The product

Turn scattered work into context the team can use.

DM helps human + AI teams carry forward what matters: evidence becomes a candidate, people review what becomes trusted, and the relevant context is ready for the next task.

Create a DM account How it works

What you are seeing: real DM UI with synthetic demo data. Public walkthroughs label simulated backend responses; no customer data is shown.

DM Prepare Context screen for the task Provision the Aurora-backed Ledger storage path, showing relevant knowledge items, missing context, conflicts, and why the selection applies. — The task starts herePrepare Context brings the relevant pieces together.A real DM screen from a seeded demo workspace — not a customer workspace.

One recurring task

From evidence to reusable context.

The product loop is easy to follow: capture what happened, review what DM understood, approve what is still true, then bring it into the next piece of work.

01
Bring in the evidence
A source, note, or agent report is evidence first. It does not silently become organizational truth.
See the evidence intake lesson
02
Review the candidate
DM shows what it understood, the supporting evidence, and the checks to consider before approval.
See the review lesson
03
Approve what the team can rely on
A person can approve, edit, reject, or request evidence. Agent-generated learning remains a proposal until then.
Watch the decision walkthrough
04
Prepare the next task
The next person or connected agent gets the current, relevant, permitted subset — not the entire knowledge base.
Learn about Prepare Context

DM review queue showing a candidate decision extracted from an agent proposal, with Reject, Request evidence and Approve as Decision actions. — 02 · Human reviewA candidate is understandable before it is trusted.Review UI with synthetic data; approval remains a human decision.

An approved, active DM decision for the Helix ledger service, showing what was decided, its rationale, and tabs for authority, evidence and lineage, follow-through and history. — 03 · Reusable recordApproved context keeps its rationale and history.A real DM UI capture from the same seeded demo workspace.

Watch the workflow

See one decision move from evidence to reuse.

This approved public walkthrough focuses on the human review path. It includes native controls, captions, chapters, and a full descriptive transcript, so the steps do not depend on playback.

Founder-approved public demo · DM-02

From evidence to a trusted decision

Show one end-to-end decision lifecycle with evidence and human approval. A meeting note is not an approved decision. Here is the difference.

1 min 41 s

Chapters

Descriptive transcript

Read the full transcript, including what happens on screen

Demo environment · synthetic workspace and data · backend responses are simulated. A meeting note is not an approved decision
The extraction result, the checks, the approval, the approved decision and the Ask DM answer are SIMULATED: this guide uses a synthetic workspace and fixture data, not a deployed backend or a real AI run. It shows the sequence DM is designed around: a source becomes a proposal, a person approves it, and only then can it be recalled as approved knowledge. It does not claim every decision in a real workspace is captured or available.

A banner reading "Demo environment · synthetic workspace and data · backend responses are simulated" stays on screen for the whole video.

00:00–00:10 · launch

On screen: Start on "Add to DM" (/add). Click "Source-based intake". The next view is "Source-based intake" (/add/source). A highlight labelled "Source-based intake" marks "Source-based intake You have source material. DM understands it and p…".

Narrator: A meeting note is not an approved decision. Here is the difference.

Narrator: Start in Add to DM, with a source you already have.

00:10–00:13 · material

On screen: Click "Add source material". The next view is "Add source material" (/intake/bundles/new). A highlight labelled "Add source material" marks "Add source material Paste text and/or attach files together. DM analy…".

Narrator: Choose Add source material.

00:13–00:19 · paste

On screen: Type "Sync notes: we agreed that any migration touching production schema needs two reviewers. A single-reviewer migration dropped an index in the spring. Priya owns the rule." into the "Paste text" field. A highlight marks "The source note".

Narrator: Paste the note. It holds the choice, and the reason behind it.

00:19–00:23 · analyze

On screen: Click "Analyze with DM". The next view is "Pasted notes" (/intake/bundles/bundle-two-reviewers). A highlight marks "Analyze with DM".

Narrator: Select Analyze with DM.

00:23–00:32 · candidate

On screen: Click "Review". The next view is "Production schema changes" (/candidates/cand-two-reviewers/review). A highlight labelled "Nothing is approved yet" marks "Source material produces candidates for review. Approved candidates b…".

Narrator: DM proposes one decision candidate. It is a proposal, so it still needs a person.

Narrator: Select Review.

00:32–00:40 · understand

On screen: Click "Continue". The next view is "Production schema changes" (/candidates/cand-two-reviewers/review). A highlight labelled "Statement and reason" marks "Any migration that touches production schema needs approval from two…".

Narrator: Check what DM understood, and where it came from.

Narrator: The statement and its reason match the note.

