Compare
Agnes vs Glean, side by side
What each product does today, taken from its own documentation. Previews are marked as previews, and where the answer depends on your setup, we say so.
GleanScorecard
14 questions, one verdict each
Every Agnes answer links to the changelog entry where it shipped.
- Warehouse sources: Even
- Data foundation: Agnes
- Access per person: Depends
- Semantic layer: Agnes
- Owning definitions: Agnes
- Company memory: Agnes
- Where your definitions live: Agnes
- Document indexing: Glean
- AI tools people already use: Even
- Model choice and your own key: Even
- Agents that do the work: Even
- Sharing results: Agnes
- Where it runs: Agnes
- Security operations: Glean
- Available
- Available, admin-controlled
- Preview or limited
- Not documented or not built
Showing 14 of 14 questions.
Data and sources
Warehouse sources
Can it answer from our warehouse?
EvenBoth query warehouses live, without copying them. Glean signs in as each person on all three; Agnes adds Keboola and scheduled copies. The bigger difference is who owns the definitions.
GleanAvailable: Live Snowflake, BigQuery and Databricks queries through each warehouse's own tools. By default each person signs in with their own warehouse account.
AgnesAvailable: Live queries, zero copy: each question runs inside Snowflake, BigQuery or Databricks and only the answer comes back, so numbers are always current and nothing is duplicated. Keboola tables and scheduled copies when you want them. BigQuery can also run under each person's own Google account.
Data foundation
What if half our data still sits in an ERP like SAP?
Glean starts from tables that already exist. Agnes can build them.
GleanNot documented or not built: No data preparation product documented. SAP is not in Glean's connector catalogue; NetSuite, Dynamics 365 and Workday are.
AgnesAgnes aheadAvailable: Built on Keboola's data foundation: 1,500+ connectors, ELT and ETL, transformations, scheduled automation, and Kai, the AI agent that builds the pipelines with your team.
Access per person
Does each person see only what they are allowed to?
DependsGlean for permissions across many document apps. Agnes for row and column rules on data.
GleanAvailable: Permissions mirrored from each connected app and checked for the signed-in person in search, answers, agents and its MCP server.
AgnesAvailable: Group grants on every data package, table, plugin and agent, plus sharing for reports and apps. Row filtering and column masking, set per group by your admins, on live tables and on copies held on the server, and BigQuery under each person's own account.
Definitions and semantics
Semantic layer
Do our agreed definitions reach every tool?
GleanNot documented or not built: No metric layer of its own documented. For governed numbers it uses Snowflake semantic views and Databricks Genie spaces.
AgnesAgnes aheadAvailable: Several semantic models in one layer, with a business glossary, metric definitions and a Knowledge Graph. Reads Snowflake semantic views, Databricks metric views, Keboola, git and uploads.
Owning definitions
Who approves what a business term means?
GleanPreview or limited: For documents: verified badges and short approved answers. No approval step for metric definitions documented.
AgnesAgnes aheadAvailable: Each definition has one source that owns it: git, Keboola or your warehouse. Every correction waits for an admin's approval, and metrics can carry approvals and owners.
Company memory
Does it get smarter for everyone, not just one person?
Shared and reviewed in Agnes, per person in Glean.
GleanAvailable: Memory per person: what you tell it and what it learns from your own chats and work.
AgnesAgnes aheadAvailable, admin-controlled: Shared memory from conversations with a review queue, plus personal memory with your approval.
Where your definitions live
What can we take with us later?
GleanPreview or limited: The index lives in the Glean deployment. Exports exist for documents, audit logs and agent traces; no bulk index export documented.
AgnesAgnes aheadAvailable: Definitions stay in your git, Keboola or warehouse. Models in Agnes export as open Ossie YAML.
Documents
Document indexing
Can answers use our contracts, decks and notes?
Glean covers far more apps.
GleanGlean aheadAvailable: 107 connectors, each with permission-aware indexing, live tools or both. Sensitive-content scanning is part of Protect+, licensed site-wide for an annual fee.
AgnesAvailable, admin-controlled: SharePoint and uploaded files, indexed where they live, with names pseudonymized before storage in the folders you choose. A Knowledge Graph keeps each fact's source quote.
AI access and models
AI tools people already use
Can people ask from Claude, ChatGPT or their editor?
EvenBoth cover the main assistants, coding tools, Slack and Teams. Glean also lists Zoom and more MCP hosts.
GleanAvailable: An MCP server with 22 documented hosts, including coding tools, ChatGPT, Claude, Copilot Studio and Gemini Enterprise, plus assistants inside Slack, Teams, Zoom and other apps.
AgnesAvailable: Claude, ChatGPT, Cursor and GitHub Copilot Chat in VS Code over MCP, with full setup for Claude Code and Codex, under each person's permissions. Answers in Slack and Microsoft Teams.
