code audit
detect what shipped
brand and reach(soon)
lift quality
perfect memory(soon)
decision graph
// perfect memory (coming soon)

structured decisions. not flat recall.

this is not a generic memory product. perfect memory is a decision graph — it captures what your AI agent decided, the alternatives it considered, why it chose what it chose, and how each decision got superseded over time.

generic memory tools (Mem0 et al.) store facts and recall them. that's useful for chatbots; it's not what a guardian does. a guardian remembers the provenance of every choice so the team (or the next agent session) can revisit a decision with full context, not just the conclusion.

four primitives.

small surface area on purpose. each primitive captures or queries one aspect of the decision graph.

record_decision
record_decision(context, alternatives, rationale, chosen)

Capture a decision as it's made — what you were deciding, the alternatives you considered, the rationale, what you picked. Not the implementation, the decision itself. The signal you need a month later when someone asks "why did we do it this way?"

query_decisions
query_decisions(query, scope)

Semantic search over decision rationales. "Why did we pick Stripe over Lemon Squeezy?" returns the original record with full context. Scope to a project, an epic, or across all projects (pro tier).

mark_superseded
mark_superseded(decision_id, superseded_by, reason)

Decisions don't die, they get overruled by later ones. Mark the supersession explicitly with a pointer + reason. The audit trail shows the lineage, not just the latest state.

get_decision_lineage
get_decision_lineage(decision_id)

Walk the supersession graph backwards — "what was the first decision that led here, and what changed each time?" Useful when revisiting an architecture choice that's been re-decided three times.

three tiers planned.

free covers most indie use; pro adds the cross-project + semantic layer; regulated exists for teams that need a tamper-evident audit trail.

free
£0
per-project decision log
  • +record_decision / query_decisions / mark_superseded scoped to one project
  • +plain-text storage in Supabase
  • +AI editor surfaces decisions inline as you work
pro
TBD
cross-project lineage + semantic search
  • +all four primitives across every project you own
  • +pgvector semantic search over decision rationales
  • +cross-project lineage queries — see how a pattern travelled
  • +export the decision graph as a markdown audit trail
regulated
enquire
hash-chained + qualified timestamps
  • +cryptographic hash-chain over the decision log (tamper-evident)
  • +qualified timestamps via a trust service provider (eIDAS / equivalent)
  • +signed agent identity per record (which AI made which decision)
  • +compliance evidence for ISO 27001 / SOC2 / regulated industries
perfect memory ships Q3/Q4 2026.
Conceptually further along than implementation right now — primitives + tier design locked. Build timing pulls forward if a paying customer asks. Register interest to hear when it lands.
register interest