The Interface Split

Microsoft launched Web IQ on June 2. Not a Bing update. A separate search system built for agents. No rankings, no blue links. Structured evidence objects an agent can reason with mid-task without opening a browser.

It already powers parts of ChatGPT’s web search and Microsoft’s Copilot. This is production infrastructure, not a research preview.

The move is familiar if you’ve watched infrastructure evolve before. APIs split from user interfaces decades ago. Nobody expects a machine to navigate a dashboard or a human to call a REST endpoint. The users needed different things, so the systems diverged. That was obvious — in retrospect, it always is.

Search is now getting the same treatment. So will data access, internal tools, and every system that both humans and agents touch. The question is whether your organization recognizes the split or keeps forcing agents through human-shaped interfaces and wondering why they underperform.

I run agents that pull web data daily. Until recently, the workflow was: search like a human, parse the page, hope the right paragraph is in there. Web IQ replaces that with a direct pipeline from index to agent. No browser in the middle.

If your agents still access external information the same way your team does, you have found your bottleneck. You just haven’t named it yet.