Agentivity documentation

Agentivity is a platform to build and run AI agents, AI teams and workflows, without writing code. These pages explain what it is, how to install it and how to use it.

1. What Agentivity is
2. What a Solution is, and how to mark one
3. What you can do with Agentivity
4. Three ways to use Agentivity
5. Choose between an agent, a team and a workflow

Also in this section: how a person stays in control, what the Community edition includes and the words Agentivity uses and where each kind of information goes.

1. Install Agentivity with Docker: run the free Community edition on your own machine or server.
2. Follow the welcome guide: the four short steps you see the first time you open the Studio.
3. Try a ready-made solution: import a showcase and see how it is built.
4. Build your first team with the Assistant: describe what you need and get a draft.

Agents

• What an agent is made of
• Create an agent
• Write the role and the instructions of an agent
• Choose or change the model of an agent
• Give tools to an agent
• Give an agent knowledge and a memory
• Let an agent ask a person before it decides
• Run an agent and read its answer

Teams

• Choose how a team works together
• Create a team of agents
• Name, group and give an icon to team members
• Use a team or a workflow as a member
• Run a team and read its work

Workflows

• What a workflow is made of
• Create a workflow
• Add an AI step to a workflow
• Use an agent or a team as a step
• Ask a person to approve or answer
• Ask a person to fill in a form
• Add a condition to a workflow
• Repeat steps for each item of a list
• Run a workflow and follow each step
• Keep texts and find them by meaning

Data Tables

• What a Data Table is for
• Create a Data Table
• Look at a Data Table and change its columns
• Read and write a Data Table from a workflow
• Let an agent read and fill a Data Table

Knowledge Bases

• What a Knowledge Base is for
• Create a Knowledge Base
• Keep a Knowledge Base up to date
• Look inside a Knowledge Base and try a search
• Change, empty or delete a Knowledge Base
• Give a Knowledge Base to an agent or a prompt

Prompts

• What Prompt Studio is for
• Write a prompt and its inputs
• Try, test and improve a prompt

• The four showcases and what each one teaches
• Invoice Clerk: one agent with tools and a table
• Shopping Lab: nine agents one after another
• Trip Manager Team: a manager and its specialists
• CV Screening: a workflow with AI steps

• What a VibeApp is
• A VibeApp or a classic app, what changes
• One app for one job, or one portal for all
• Show a team in the portal
• For developers, the React SDK: how a VibeApp talks to Agentivity, what the SDK offers and how to work with it, start with the SDK, run a solution from your app, show the team at work, the widgets an agent can show, add your own widgets, give the app your brand, before you put it online

• Find your way around the Studio, then each screen: Home, agents and teams, workflows, an editor, Executions, Templates, Settings, status bar and theme
• Connect an AI model
• Add, change or remove credentials
• Share your work as a Template
• Organise your work in folders
• Reference: every node of the Nodes catalog, the routes of the API, and the settings of the server

• Update Agentivity and back up your data
• Keep your Agentivity server private
• Fix a failed installation
• If an antivirus blocks the connections of Agentivity

• Common questions about Agentivity
• Questions about building and running
• Questions about licence, security and editions

• These pages describe what Agentivity does today. Anything planned is labelled as planned.
• For examples of what people build with it, see the use cases.
• To see how the Studio works before you install it, read how it works.