00:40–00:44 · place

On screen: Click "Next". The next view is "Production schema changes" (/candidates/cand-two-reviewers/review). A highlight labelled "Next" marks "Next →".

Narrator: Place is optional here, so continue.

00:44–00:50 · check

On screen: Click "Continue to Refine". The next view is "Production schema changes" (/candidates/cand-two-reviewers/review). A highlight labelled "Blocking issues" marks "0".

Narrator: Check lists any blocking issues. DM found none here.

00:50–00:55 · refine

On screen: Click "Continue". The next view is "Production schema changes" (/candidates/cand-two-reviewers/review). A highlight marks "Edit if needed".

Narrator: Refine is where you would correct the wording. It is right as written.

00:55–00:59 · confirm

On screen: Click "Approve". The next view is "Pasted notes" (/intake/bundles/bundle-two-reviewers). A highlight marks "Approve".

Narrator: A person approves. Select Approve.

00:59–01:05 · approved-bundle

On screen: Click "View approved knowledge". The next view is "Knowledge" (/knowledge). A highlight labelled "Approved knowledge" marks "View approved knowledge".

Narrator: DM records the approval as durable knowledge. Open it.

01:05–01:10 · library

On screen: Click "Select the decision". The next view is "Knowledge" (/knowledge). A highlight labelled "The new decision" marks "Any migration that touches production schema needs approval from two…".

Narrator: It now sits in the approved Knowledge library. Select it.

01:10–01:13 · open-decision

On screen: Click "Open decision". The next view is "Two reviewers for production schema changes" (/decisions/dec-review). A highlight labelled "Open the decision" marks "Open decision".

Narrator: Open the decision.

01:13–01:18 · approved

On screen: A highlight labelled "Approved decision" marks "What was decided".

Narrator: The decision has its reason and its source on record.

01:18–01:24 · open-ask

On screen: Click "Ask DM". The next view is "Ask DM" (/ask-dm). A highlight marks "Ask DM".

Narrator: Later, a new teammate wonders why. They open Ask DM.

01:24–01:28 · ask

On screen: Type "Why do production migrations need two reviewers?" into the "Ask a question about decision memory" field. A highlight marks "Your question".

Narrator: They type the question in their own words.

01:28–01:31 · send

On screen: Click "Send". The next view is "Ask DM" (/ask-dm). A highlight marks "Send".

Narrator: Select Send.

01:31–01:40 · answer

On screen: A highlight labelled "Answer from the approved decision" marks "Any migration that touches production schema needs approval from two…".

Narrator: The answer comes from the approved decision, and shows the source it rests on.

Narrator: The team's reasoning is there when it's needed.

Open the full lesson with quick steps

Then put it to work

Prepare context for the next task.

Continue with the published Prepare Context lesson to see how current, applicable context is assembled around a task.

Learn about Prepare Context

Advanced detailWhat DM supports today — and what is still in progress

The workflow above is the shortest path through the product. These details explain the current capability boundaries for technical evaluators and buyers.

DM capabilities, value, availability, and implementation
Capability	Value	Status	Implementation
Decisions	Keep current decisions and rationale available to the work.	Ready	DM
Rules	Give humans and connected agents the constraints that apply.	Ready	DM
Skills	Reuse proven ways of working.	Ready	DM
Repository intelligence	Understand code structure, dependencies, relationships, and architecture around the task.	Ready	Graphify integration
Architecture context	Keep architecture boundaries and important system context available around the task.	Ready	Graphify-backed repository intelligence
Memory	Carry useful context from previous work across sessions and tasks.	In progress	Mem0 integration
Prepare Context	Bring the relevant pieces together for the task at hand.	Ready	DM
Agent access	Use DM through supported MCP clients.	Ready	MCP
Search / Ask DM	Ask and search the workspace for approved context that applies.	Ready	DM

In the task

Current, approved, in-scope Decisions and Rules; relevant Skills; repository and architecture context where available; authorization context where it exists.

Outside the task

Superseded or expired decisions, irrelevant records, and context the actor is not allowed to use.

Connected agents and MCP

Supported MCP clients can read already-approved context and submit candidate updates for review. Connecting an agent is explicit. DM does not claim automatic universal integration or runtime enforcement.

dm.prepare_task_memory

Prepare approved context that applies to a task.

dm.submit_memory_update

Propose new learning for human review.

dm.search_decisions

Search approved Decisions in the active workspace.

dm.propose_decision

Submit a Decision candidate for human review.

Provider boundary: Graphify is the current repository-intelligence integration. Mem0 is the current memory integration path and remains in progress. Neither provider is the product.