Deploy Agents
Hand a task to an autonomous agent and get on with your day. It works through visible steps using your own data, then reports back — or acts on your behalf if you let it.
- 192 ready-made templates, or write your own
- Report mode brings findings back; managed mode acts for you
- Agents run in parallel — more of them on higher tiers
- Every agent publishes the exact instructions it runs on
An assistant answers a question. An agent does a job. The difference matters when the work takes longer than you want to sit and watch — reading a month of spending, going through a reading list, reviewing a diff, pulling a week into a brief.
You give an agent a mission in plain language and it works through it step by step, drawing on the data you have already given Tase. You get the result when it is done.
Start from something that already exists
There are 192 agent templates in the library, grouped into 19 categories — health, finance, product, law, operations and the rest. Each one is a persona with a job, and each is a starting point you can edit rather than a black box.
We publish the full instructions every agent runs on, on its own page, before you deploy it. You can read exactly how it is told to behave, and change any of it.
Two modes, and the difference is who decides
- Report mode: the agent does the legwork and brings back findings. Nothing in your data changes until you say so.
- Managed mode: the agent acts — filing what it found, organising, finishing the job.
Most people leave anything touching money or health in report mode and let the research agents run managed. That is the right instinct and the product does not argue with it.
They run in parallel
What you would actually use it for
- Turning a month of transactions into a plain-language answer about where the money went
- Reading the articles you saved and returning the three that were worth it
- Reviewing a code diff and explaining what it would change, and why
- Drafting a plan out of notes that are currently scattered across a week
Can an agent see all my data?+
Only the areas it is scoped to. Each agent declares what it reads — health, finance, tasks, goals or your second brain — and that is published on its page.
What stops an agent doing something I did not want?+
Report mode, which is the default for anything consequential, changes nothing on its own. Beyond that, you set per-domain autonomy rules in Settings.
Can I write my own agent?+
Yes. The templates exist so you do not have to start from nothing, but a custom persona and instruction set is a first-class thing.
Do agents keep running when I close the app?+
Yes. That is the point — you hand off the work and come back to the result.
Try it
Free to start, no card. Everything on this page is in the app today on iPhone and the web.
Start free