As we eagerly await Oktane 25, we at Fctr Identity are thrilled to share that Tako AI has been hard at work preparing some game-changing features for the identity management community. While everyone’s buzzing about what Okta will unveil at their annual event, we couldn’t wait to share our latest innovations that are reshaping how organizations interact with their identity infrastructure.
Tako AI v1.3 “Takotastic” represents our most significant leap forward in delivering consistent, deterministic results through advanced multi-agent orchestration. Here’s what’s new and why it matters for your identity operations.
🔐 OAuth 2.0 Support: The Token Method Okta Recommends
Finally here – and it was worth the wait.
After countless user requests and following Okta’s best practice recommendations, Tako AI now supports OAuth 2.0 authentication with Private Key JWT. This isn’t just another authentication method; it’s the enterprise-grade security foundation that modern identity platforms demand.
Why This Matters:
- Okta’s Recommended Approach: Aligns with Okta’s security best practices for programmatic access
- Enhanced Security: Private Key JWT authentication provides superior security compared to API tokens
- Scalable Architecture: Supports large-scale deployments with proper credential management
- Future-Proof: Positions your Tako AI deployment for upcoming Okta security enhancements
This authentication upgrade ensures Tako AI integrates seamlessly with enterprise security policies while maintaining the same powerful natural language querying capabilities you rely on.
🧠 Execution Details UI: Watch Tako AI’s Mind at Work
From black box to crystal clear – see every step of Tako AI’s reasoning process.
The new “Thinking UI” transforms Tako AI from a mysterious black box into a transparent, observable system. Our expandable execution panels provide unprecedented visibility into how Tako AI plans, executes, and assembles your identity queries.
Revolutionary Transparency Features:
Unlike traditional loading spinners that leave users wondering “what’s happening?”, Tako AI’s Thinking UI builds confidence through visibility. When you can see the Planning Agent analyzing your request, the API Agent making calls, and the Results Agent formatting responses, you understand not just what Tako AI is doing, but why.
- Real-Time Progress Tracking: Watch each agent as it processes your request
- Step-by-Step Breakdown: See exactly how complex queries are decomposed and executed
- Timing Analysis: Understand performance characteristics of different operation types
- Error Visibility: Clear error reporting with context for troubleshooting
- Payload Inspection: Progressive disclosure of data flows between agents
This transparency is crucial for enterprise deployments where understanding system behavior directly impacts trust and adoption rates.
🔍 Access Analysis Special Tool: “Can User X Access Application Y?”
The most common question Okta admins and help desk teams face – now answered instantly.
Every day, identity administrators and help desk staff field the same question: “Can this user access that application?” Tako AI’s new Access Analysis Tool provides comprehensive answers in seconds, not minutes of manual checking across multiple Okta screens.
How It Works:
Simply ask: “Can john.smith@company.com access Salesforce?”
Tako AI’s special tool performs a complete access analysis by collecting user status, application details, assignment discovery, policy evaluation, and network context. The tool collects comprehensive raw data and lets Tako AI’s LLM analyze the complete access picture, including complex policy rule priorities and conditional access scenarios.
Perfect for help desk teams providing instant answers to replace manual multi-screen checks, security audits validating access patterns, troubleshooting access issues, and compliance documentation with full reasoning.
🎯 Relationship Analysis Agent: Reliability and Repeatability
The breakthrough that transforms probabilistic AI into deterministic identity operations.
While Large Language Models are inherently probabilistic, Tako AI introduces the Relationship Analysis Agent – achieving ~95% consistent multi-step result assembly through advanced entity correlation.
Real-World Example:
Query: “Find users who logged in the last 30 days and get all their assigned apps, groups and roles”
This requires 4 execution steps generating separate datasets:
- API Query: User details with groups and applications (5 records)
- API Query: Users with role assignments (2 records)
- API Query: Detailed role assignments per user (2 records)
- API Query: User profile enrichment (2 records)
The Solution:
The Relationship Analysis Agent samples data from each step and creates correlation maps:
- Cross-Step Joins: Links
user_idacross all datasets - Field Mappings: Normalizes
okta_id↔user_idinconsistencies - Aggregation Rules: Groups role assignments by user
- Processing Template: Defines “ENTITY_AGGREGATION” blueprint
This eliminates guesswork and ensures consistent, accurate results every time—transforming probabilistic AI into reliable identity operations.
🛡️ Additional Enhancements
Beyond the major features, Tako AI includes crucial improvements for enterprise environments:
Cross-Platform Reliability:
- ARM64 Support: Full compatibility with Apple Silicon and AWS Graviton processors
- Enhanced Docker Support: Multi-architecture containers for seamless deployment
AI Provider Ecosystem:
- Multi-Provider Testing: Validated across OpenAI, Anthropic, Google Vertex AI, and AWS Bedrock
- AWS Session Token Support: Enhanced security for temporary credentials and STS assume role scenarios
Developer Experience:
- Okta EventTypes Tool: Sophisticated system log filtering for precise audit queries
- Enhanced Error Messages: Clear, actionable feedback for troubleshooting
- Comprehensive Documentation: Updated guides for all supported AI providers
Looking Forward: Oktane 25 and Beyond
As we prepare for Oktane 25, Tako AI positions the identity management community at the forefront of AI-powered operations. While we’re excited to see what Okta announces at their annual event, we’re confident that Tako AI’s multi-agent architecture provides a solid foundation for integrating with whatever innovations emerge.
What’s Next:
- Supervised Write Operations: Human-in-the-loop provisioning workflows (coming Q4 2025)
- Advanced Policy Analysis: AI-powered compliance and risk assessment
- Predictive Identity Modeling: Proactive access management based on usage patterns
Get Started with Tako AI
Ready to experience the future of identity management? Tako AI is available now with our signature 10-minute Docker deployment.
Resources:
- GitHub Repository: fctr-id/okta-ai-agent
Connect with the Team:
- General Support: support@fctr.io
- Direct Development: dan@fctr.io
- Feature Requests: GitHub Issues
The Future of AI-Powered Identity is Observable, Deterministic, and Secure
Tako AI “Takotastic” release represents more than just feature updates – it’s a fundamental shift toward transparent, reliable, and secure AI-powered identity operations. As the identity management community gathers for Oktane 25, Tako AI stands ready to amplify whatever innovations emerge from Okta’s latest announcements.
The future of identity management isn’t about replacing human expertise – it’s about augmenting it with AI systems that think transparently, act consistently, and secure reliably.
© 2025 Fctr. Built with ❤️ for the identity management community.
This looks like a phenomenal release and I am very excited by the progress. The “Can User X Access Application Y?” example is such a compelling use case. Especially if this app can provided as a front-end to regular users to demystify access.
Can I also ask “Can user X *request* application Y?”
Not yet on the request part.
But a great idea … let me look into it.
Thanks! I am very excited about the potential for a custom/stripped down Tako interface for general users. One of the biggest obstacles we face to eager adoption of self-service access requests is that even those champion users stumble on, “I’m going to go tell them to request access. Wait, are they able to request that?”
Some of it is an access request design challenge, other times it would be nice for users to be able to ask and get an answer!