Model choice and your own key
Whose AI model runs, and on whose contract?
EvenBoth run on your own key and contract. Glean offers more model families for chat directly; Agnes chat runs on Claude or through your gateway.
GleanAvailable: OpenAI, Azure OpenAI, Gemini and Claude, on Glean's keys or your own, plus open models on Glean's key.
AgnesAvailable: Bring your own key: Claude on your Anthropic key or in your own Google Cloud through Vertex AI, or chat through your own AI gateway. Other AI work runs on OpenAI, Gemini or any OpenAI-compatible provider, routed task by task.
In the changelog
Agents and shared results
Agents that do the work
Can it run recurring jobs, not only answer?
EvenGleanAvailable: Describe the goal or build step by step. Runs on a schedule or when content changes; agents with their own identity are in beta.
AgnesAvailable: Describe the job and the builder sets up the agent. Agents run on a schedule or when another system starts them, colleagues can share them with admin approval, and spending has a budget.
Sharing results
Can people share what they built from an answer?
Agnes data apps query live data on a server and can show each viewer their own data.
GleanAvailable: Documents and interactive pages inside the company, refreshed on a schedule. Pages run in the viewer's browser only.
AgnesAgnes aheadAvailable: Reports at a stable link for the right people, pages that keep their numbers current, and data apps from a chat request.
Running it
Where it runs
Who operates it, and where?
If your IT wants to run it, or you need Azure. Some buyers prefer that the vendor runs it.
GleanAvailable: Single-tenant, hosted by Glean or in your own Google Cloud or AWS account, always operated by Glean.
AgnesAgnes aheadAvailable: A single-tenant instance for your company only: managed for you on Keboola's Google Cloud infrastructure, or on your own servers in your cloud, AWS and Azure included.
Security operations
What does our security team get?
Glean's security tooling is more mature.
GleanGlean aheadAvailable: An admin console and audit logs to CSV or your SIEM. For Customer Hosted, also a vulnerability report per release and Wiz or CrowdStrike sensors.
AgnesAvailable: A record of every agent run and admin action, sign-in through Google, Microsoft Entra, Okta or Keboola, and new access tokens that expire within a year.
Glean today
What Glean gives you today
Glean is a mature product, and a buyer who already likes it will check our claims. This is what Glean's own documentation describes today.
- Available
Search across 100+ apps
107 connectors, each with permission-aware indexing, live tools or both, checked for the signed-in person in search, answers and agents.
- Available
Answers from your warehouse
Live Snowflake, BigQuery and Databricks queries under each person's own login, through each warehouse's semantic views or Genie spaces.
- Available
Reach into other tools
An MCP server with 22 documented hosts, from coding tools to ChatGPT, Claude, Copilot Studio and Gemini Enterprise, and assistants inside Slack, Teams, Zoom and other apps.
- Available
Model choice
OpenAI, Azure OpenAI, Gemini and Claude, on Glean's keys or your own, plus open models on Glean's key.
- Beta
Agents with their own identity
Agents that run on events or schedules under a scoped service identity, with a presence in Slack. Regular agents run as the person who started them.
- Paid add-on
Protect+
Scans indexed content for sensitive data and hides flagged items. Content fetched live is not scanned.
Where Agnes adds value
Where Agnes adds value for a Glean shortlist
What Agnes brings when Glean is on your shortlist. Every card is in the released product; some are admin-controlled, so you decide who gets them and when.
- Available
Live queries, zero copy
Agnes asks your warehouse directly: each question runs inside Snowflake, BigQuery or Databricks and only the answer comes back. Numbers are always current, there is no pipeline to keep in sync, and your warehouse's governance stays in force.
- Available
Definitions you own
Semantic models, a glossary and metric definitions, each owned by one source: git, Keboola or your warehouse, plus a Knowledge Graph. Corrections wait for an admin's approval.
- Available
A data foundation, not only a search layer
Data still in an ERP gets a real pipeline through Keboola: 1,500+ connectors, transformations, automation, and Kai building it with your team.
- Available
Run it your way
A single-tenant instance managed for you on Google Cloud, or on your own servers in your cloud, AWS and Azure included, operated by your IT.
- Available, admin-controlled
Company memory people review
Agnes learns shared knowledge from conversations into a review queue, so the whole company gets smarter, not one person.
- Available
Results that keep working
Reports at a stable link, pages that keep their numbers current, and data apps from a chat request that query live data on a server.
Every item below is in the released product. Some are admin-controlled: your admins decide who gets them and when, which is part of governing AI across a company. If your job is search across many document apps and you want the vendor to run it, Glean alone may be the better answer.
Which fits you
Glean, Agnes, or both
The right answer depends on the job you are hiring for. These are the three answers we give most often.
