Daily brief · 11 November 2026
What moved on 11 November
Agents are being re-pointed at browsers, inference is splitting into two price tiers, and procurement has quietly become the gatekeeper of the whole category. Three moves, one number to watch.
Three moves
1. The agent conversation moved to the browser
The demos that circulated this week were not chat windows. They were browsers being driven by something that had to hold a session, respect a permission scope and leave a record. That is a harder engineering problem than prompt design, and it explains why the agent category is consolidating around a handful of infrastructure players while the wrapper layer thins out.
2. Pricing split into two tiers, and the router became the product
Frontier pricing held; small-model pricing kept falling. Buyers responded by splitting workloads rather than switching vendors, which puts a routing layer between them and the model. Nobody has audited that layer yet, and it is where the next cost surprise will come from.
3. Procurement is doing the product filtering
The question changed from 'can it do the task' to 'can we reconstruct what it did'. Vendors with a readable action log are clearing review faster than vendors with a better demo, which is an uncomfortable fact for anyone selling on capability alone.
The index, at 10:00 UTC
| Index | Value | 24h |
|---|---|---|
| Momentum | 742 | ↑ 1.8% |
| Capital | 688 | ↑ 2.4% |
| Capability | 811 | ↑ 0.6% |
| Adoption | 703 | ↓ 0.4% |
Momentum is the composite. Capital and capability are running ahead of adoption, which is the shape of a market that is building faster than it is buying — and the reason procurement, not product, keeps showing up in the top ten.
What to watch tomorrow
- Whether any major vendor publishes a routing policy instead of a price list.
- The first procurement rejection that cites audit logs rather than accuracy.
- Whether voice input shows up as a default in an operating system update, not an app.
How to read this
The board ranks themes, not companies, and the score measures attention rather than merit. A high reading means a lot of credible people are talking about it today; it does not mean the technology works. Where a claim could not be traced to a primary source, it was left out.
Figures are editorial estimates, not market data, and nothing here is investment advice. The full method and weights sit on the index page.
The index is rebuilt every morning at 10:00 UTC. If you want the previous days, they are listed on the index page.