What this is
A page per tool we write about, holding only what we have published: the surface we measured, and the results from our own tests. Vendor pages change. These notes are the snapshot we stand behind, dated.
Tools
Clay
Research and enrichment: one table where an agent or a person fills columns from many providers.
EmailBison
The send layer: campaigns, leads, replies, mailboxes and warmup behind an API-first product.
Grok Bot
A bot platform we use as the front door of an AI management team, not as a single assistant.
Claude Code
A coding agent we run as the interchangeable home base: a CLI, MCP on both sides, and a harness an agent can drive.
Codex
A coding agent we run as the other home base, and the grader behind our published benchmark labels.
TinyFish
A search API we run in production next to Google for funding-round discovery.
Keenable
A search API that measured best as the second source next to Google for funding rounds.
DeepSeek
A search API whose raw results were the cleanest of the four we benchmarked.
How to read a note
Surface claims come from the 47-tool table: whether there is an MCP server and who built it, whether there is a CLI, and how far the API goes. Result claims come from a measured test with a stated method. Where we have neither, the page says so rather than repeating a vendor promise.
