Things That Think / AI agents & automations
AI agents & automations
Software that does a job on its own: reads the inbox, fills the sheet, chases the invoice, answers the question, runs every night at three.
Who needs this
Teams drowning in repetitive work: support triage, lead routing, moving data between tools, weekly reports, anything with a rule and a trigger attached to it.
What you get
- An agent running on infrastructure you control, using your real accounts and data, never a demo dataset
- The tools it can use, wired to your actual APIs
- Guardrails: what it may do, what it must never touch, and what a person sees before it acts
- A log of every run, replayable, so you can see why it did what it did
- A weekly cost estimate based on your real usage, not a guess
How long it takes
2–6 weeks for a first useful version.
The price is written into the quote after the call, and it is fixed before any code is written. There is no price on this page because it would be a guess.
Book a callWhat is included
- Agents that act on a trigger. It reads the inbox at three in the morning and does the job, or it does not ship.
- Your APIs as tools. The agent calls the systems you already pay for, with the credentials it needs.
- Guardrails and approvals. What it may do, what it must never touch, and what a person sees before it acts.
- Memory across sessions. It remembers the last conversation without remembering everything forever.
- Replayable run logs. Every run is recorded, so you can see why it did what it did.
- Real cost per run. A weekly estimate from your actual usage, not a guess.
- Retries and backoff. A flaky API does not become your agent's failure.
- Evaluation of the output. A way to tell whether the answer was any good.