AI agents by use case
The same work turns up in every industry under a different name. These group the agents by the job rather than by the trade, so a revenue leakage auditor and a freight invoice auditor sit together.
Customer support
18 agents
Agents that read a queue and decide what actually needs a person. They triage, draft, and escalate — and they are built to refuse to promise a refund, an extension or an exception on your behalf.
Lead generation and qualification
11 agents
Agents that answer an enquiry before it goes cold, qualify against criteria you set, and find the disqualifier early. They work from what the prospect actually said rather than from a scoring model nobody can explain.
Compliance monitoring
78 agents
Agents that check work against a rule that exists — a policy, a regulation, a signed obligation — and report what fails and why. Each documents its reasoning so an auditor can read it, and none of them decides on your behalf.
Reporting and analysis
60 agents
The largest group in the catalogue. Agents that read the source rather than the summary, state their assumptions, and say what happens if those assumptions are wrong — instead of producing a confident number with nothing behind it.
Contract and document review
25 agents
Agents that read long documents for the clause that matters: the obligation running your way, the exclusion that applies, the term that lapses if nobody acts. They find and cite; they do not advise.
Hiring and people operations
9 agents
Agents that screen against the criteria you wrote down, structure an interview, and surface what the process is actually selecting for. They are built to leave the judgement with a person.
Research and due diligence
37 agents
Agents that go and read the sources, then report what they found and where they found it. Each one cites what it used, separates what the source says from what it inferred, and says plainly where the evidence ran out — so the answer can be checked rather than trusted.
The catalogue does not cover this well enough yet. Rather than publish a page with two agents on it, we would rather build what is missing. Describe what you need