Understand the user population.
“Show me recently created admins without MFA.”
“List suspended users and who manages them.”
“Find contractors who have not logged in for 30 days.”
“Now narrow that list to users with Slack assigned.”
Tako AI
v3.0 Beta
Autonomous AI Agent for Okta.
Now with smarter routing and stronger follow-up investigations.
Built for Real Investigations
Move beyond one-off lookups. Tako helps identity teams answer questions, troubleshoot access issues, and keep following the trail across multiple steps using simple natural language.
- Investigate identity issues without restating context
- Web, CLI, and Slack access in one product
- Resolve access and policy questions faster
- 100% Private—Data stays on your infrastructure
- Works with OpenAI, Claude, Gemini & Ollama
Setup & Investigation Walkthrough
Setup, architecture, and live investigation workflows.
Interface
Web-Based Interface.
Tako's web interface makes longer investigations easier to follow, with clearer progress, better tables, and cleaner formatted answers.
Natural language
Start with the question.
Ask for the identity context you need without learning another query language.
- “Show me active admins without MFA.”
- “Now narrow that list to Slack users.”
- Keep the investigation connected across follow-ups.
Investigation workspace
Review a clear answer.
Progress, tables, traceability, and export options stay organized for the operator.
- Smarter routing selects the next useful step.
- Synced and live context are combined as needed.
- Results arrive in readable, operator-friendly formats.
Use cases
Autonomous Operations.
Ask Tako anything in natural language, then keep drilling into the answer with follow-up questions.
Find identity risk quickly.
“Show admins missing phishing-resistant MFA.”
“Find users who still rely on SMS.”
Trace access and change.
“Can john.smith@company.com access Salesforce?”
“What changed for this user's access in the last week?”
Follow assignments.
“Compare Engineering and Contractors group membership.”
“Find groups with more than 50 members.”
“Which apps are assigned through group membership only?”
“Show active SAML apps with expiring certificates.”
Workflow
Investigation Engine.
Tako v3.0 uses smarter routing, follow-up context, and clear synthesis to support longer identity investigations without losing the thread.
Smart Routing
Tako chooses the best next step for each question instead of forcing every request through the same path.
Live + Synced Context
It uses synced tenant data for speed and brings in fresh checks only when the answer needs it.
Follow-Up Continuity
It carries forward the right context so the next question stays connected to the investigation already in progress.
Clear Final Answers
It brings everything together into a clear response with traceability, export options, and operator-friendly formatting.
Resilience
Self-Healing Engine.
Watch Tako recover from runtime issues and keep moving. Built-in guardrails help prevent small errors from breaking longer workflows.
107 → 5
Filtered 107 possible operations down to the exact 5 needed for the query.
Exception found
Tako wrote a test probe to fetch group members but missed a required argument.
Recovered
Analyzed stack trace, understood the TypeError, and rewrote the function on the fly.
8,352 chars
Generated final, robust production code ready for deployment.
Self-correction traceResolved
Semantic Mapping: 107 endpoints analyzed → 5 optimal paths identified. Exception Detected: TypeError halted probe. Initiating auto-recovery. [0.42s] Executing list_group_members() TypeError: missing 1 required positional argument: 'group_id' Self-Correction: Parameter 'group_id' mapped from user context. Refactoring API Call... [FIX] Injected group_id: 00g...x21 [TEST] Request 200 OK Synthesis Complete: Production-ready script signed.
Features
Architected for Reliability.
Tako v3.0 focuses on consistency, flexibility, and operator clarity across real-world Okta investigations.
Smarter Investigation Flow
Better routing keeps work focused, reduces unnecessary steps, and helps Tako stay useful as questions become more complex.
Multi-Arch Support
Native Docker images for both AMD64 and ARM64. Deploy on everything from high-compute AWS Graviton clusters to local Mac mini workstations.
Follow-Up Continuity
Earlier results carry forward so teams can refine a question instead of starting over each time.
Cost-Efficient Models
Use efficient models like Gemini 3 Flash and Claude 4.5 Haiku while keeping stronger reasoning where it matters most.
Hybrid Data Modes
Flexible retrieval lets Tako use synced data for speed, fresh checks for recency, or both when the question needs it.
Cleaner Operator Experience
Cleaner tables, clearer progress states, and more readable output make longer investigations easier to review and act on.
Security
Privacy-First Architecture.
Your data never leaves your infrastructure. AI providers receive only the minimum context needed to help answer the question.
Local Docker Deployment
Runs entirely on your infrastructure. All identity data remains within your security perimeter.
Minimal LLM Exposure
The reasoning engine only receives your query and minimal data samples—never your complete Okta dataset.
Isolated Code Sandbox
All AI-generated code executes within secure, isolated containers, ensuring that every operation is validated and contained within a hardened perimeter.
Onboarding
Get Started with Tako AI.
Deploy Tako AI Agent in minutes and start investigating Okta questions across web, CLI, and Slack.
Clone the repository.
Clone the repository and follow our quick setup guide to deploy in your environment.
- Smarter investigations
- Web, CLI, and Slack access
- Docker deployment
- Local-first data handling
Use the guides.
Comprehensive guides, API references, and examples to help you get the most out of Tako.
- Setup walkthroughs
- Security and auth guides
- Slack + CLI configuration
Open source community
Join the Open Source community.
Report bugs, request features, or get direct support from our engineering team via GitHub or Email.
Community-driven development · MIT-Licensed · Production-Ready