Eight example companies run through the AI control finder

How were the eight profiles chosen?

They cover the combinations we expect readers to bring: browser AI alone, coding agents with production access, a Palo Alto Networks shop, a SentinelOne shop, cloud workspaces behind an SSE, a handful of people using AI, and unmanaged devices where nobody is sure what is in use. The default profile on the finder page is the first: browser AI and coding agents on managed devices, with a preference for coaching.

What does each profile produce?

What do the results show?

First, the starting control type follows the AI tools in use more than anything else. Profiles with coding agents start at on-device control, because coding agents act on the laptop. The profile with browser AI and coaching produces a tie between the device and browser positions, and both are picked.

Second, the first group of products changes with the answers. A company already running Palo Alto Networks or SentinelOne sees the product built into that platform come first, Koi or Prompt Security, because the finder gives four points for a product built into something you already run and both products' pages describe the coding agents they cover. A company behind an SSE with cloud workspaces sees the product that describes proxy, gateway and browser positions come first. A company that wants to coach browser AI users sees the product built around block, warn and log come first.

Third, Bay is in the first group in some profiles and not in others. It comes first alone where coding agents meet production access and a preference for blocking, because its pages describe tool-call decisions, rollout through existing EDR or MDM and an action record. It is second in the Palo Alto Networks and SentinelOne profiles, behind the product built into that platform, and it is not in the first group where browser AI, coaching or an SSE dominate, because it describes no browser, network or gateway control.

Fourth, two profiles point to policy before a product: a handful of people using browser AI with general information, and unmanaged devices with no clear view of AI use. For those, the finder opens with a panel saying you may not need a product yet, then lists products in case you still want to look.

Which answer moves results the most?

Across the eight profiles, the AI tools question moves the starting control type the most, and the stack question moves the product order the most. Data sensitivity comes next: regulated data and production access switch on the check for an action record, which separates products whose pages describe who or what took each action from those that do not. The appetite question matters less to the control type than people expect, but it decides which response check applies, and that often splits products that otherwise look alike.

Why show equal fits instead of breaking ties?

Several profiles end with two or five products on the same fit. We could break the tie with a hidden rule, but any rule we chose would be a preference dressed up as a result. Equal fits are listed alphabetically and tagged as equal, and the product notes and head-to-head pages show where they differ.

What should you do with this?

1. Find the profile closest to your company and read its result.
2. Then run the finder with your own answers; one changed answer can change the order.
3. Open the first group in the line-up page and read the rows tagged Differs.
4. Take the limits listed in each product's notes into your vendor conversations.

What are the limits of the examples?

Real companies rarely fit one profile. A software company may have coding agents in engineering, browser AI in sales and contractors on their own laptops, all at once. The finder takes one set of answers at a time, so the useful approach is to run it once per group with a distinct situation and compare the results. Where the groups point to different control types, that is a sign you may need two controls, or one product that describes both positions. The note on chat tools and coding agents covers that choice in more detail.