INTREXA / Products & R&D
AI products for real work.
Research for what comes next.
INTREXA builds focused AI SaaS products and investigates the security and systems questions behind them. AXIS is our first product. FORGE is in development.
- ALLOWread:catalogpurchasing-bot → catalog/items✓ receipt
- ALLOWfile:writereport-agent → s3://corp-data/q2.csv✓ receipt
- DENYdelete:backuppitch-demo → backups/nightly✓ receipt
- REVIEWapprove:paymentpurchasing-bot → payments/vendor-invoice✓ receipt
- ALLOWread:calendarpitch-demo → calendar/team-standup✓ receipt
— The platform
Every action, decided and recorded.
Before an agent touches a tool, AXIS verifies who delegated what, scores trust and risk, evaluates policy, and returns allow, deny or human review — with a signed receipt you can replay later.
- ✓Works with LangChain, CrewAI, LangGraph, OpenAI Agents SDK, MCP, A2A and plain HTTP
- ✓Open-source Python SDK: pip install agf-sdk
- ✓Built on AAP, an open standard anyone can implement
from agf import AgentGovernance agf = AgentGovernance(api_key="…", org_id="org_acme") result = agf.authorize( agent_id="did:agf:agt_01abc", action="read:calendar", resource="calendar/team-standup", ) if result.allowed: # proceed
— Inside AXIS
The interface, the evidence, and the rules.
Real screenshot: the AXIS dashboard's analytics view, demo organization.
- decision
- ALLOW
- action
- read:calendar
- resource
- calendar/team-standup
- chain
- human → agent → sub-agent
- evidence
- structured, replayable
- signature
- ✓ verified
Every decision leaves evidence, not a log line.
DELEGATION · CONCEPT
Authority only ever narrows.
Illustrative example of the AGF delegation model (Spec 01).
PROTOCOLS
In front of every call.
Gateways put an authorization decision ahead of tool calls, agent-to-agent delegation and plain APIs.
— The company
A product company with a research engine.
Intelligent software is moving from answering questions to taking actions. That shift creates opportunities for useful products and new demands for authority, safety, and accountability.
INTREXA develops products for specific operational needs while researching the deeper problems that will shape future systems. Each product has its own identity; the company connects them through engineering and R&D.
INTREXA TECHNOLOGIES (OPC) PRIVATE LIMITED · Chennai, India
— Products
Purpose-built AI software.
We give each product a clear job. The portfolio shows its stage so visitors can tell what exists today and what is coming next.
AXIS
Authorization infrastructure for AI agents. Evaluate authority before sensitive actions, enforce decisions at execution, and connect outcomes to evidence.
FORGE
Our next product is taking shape. We will publish its purpose and capabilities when the experience is ready to show.
The portfolio can grow as new research and customer needs lead to focused products.
— How we build
Research informs products. Products test the ideas.
Our work moves from a concrete problem to a model that can be tested, then to software a team can use. AGF and AXIS show that path in practice.
01 / INVESTIGATE
Define the problem
Study real operational boundaries and identify what current tools leave unresolved.
02 / MODEL
Make the rules explicit
Develop specifications, prototypes, and evaluation methods that can be examined and improved.
03 / SHIP
Build the product
Turn validated ideas into focused software with a clear user, workflow, and responsibility.
— Explore the system
From an open model to an enforced decision.
See how a research question becomes a practical control. This interactive illustration shows two views of the same work.
Illustrative flow · no live authorization request is made
A shared model defines how authority, delegation, and evidence should be represented.
INTREXA R&D
Work on the questions behind the next product.
Our current focus is the behavior of autonomous systems: who gives them authority, how decisions remain valid, and how systems can trust one another. Research outcomes may become open work, prototypes, or products.
See our open work at AGF ↗How an agent's identity and delegated permissions can be represented and checked.
How an approved action remains valid when it actually reaches a tool or resource.
How agents, tools, and organizations can exchange authority across boundaries.
How to observe and contain harmful behavior as systems operate with less supervision.
— Published work
Ideas you can inspect.
Selected writing from the AGF open work supported by INTREXA.
Authorization gates for OpenAI Agents SDK tools
A concrete tool boundary, outcome reporting, and a caveat discovered in integration.
Read article ↗IDEAS · JUL 29, 2026Explainability tells you why. Authority tells you whether.
Why a plausible agent decision still needs an explicit authorization answer.
Read article ↗STANDARDS · JUL 12, 2026The AGF specifications are now public
Open specifications for delegation, revocation, audit, and interoperability.
Read article ↗
— Open work / AGF
Shared foundations for agent governance.
Agent Governance Foundation (AGF) is the open framework and specification work supported by INTREXA. It explores identity, delegation, authorization, revocation, and evidence. AXIS is a separate commercial product that implements agent authorization for operational use.
Read the AGF framework ↗— Explore INTREXA