A clear, approved record with a stable ID. Changes create lineage instead of silently overwriting history.
Decision Memory
by AuzzurA
Stop re-deciding
what was already
decided.
Approved decision records — not meeting notes.
Decision Memory captures approved decisions with their full context — rationale, evidence, owners, follow-through, outcomes, and lineage — so decisions stay traceable across meetings, tools, stakeholders, and time.
Meetings, Slack, docs, tickets, and emails are inputs. The durable object is the approved decision record.
The gap
Work gets recorded. The decision doesn't.
Meeting notes and meeting summarizers tell you what was discussed. Jira tells you what to do. Neither keeps the decision itself — what was chosen, why, based on what evidence, who approved it, and what changed later.
Meetings, Slack, docs, and tickets are inputs. None of them captures approved decision context as a durable, governed object.
Step 1
Discussed
What was said — rarely what was chosen
Too long · rarely re-read
Meeting summaries
Recorded, but walls of text. Nobody revisits them when the decision actually matters.
Fragmented
Slack & chat
Decisions buried in threads and DMs — lost across channels and silos.
Step 2
Done
What to execute — not why it was chosen
Tasks only
Jira & tickets
Tracks work. The decision and rationale behind the ticket aren't attached.
Drifts over time
Docs & wiki
Written once, updated rarely. Knowledge fragments as decisions evolve.
Step 3 · the missing layer
Decided
Approved decision context — not captured anywhere today
Approved decision records
What was chosen, why, by whom, with which evidence — durable, searchable, and human-approved.
Knowledge fragmentation
Too long to re-read
Same decision reopened
Lost rationale
Decision Memory fills the gap
What It Is
Decision Memory is the context layer for important decisions.
Meetings, chats, tickets, and documents are evidence. Decision Memory turns them into approved decision records that preserve what was decided, why it was decided, what supported it, who owned it, what actions followed, and how it changed over time.
Why it was decided — the reasoning, tradeoffs considered, and alternatives that were rejected.
Linked meetings, docs, tickets, and chat threads — selectively attached, with access controls.
Who owned the decision, who approved it, and which stakeholders were part of the context.
What actions came from the decision and what happened after execution in practice.
How the decision evolved, what it conflicted with, and what superseded or reversed it over time.
How It Works
Three steps from noise to trusted memory.
AI does the heavy lifting. Humans stay in control at every gate. Every approved decision becomes durable, traceable, and queryable context.
1
01
Evidence Intake
Evidence comes in.
You choose what matters.
Import or connect selected meetings, documents, tickets, chat threads, and emails. Decision Memory never silently captures everything — you choose what becomes evidence before anything is processed.
Meeting transcript
Google / Notion doc
Jira / Linear ticket
Slack / Teams thread
Email
2
02
Human Gate · Review
AI proposes.
You review and decide.
AI surfaces candidate decisions from your selected evidence. Review the decision, source context, rationale, missing information, and linked evidence — then approve what is accurate, reject what is not, or request more evidence.
3
03
Decision Memory & Recall
Approved. Traceable.
Ready to recall.
The approved decision becomes durable decision context with rationale, evidence, owners, tasks, outcomes, and lineage. It becomes available in the Decision Repository, searchable through Ask DM, comparable against future decisions, and reusable as trusted context for people and AI tools.
Product
From intake to trusted recall.
Interactive preview with sample decision data — not a live backend demo. Browse intake, review, decision records, and Ask DM.
Evidence intake
Human review gate
Decision record
Ask DM
Context Graph
Every decision keeps its context map.
Decisions connect to the evidence that shaped them, the tasks they spawned, the outcomes they produced, and the decisions they touched — forming a queryable context map people and AI tools can recall later.
Evidence linked to decision
Tasks & follow-through
Conflict signal
Decision lineage
Conflicts & Risks
Detect re-decisions before they become hidden conflicts.
DM surfaces signals when a new decision may reopen, conflict with, or supersede a prior decision — before it becomes costly rework.
Conflicts & Re-decisions
2 open
Re-decision
Identity & customer auth
"Drop in-house Keycloak; migrate all customer auth to WorkOS"
Reopens d4 — "WorkOS for SSO/SAML; defer broader IdP migration" · approved May 4
Conflict signal
Data platform
"Migrate Ledger to DynamoDB for write-scale"
Touches same system as d7 — "Use Postgres 16 Aurora as primary OLTP store" · approved May 14
Resolved
Multi-tenant schema
"Adopt per-tenant schema isolation for ledger v2"
Superseded d3 — "Shared schema with tenant_id column" · kept in lineage
Risks & Open Questions
8 open
Risk
d7 · Postgres Aurora
EU data-residency compliance not verified for Aurora config
Raised during approval review · owner: Theo Lindqvist · open since May 14
Risk
d4 · WorkOS SSO
No fallback plan if WorkOS API has extended downtime
Flagged by security review · owner: Maya Okafor · open since May 7
Question
d9 · Multi-tenant
Will per-tenant schema approach scale past 10k tenants?
Requires load-test evidence · linked task: "Load test EU region" · pending
Question
d7 · Postgres Aurora
Is 99.99% SLA contractually guaranteed for Aurora us-east-1?
Needs vendor confirmation · owner: unassigned · open since May 14
Why Decision Memory
Not notes. Not a chatbot. Not a search tool.
