Query reuse
Optionally rerun or adapt successful retrieval scripts against configured sources, rather than generate each query from scratch.
Tako AI for Okta
v3.2 Beta releaseYour Okta questions.
Clear answers.
Ask about users, groups, applications, policies, and devices in plain English. Review the results, ask a follow-up, and export what you need. A self-hosted assistant built for Okta identity teams.
- Web
- CLI
Slack
Microsoft Teams
Tako AIExample conversation
You ask
Find the SAML certificate expiry date for all active SAML applications.
AI Summary
Four active SAML applications.
Two expire within 90 days.
| Application | Certificate expires |
|---|---|
| Atlas CRM | 2026-10-18 |
| Beacon Analytics | 2026-11-30 |
Less time finding data.
More time using it.
Move from a question to a useful result, without piecing together API calls and spreadsheets.
Okta API coverage
Okta-specific context and guidance across 107 read-only API endpoints help turn a question into a retrieval plan.
Conversation history
Follow up on an answer, narrow the question, and return to saved conversations in the web interface.
Retrieval progress
Retrieval progress makes the work visible. AI Summary helps explain the results once they arrive.
Interactive results
Tako handles large results without sending every record to the model. Choose columns, sort, group, and export CSV or generated Python for further analysis.
Web, CLI, and chat
Use the web, automate queries and syncs with the CLI, or add optional Slack and Teams conversations with follow-ups and CSV downloads. Configured restrictions control bot access.
Live Okta data.
Synced data.
Or both.
Tako selects a source to suit the question. You can also ask it to use the live Okta API when you need a current lookup.
- Query live
- Retrieve data directly from Okta without an initial sync. Access follows the permissions of your configured Okta credentials.
- Use a local copy
- Sync users, groups, apps, policies, devices, authenticators, and assignments to SQLite on your infrastructure for SQL queries.
- Combine sources
- Use synced and live data together when the question calls for it. Synced results reflect the last sync.
One question.
A better starting point.
Ask.
Describe the Okta question in your own words.
Retrieve.
Tako plans and checks retrieval from your configured sources.
Refine.
Ask a follow-up, review the answer, and export the results.
Tako AIExample conversation
List active users in Engineering with no MFA enrolled.
Review the matching users.Inspect profiles and enrollment. Choose columns. Export a CSV.
Show the roles for those users.
More questions to explore
- Which groups does this user belong to?
- Which applications are assigned to those groups?
- Using the live Okta API, list five active users.
Results come from your configured Okta permissions and data sources.
Your environment.
Your deployment choices.
Run Tako in Docker on AMD64 or ARM64. Synced data, saved conversations, and logs live on your infrastructure. Configure the AI provider and channels your team needs.
Choose your AI provider
OpenAI, Anthropic, Google AI Studio or Vertex AI, Azure OpenAI, AWS Bedrock, and OpenAI-compatible endpoints, including local models through Ollama.
External AI providers can receive query and identity context. A compatible local model keeps inference on your infrastructure; Tako still needs connectivity to Okta. Optional Slack and Teams workflows also process data through those services.
Compare provider options107+ GET API endpoints
Query users, groups, applications, policies, devices, and authenticators through Okta’s read-only API endpoints. Connect with OAuth 2.0 or an API token; your configured roles and scopes control access.
AST-based code validation
Generated Python is parsed and checked before execution. Import allowlists, restricted calls, and blocked patterns constrain retrieval code. Result analysis uses allowed built-ins and methods, with runtime limits on retrieval scripts.
CSV export
Explore results with column selection, multi-column sorting, and grouping. Export loaded rows and columns as CSV, or fetch all records to export the complete saved result.
Put your first question to work.
Start with the repository, then configure Tako for your Okta environment.
- Docker with Docker Compose, or the documented local installation.
- An Okta Identity Engine tenant with OAuth 2.0 or API token authentication and the read permissions your questions need.
- Access to a supported cloud or local AI provider.