Agnes, when
the numbers have to match what finance agreed, half your data is not in a warehouse yet, or your IT wants to run the instance, including on Azure.
Both, when
Glean already serves search across your apps and you also need governed numbers and agents on your data. People reach either from the assistant they use.
Glean alone, when
the job is search across many document apps, your numbers already live in governed warehouse views, and you prefer that the vendor runs it.
The practical next step is to bring a set of your own questions and run them both ways.
Bring your questionsWhere your data goes
You choose the AI provider
Every model call goes to the provider your deployment is configured to use. Pick one below to see who reads a question and whose contract it runs on.
A person asks
In Agnes, or in Claude, ChatGPT, Cursor or VS Code, under that person's permissions.
Your Agnes instance
Single-tenant, yours alone. It queries your warehouse live, with zero copy: only the answer comes back. Indexed documents sit here.
The model you chose
The question, with the data and passages found to answer it, goes to the provider you configured.
Anthropic APIThe answer comes back
With its sources, so people can check where each number came from.
- Who reads the question
- Anthropic reads each question to answer it.
- Whose contract
- Chat runs under your own Anthropic key, so the model relationship is yours.
We don't claim your data never leaves your cloud: the AI provider has to read a question to answer it.
If people use Agnes from Claude Code, Claude Code talks to its own model provider as usual, and the tables a person is allowed to see can be copied to their laptop for fast local analysis.
8 of 8 questions shown.
Yes. Glean queries Snowflake, BigQuery and Databricks live, using each warehouse's own tools and each person's own login. The difference is who owns the definitions: Glean relies on what your data team builds in the warehouse, while Agnes keeps the definitions itself and checks questions against them.
It doesn't have to. Agnes queries your warehouse live, with zero copy: each question runs inside Snowflake, BigQuery or Databricks and only the answer comes back, so numbers are always current. If you want one, a scheduled copy is kept on your single-tenant Agnes instance and refreshed on a schedule, which makes repeated questions faster and spares warehouse cost. Keboola tables are usually kept as copies, refreshed when the data changes.
Agnes is built on Keboola's data foundation, so data that is not in a warehouse yet gets a real pipeline: connectors, transformations and scheduled automation. Glean documents no data preparation product of its own, and SAP is not in its connector catalogue, although partners or custom connectors may cover it.
Yes. Your IT can run Agnes on a virtual machine in your own cloud account, including AWS or Azure, and then owns operations, backups and upgrades. Or we run a single-tenant instance for you. We'll walk you through how running it yourself is packaged and supported on a call.
Glean's Customer Hosted option runs in your own Google Cloud or AWS account, single-tenant, and Glean operates it. Glean says customers do not deploy or patch it themselves. If you would rather the vendor runs it, that is a fair reason to prefer Glean.
Yes. Glean can stay your search across company apps, and Agnes adds governed numbers, owned definitions and agents on top of your data. Both connect to Claude and ChatGPT over MCP, so people can reach either from the assistant they already use.
Both let you bring your own key, so the model contract is yours. Agnes chat runs on Claude, on your Anthropic key or in your own Google Cloud through Vertex AI, or through your own AI gateway and the models it serves, and other AI work runs on OpenAI, Gemini or any OpenAI-compatible provider, routed task by task. Glean offers OpenAI, Azure OpenAI, Gemini and Claude for chat, on its keys or yours.
Glean facts on this page come from Glean's own documentation, read on 12 November 2026; we did not test Glean hands-on. Agnes facts come from the Agnes product documentation and the public changelog. Products change; if something here is out of date, tell us on the demo.
- Glean connectors (opens in a new tab) Last updated 6 October 2026. The connector catalogue.
- Connect Snowflake to Glean Assistant (opens in a new tab) Last updated 12 October 2026. Live warehouse queries, with BigQuery and Databricks pages alongside.
- Glean security principles (opens in a new tab) Last updated 11 November 2026. Permission checks across search, answers and agents.
- About Glean deployment (opens in a new tab) Last updated 29 September 2026. Glean Hosted and Customer Hosted, and who operates them.
- Glean memory (opens in a new tab) Last updated 12 October 2026. Saved and learned memories, per person.
- Glean model hub (opens in a new tab) Last updated 10 November 2026. Models and bring-your-own key.
- Glean MCP server (opens in a new tab) Last updated 3 November 2026. Tools and supported hosts.
For how Agnes runs and where model calls go, see the deployment page.
AI Benchmark
See how Agnes did on real business questions
Cuesta Partners asked Agnes, ChatGPT and Claude the same 10 questions, three times each, and graded every answer themselves. One firm, one set of questions: directional, not a general ranking.
Share of graded answers that passed- Agnes100%
ChatGPT + SharePoint78%
Claude + SharePoint70%
Same 10 questions, 3 runs each, graded by Cuesta.