Guides
A guide is a filter over the index, not a separate opinion. Each one names the capability, the criteria a product had to meet, and the date of the pass. If too few products qualify, the guide says so instead of padding the list.
An agent that needs you to approve every step is a chat window. This list is ordered by our score, which weighs evidence completeness and adoption — not by who has the loudest launch video.
AI Agents · 4 criteria
12
qualified · pass 30 Oct 2026
The best note taker is the one you forget is running until you need the quote. We rank on capture, search and whether the price is visible before you sign up.
Memory · 4 criteria
12
qualified · pass 30 Oct 2026
Completion, chat and autonomous agents solve different problems. This list mixes all three but labels which one you are getting.
Coding · 4 criteria
8
qualified · pass 30 Oct 2026
Translation is a latency product. Everything here claims real time; the difference is whether it survives a real conversation.
Communication · 4 criteria
10
qualified · pass 30 Oct 2026
Three families, three very different evidence bases. Cheapest first is usually the right order here, and we say which family each product belongs to.
Focus · 3 criteria
9
qualified · pass 30 Oct 2026
Always-on capture is a consent problem as much as a hardware problem. This list includes what shipped and what stalled.
Memory · 4 criteria
12
qualified · pass 30 Oct 2026
A tracker tells you what happened. An intervention changes it. Both appear here, labelled.
Health · 4 criteria
8
qualified · pass 30 Oct 2026
The panel is the easy part; the reference ranges and the follow-up are the product. We say what each operator publishes.
Longevity · 4 criteria
3
qualified · pass 30 Oct 2026
These are the products where AI is already doing real work. Latency and offline behaviour matter more than model size.
Accessibility · 3 criteria
4
qualified · pass 30 Oct 2026
A confident paragraph with no citation is a liability. Everything here links to what it read.
Decision-Making · 4 criteria
1
qualified · pass 30 Oct 2026
The category is half product, half provocation. We list what exists and note where the claim is only a landing page.
Digital Twins · 4 criteria
3
qualified · pass 30 Oct 2026
Speaking practice is the bottleneck, and it is the thing most apps avoid because it is expensive to serve.
Learning · 3 criteria
5
qualified · pass 30 Oct 2026