Niches · Company knowledge & data catalogues
Find the table, dashboard or topic that holds the data
No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.
What counts as done
A short ranked list of assets with fully qualified name, platform, description, owner and certification or usage signal, with a count first when matches exceed 20.
Measured, for our own tests: in at most two calls and under 5,000 tokens of tool output. No docs we read state these figures, so they don't count against a listing here.
How agents ask for it: which table has the real customer data
, is there a dataset for churn by region
, where does this dashboard read from
Does the job, with a limit 8
It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.
| Listing | Why |
|---|---|
| Alation C Alation, Inc. | catalog_context_search_agent takes a natural-language question and builds the signature itself, but result size and fields aren't stated, and the Aggregated Context API behind it has a one-time allowance of 2,000 calls before final 429s. source |
| Atlan C Atlan Pte. Ltd. | search_assets filters on certification, tags, domains, terms and dates and can return a count only, but descriptions come back only when requested, get_assets needs GUIDs from a prior search, and 41 tools sit on one endpoint. source |
| Basedash MCP B BaseDash Inc. (Delaware, United States) | get_data_sources, list_dashboards and get_dashboard give an inventory and ask_question can name tables, but no tool returns a ranked asset list with qualified name, owner or certification, or a count first. source |
| Collibra B Collibra Inc. | search_asset_keyword filters by status, community, domain, type and creator, and discover_data_assets takes natural language, but the latter spends Collibra Units and needs dgc.ai-copilot, and no count-first summary or token bound is documented. source |
| DataHub C Acryl Data, Inc. (d/b/a DataHub) | search takes /q syntax, filters, usage sorting and num_results=0 for facets, but the default tools carry about 26,000 characters of definitions, num_results is silently capped at 50 (issue 215) and document search once returned empty lists silently (acc-27). source |
| Marmot B Marmot Data Ltd | discover_data returns a summary when over 20 match, pages with offset and limit, and takes an mrn, with 11 small tools, but asset lists show name, MRN and link without owner or certification, so the success fields need a second lookup. source |
| OpenMetadata C Collate, Inc. | search_metadata takes fields, aggregations and size, and semantic_search covers vague queries, but the tool definitions on main are about 73,000 characters, far past the 5,000-token bar before any result. source |
| Secoda E Secoda, Inc. | search_data_assets takes a query and page only, with no platform, owner or type filters and no count-first mode; the result fields come from Secoda's API and aren't shown in the repo. source |
Couldn't check 2
The docs were closed to us, or silent. These carry no mark either way.
| Listing | Why |
|---|---|
| ClicData MCP F ClicData SAS | Whether a tool lists or searches data sets, with owner or description fields, couldn't be read: the help article needs JavaScript and tools/list needs a signed-in account. |
| Precisely AI Studio F Precisely Incorporated | The page lists a Data Governance category, but which agents or skills it holds and whether any searches a data catalogue couldn't be read; the help centre renders only with JavaScript. |
Doesn't do this job 10
Listings in the category that do something else, with the reason.
| Listing | Why |
|---|---|
| Cohere North D Cohere Inc. | North is sold as agents and search over work apps (Slack, SharePoint, Outlook, Jira, Notion, GitHub); no data catalogue is described. source |
| Dust B Permutation Labs | Dust indexes documents and app content into Spaces; it has no data catalogue of tables with platform, owner or certification. source |
| Gemini Enterprise B Google LLC | An assistant over connected apps, not a data catalogue; no asset search with qualified name, owner or certification in the pages read. source |
| Glean B Glean Technologies, Inc. | Not a data catalogue. The data_analysis tool routes questions to a BI tool or warehouse; nothing returns ranked tables with qualified name, platform, owner and certification. source |
| Guru D Guru Technologies, Inc. | A card and document knowledge base; no tables, dashboards or data catalogue. source |
| Hexis Cloud D Bevelites GmbH | A git repository of pages, skills and tool manuals; no tables, dashboards or catalogue. source |
| Onyx B DanswerAI, Inc. | A document search index, not a data catalogue; the MCP tools are document search, web search and URL fetch, with no table, owner or certification fields. source |
| Overclock F Overclock Technologies Limited | Indexes documents and email into a knowledge graph; no tables, dashboards or catalogue assets. source |
| Procore AI Go F Procore Technologies, Inc. | Not a data catalogue; it answers from daily logs, drawings, RFIs, specifications and submittals in one Procore project. source |
| Lens (Thought Industries) F Thought Industries, Inc. | No tables, dashboards or catalogue; sources are websites, files, Zendesk, Academy, Notion, Drive, S3, YouTube, Confluence and GitHub. source |
What stops agents finishing
- assets have no descriptions so search returns lookalikes
- search tool returns an empty list on a parsing bug rather than an error
- tool definitions alone take 26,000 to 50,000 characters of context
Around it
Before: chat.user-question. After: databases.sql-query, company-knowledge.lineage-trace.
Where this comes from
Weight 15 of 100 in Company knowledge & data catalogues: 8 of 38 counted accounts (Marmot and OpenMetadata deployments, the 47-tables account, DataHub search and document issues, Atlan, Marmot API-key cost, internal data-lake assistant)
The same as data: /api/v1/use-cases/company-knowledge.find-data-asset.json, or the MCP tool get_use_case with id company-knowledge.find-data-asset. Other jobs in this category: /api/v1/niches?category=company-knowledge.
For companies
Do agents find, use and choose your tools?
An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.