Skip to content
Agentivity
Star on GitHub Start free

What you can do with Agentivity

Agentivity handles work that has several steps and needs judgement on the way: reading, comparing, writing, checking, and asking a person when a decision is theirs. This page shows what that looks like.

Agentivity offers four showcase solutions. You import one in a few clicks and you see how it is built. See Try a ready-made solution.

ShowcaseThe needHow it is built
Invoice ClerkRead the supplier invoices you upload, file each one and check its figures.One agent and one data table.
CV ScreeningCompare a CV with a job description and propose what to do with it.One workflow.
Shopping LabCompare a product across stores and write a recommendation.One team of nine agents that work one after another, and five data tables.
Trip Manager TeamPlan a full trip: flight, hotel, car, dining, transport. The bookings are simulated.One team of thirteen agents led by a manager, and two data tables.

The four examples use the three building blocks: one agent, one workflow, two teams.

  • Reading and filing. Read documents, take out the facts, check them, store them in a table.
  • Preparing and writing. Gather information, write a note, a reply or a report, and have another specialist review it.
  • Comparing and recommending. Look at several options against your criteria and say which one fits.
  • Handling a request from start to finish. Understand what a person asks, send each part to the right specialist, and come back with a complete answer.
  • Work that needs your approval. Do everything up to the decision, then stop and wait for a person.

You do not need a large project to begin.

  1. One agent that takes one repetitive job off your hands.
  2. One team, when the job needs several skills, or when one specialist should check the work of another.
  3. A workflow, when the steps must be the same every time and a person must approve before the work continues.
  4. Teams that use other teams. A member of a team can be another team or a workflow.
  • It does not start by itself. An execution starts when a person starts it in the Studio, or when an application asks for it. Starting executions on a schedule is planned, and it is not available yet.
  • It has no login. Agentivity does not know who is using it. Anyone who can reach your server can use everything on it. Keep it on a network you trust.
  • It does not provide the AI model. You connect your own.