs least new infrastructure

The rule: Let the market tell you what to build. The product layers (Vault / Bottle / Verse) are ready to build — but only the right one, for the right vertical, at the right time.