Compared with notes, chatbots, search, and task tools, Decision Memory focuses on approved decision context — with evidence, ownership, follow-through, outcomes, conflicts, and lineage.
| Capability | Decision Memory | Meeting Notes | AI Chatbot | Enterprise Search | Project Mgmt |
|---|---|---|---|---|---|
| Preserves decision rationale | ✓ Structured record | ✗ Free-text only | ✗ No memory | ✗ Documents only | ✗ No context |
| Links evidence to decisions | ✓ Selective & access-controlled | ✗ | ✗ | ~ Docs only | ✗ |
| Human approval gate | ✓ Required before memory | ✗ | ✗ | ✗ | ✗ |
| Ownership & accountability | ✓ Tracked per decision | ~ Attendee list | ✗ | ✗ | ~ Assignee only |
| Conflict & re-decision detection | ✓ Signals surfaced | ✗ | ✗ | ✗ | ✗ |
| Decision lineage & history | ✓ Full lineage graph | ✗ | ✗ | ✗ | ~ Version only |
| Queryable & grounded answers | ✓ Grounded in approved decisions | ✗ | ~ Hallucination risk | ~ Keyword only | ✗ |
AI-Native
AI-native organizations need trusted decision context.
"Your AI tools are only as strong as the context they can trust."
Decision Memory helps people today and prepares your AI ecosystem for trusted decision recall tomorrow. AI copilots need more than documents — they need approved decision context with evidence and human sign-off.
Approved context, not raw documents
AI recall is grounded in human-approved records — not unfiltered data that may be outdated or incorrect.
Future-ready for agentic workflows
When AI agents need to act on organizational decisions, DM provides the trusted, policy-aware context layer they require.
Decisions, not just documents
Most AI tools ingest documents. DM gives AI the structured decision layer — with rationale, evidence, and human sign-off already attached.
Security & Governance
Human approval before durable memory.
Every design decision in DM was made with governance in mind. AI extracts. Humans approve. Only then does something become trusted recall.
Human-gated approval
Only human-approved decisions become durable memory. Evidence is selected and governed before it enters the record.
Selective evidence by default
Evidence is selected and governed. Not everything in a meeting or thread enters the decision record.
Workspace-aware access
Decision visibility follows workspace boundaries. Context is scoped, not organization-wide by default.
Policy-aware recall
Recall is policy-aware. Ask DM is grounded in approved decisions and permitted evidence — not unrestricted raw data.
Evidence states
Evidence can be restricted, redacted, metadata-only, or blocked — sensitive material protected at the source level.
Grounded answers only
Ask DM does not speculate. Responses cite which approved decision and which evidence supports the answer.
FAQ
Questions about decision memory.
Decision Memory is a decision-context workspace — not a meeting summarizer, chatbot, or enterprise search layer.
What is Decision Memory?
Decision Memory is a decision-context workspace by AuzzurA. It turns fragmented meetings, docs, tickets, chats, and emails into approved decision records with rationale, evidence, ownership, follow-through, outcomes, conflicts, and lineage — so organizations can recall why a decision was made, not just what was discussed.
Is Decision Memory a meeting summarizer?
No. Meeting notes and AI summarizers capture discussion. Decision Memory captures the approved decision itself — what was chosen, why, based on which evidence, who approved it, and what changed later. Meetings are inputs; the durable object is the approved decision record.
What becomes a decision record?
A decision record is created only after human review. AI may propose candidate decisions from selected evidence, but nothing enters durable organizational memory until a reviewer approves it. Approved records include rationale, linked evidence, owners, tasks, outcomes, and lineage.
Does AI automatically create official decisions?
No. AI extracts and proposes candidates. Humans review each candidate, can request more evidence, reject it, or approve it. Official decision memory requires explicit human approval at the gate.
What is the human approval gate?
Every candidate decision passes through review before it becomes durable memory. Reviewers see the proposed decision, linked evidence, and any re-decision warnings. Approve, reject, or request more evidence — only approved decisions enter the Decision Repository and Ask DM.
What evidence can be attached to a decision?
Evidence can include meetings, documents, tickets, chats, and emails that you explicitly select during intake. Evidence is linked to the approved record — selectively, with access controls — not dumped into an unstructured archive.
How is Decision Memory different from Jira, Notion, or meeting notes?
Jira tracks work. Notion stores documents. Meeting notes capture discussion. Decision Memory preserves approved decision context — rationale, evidence, ownership, follow-through, outcomes, conflicts, and lineage — as a first-class object designed for recall and governance.
What is Ask DM?
Ask DM is grounded decision recall. You query approved decisions in natural language and receive answers cited to specific decision records and permitted evidence. Ask DM does not speculate from raw documents or unrestricted data.
Can AI tools use Decision Memory later?
Yes — that is the direction. Decision Memory is built as a trusted decision-context layer that people and AI tools can reference once decisions are approved. AI-native organizations need durable, policy-aware decision context rather than ad-hoc chat history.
How does Decision Memory handle sensitive evidence?
Evidence can be restricted, redacted, metadata-only, or blocked at the source level. Recall through Ask DM is policy-aware and grounded in approved decisions and permitted evidence — not unrestricted raw data.
Explore
Go deeper when you're ready.
Screens, workflows, and scenarios live on dedicated pages as the private pilot grows.
Private Pilot
Request a private pilot.
Decision Memory is in private pilot with a small group of design partners. If your organization loses decision context across meetings, tools, or stakeholder groups — we'd like to compare notes.