Catalogs, semantic layers, knowledge graphs · you've built them where you had control: the warehouse. Operational systems stayed out of reach. Own your context is the same discipline at every source · LLM-maintained knowledge of schema, semantics, rules, and working examples · the substrate LLM analytics needs to be trusted.
The parts of the modern data stack that never reached operational systems. No central re-model required · the wiki compiles where the data lives and stays current as it changes.
Schema, rules, and definitions resolved here
Salesforcesales_read
Snowflakeanalytics_ro
Postgresapp_read
Jiraissues_readOperational sources · CRM, tickets, app databases · connected alongside the warehouse, which stays authoritative for analytics. One connection among every system that owns truth.
Secure connections# The agent reads the wiki before it queries.▸ wiki Opportunity · schema + meaning▸ wiki rules · what "qualified" means here▸ wiki examples · joins that ran clean$ mp.run("slipped_deals.sql")# Starts from proof · not from guesses.▍
Joins, filters and quality checks run in code before anything reaches the model, so analytics agents reason over refined results rather than raw rows.
Governed workspaceCompiled by an agent · corrections kept as revisions
Schema, semantics and rules are captured per source and linked into one growing body of knowledge · versioned, portable files, with no lock-in on your semantics.
Trust and governanceNo central re-model required. The wiki compiles where the data lives and stays current as it changes.
Point the gateway at operational sources · CRM, tickets, app DBs · alongside the warehouse.
An agent builds and maintains the wiki: schema, semantics, rules, queries that worked.
Analytics agents read the wiki first · consistent answers, whoever asks.
Two people ask about the same account from different tools. Both agents read the same captured context first, so the definitions behind both answers agree.
The discipline you built for the warehouse · catalog, semantic layer, lineage · extends to the operational systems that were always out of reach.
A new field lands upstream. The wiki updates diff by diff and health checks flag synthesis that no longer matches, before it shows up as a wrong number.
A catalog is written by people and read by people, and it bit-rots between review cycles. This is compiled and maintained by an agent, and its first reader is another agent · so it captures the automations, business rules and working joins catalogs miss.
No. The wiki compiles where the data lives, per source, and the pages link to each other. The warehouse stays authoritative for analytics · this is the same discipline applied to the systems around it.
No. The wiki lives inside your workspace and your deployment, as versioned, portable, exportable files · not rows in a proprietary metadata store.
Same data gateway, same governance model. What changes is who is driving and where the work is written.
A knowledge map per connection: schema, business rules, and joins that already worked.
Read the Adaptive Context pageLiveConnect100+ live enterprise systems behind the same six verbs, scoped per user and agent.
Read the LiveConnect pageAgent SDKGoverned connections, managed execution, and audit inside the agent you ship.
Read the Agent SDK pageOne MCPOne governed tool surface that Claude, ChatGPT, Cursor, and Copilot can use.
Read the One MCP page
Connect one operational source and see the wiki it produces · then decide how far to take it.
