At the heart of iTeri is our unique and innovative client-side agentic AI architecture. By operating directly inside the user's browser we have been able to bring a new level of context awareness and power to the world of AI copilots.
Based on cognitive science principles, our agents operate via a continuous Perception > Decision > Action loop - taking them beyond mere orchestration and into the realm of truly autonomous agentic AI.
Our unique tool-calling architecture goes beyond traditional LLM APIs to deliver new levels of speed, stability, and cost efficiency. Combined with streaming responses this architecture allows our agents to respond 70% faster than all other browser agents in the market - delivering a truly seamless user experience.
By accessing your product at the DOM-level, our agents are able to accurately target all application elements, handle dynamic UI changes, and execute actions with the full power of the browser's built in capabilities.
Our agents leverage advanced context management techniques including: sliding windows, context-aware memory recall, auto summarization, and a whole host of other approaches that give them the edge.
Unlike other agentic systems, iTeri runs inside the client-side of your product. By working in the browser, the agentic pipeline only needs to touch our server-side for agent configuration and contextual knowledge.
LLM calls can be routed to any LLM provider you choose - all under your own DPA and API contract. This can be through our secure managed service - or through your own hosted proxy - meaning we never handle ANY of your user's data.
All user auth remains untouched and continues to be handled by your product. Our agents simply act on the user's behalf and so inherit all of the same permissions and access rights.
By default, our agents can only perform actions within the areas of your product that you explicitly allow them to - by a mixture of blacklists, whitelists and guardrails you maintain full control over what your agents can and cannot do.
Our systems only need to store and serve the product guides, design choices and guardrails you provide us. By hosting your own LLM proxy, we never need to handle ANY of your user's data - ensuring that your product remains compliant with ALL of your security requirements.
iTeri is designed to work with your product out of the box. However, the true power emerges when your engineering team extends it with your own custom capabilities.
The WebMCP standard, being developed by Google and the rest of the W3C community, lets web developers expose powerful client-side tools to browser-based agents - giving them the ability to interact with your application in new and powerful ways. iTeri is one of the first agentic platforms to fully support WebMCP - giving your developers the ability to turn your agent into a truly integrated AI copilot.
Similarly, MCP and A2A are now the de-facto standards for connecting AI agents to backend tools and services. With iTeri, you can add as many of these connectors as you need, quickly and easily. Your agents can move beyond just using your standard product and can start offering your users back-end capabilities without needing to wait for new UX capabilities to be developed.