DataUnmess Memory
Give every connected AI the same business definitions and data context.
DataUnmess Memory keeps company knowledge, KPI definitions, and data catalog context beyond a single conversation, with provenance and conflict review when new information disagrees.
Workflow
How the feature works
Seed company context
Start with the company website or a concise description of products and operations.
Define business language
Store glossary terms, KPI formulas, grains, units, and material exclusions.
Catalog the data
Record datasets, keys, important fields, ownership, and source meaning.
Reuse and review
Search or ask Memory from MCP, and resolve conflicts instead of silently overwriting approved knowledge.
Durable context
Store the definitions that should survive the conversation
Memory separates reusable company and data knowledge from temporary chat context. Connected agents can search it before answering a KPI question, building a dashboard, or preparing a pipeline.
Governance
Review conflicts instead of losing approved knowledge
Low-risk additions can be learned directly. When a proposed KPI formula, dataset grain, or key conflicts with approved memory, DataUnmess creates a review item rather than overwriting the existing definition.
A user can accept the proposal, keep the approved version, or reject the change.
Across agents
Reuse one memory from Codex, Claude Code, Cursor, and other MCP clients
Memory belongs to the DataUnmess workspace, not to one AI conversation. Any authorized MCP client can retrieve the same company, KPI, and catalog context.
Existing saved dashboards can also seed knowledge through artifact backfill, helping a new workspace capture definitions that already exist in operational reporting.
Example prompts
Ask for the outcome in business language
DataUnmess MCP start
Remember that Active Customer means an account with at least one paid invoice in the last 90 days.
What is our approved Net Revenue Retention definition, and which dataset should I use?
FAQ
DataUnmess Memory questions
Is DataUnmess Memory shared across AI clients?
Yes. Authorized MCP clients use the same workspace memory, so definitions are not trapped in one conversation or provider.
What happens when new information conflicts with an approved KPI?
DataUnmess creates a conflict review item instead of silently replacing the approved definition.
Can existing dashboards help populate Memory?
Yes. Artifact backfill can learn low-risk knowledge from a saved dashboard and create review items for conflicts.
Continue exploring