Ask. A clear question defines the goal and scope.
Step 1 of 3 · AskAn assistant inside the suite (1 of 3)
Preview taken in a demonstration workspace. All data shown is fictitious.
Your instructions. Your sources. Your agent.
Create assistants with their own instructions, sources and tools. Each Suite 366 agent works only within the access you authorize, to research, prepare and help with your tasks.
- Document assistant
- Work preparation
- Scheduled research
Features
A mission, tools, a memory. Within the frame you set.
1The agent
Each agent has its role.
An agent is set up with an identity, instructions and a model, then shared with the people who use it.
A name, a face, a speciality
Each agent has its name, its face and its mission. Its speciality sets its behaviour and its access: general assistant, calendar, documents, mail or orchestrator.
Instructions and a model
Its instructions set its role, its tone and its limits. Its model is picked among those of your organization, and a slider tunes its answers from precise to creative.
Tested, then shared
A test panel talks with the agent next to its settings. The agent then stays private, or opens to a team or to the whole organization.
2Its tools
It reads, it searches. Then it acts.
In agentic mode, the agent chains several actions to carry a task through, with the tools its permissions open.
The modules of the suite
Files, mail, calendar, CRM, pages, forms, meetings, chat and workflows: the agent reads and writes where your team works.
Your documents as context
Before answering, the agent finds the useful passages in the folders opened to it and shows its sources.
Files read in full
A scanned contract, a presentation, two versions of the same document: the agent reads every page and compares the differences line by line.
The web, in a real browser
The agent searches the web, opens pages, clicks and fills in forms. You watch its browser live and take over for a sign-in.
Code in a sandbox
Python, Node.js or Bash run in an isolated environment, and the files produced land in your Drive.
Your connected services
Connectors link the agent to your accounting, your code repositories, your messaging or your databases, right by right. MCP servers add your own tools.
3Its memory
It remembers. From one conversation to the next.
The agent retains what matters for each person and links the names, companies and projects it comes across.
Your preferences, your projects
Preferences, ongoing projects, useful information: what the agent retains about you serves all your agents, and its memory page shows it.
A knowledge graph
People, companies, projects: the agent links the entities of your documents and conversations, and relies on those links to answer.
A history that lasts
Every conversation is found by search and continues as a branch. The longest ones summarize themselves: the agent keeps the context.
4Its autonomy
The work moves on while you are away.
A scheduled task, sub-agents in parallel, a team of specialized agents: the agent carries out its mission and reports back.
Scheduled tasks
Every Monday at 9 a.m. or on a given date, the agent runs its mission and notifies you. One sentence in the conversation is enough to create the task.
Sub-agents in parallel
For a broad job, the agent splits the research between sub-agents that move forward at the same time, then gathers their reports.
A team of agents
An orchestrator hands each part to the right specialized agent and assembles the answers.
5Its channels
It answers where the question comes up.
In its chat, next to a document, on Telegram or from your own applications.
A chat that shows the work
While the agent works, its face comes alive and each tool appears with its result. One button stops the turn at any moment.
Next to what you are reading
File, page, CRM, dashboard: the agent opens in a panel, with the document in view. With dictation, you speak to it out loud.
On Telegram
Linked to a Telegram bot, the agent answers from the messaging app according to its instructions. Your data stays in the suite.
Through the API
One key per agent opens a widget to place on your website and an API for your applications. The agent can also be called as a model, from an OpenAI-compatible interface.
6Its frame
You decide what it touches.
Every tool depends on a permission. The agent acts on behalf of the person using it, within that person’s own rights.
Permissions per resource
Read, create, update, delete, execute: each access is switched on resource by resource, from files to the CRM.
A chosen scope
A folder, a team, a CRM board: the agent’s access focuses on the resources you designate.
Every action can be read
Tool, arguments, result: the detail of each action stays in the conversation. The agent map shows what is running and stops a sub-agent in one click.
Usage in figures
Conversations, messages, users, response time: the dashboard of each agent tracks its activity.
7Your uses
Three ways to use it.
Document assistant
Ask questions about a defined set of resources.
Work preparation
Gather information before a meeting or decision.
Scheduled research
Run a recurring task and read its report.
Get started on video.
Tutorials from the Suite 366 team, narrated in French.
Runs on your premises.
Where this application runs, and the technologies that come with it.
- On-premises
- On your servers, with a configuration sized for you.
- DIWY
- Available on DIWY, depending on your configuration. Discover DIWY
- Requirements
- Suite 366 module with configured models and tools. A PersonIA gives an agent a colleague’s role in the team.
- Works with
In Suite 366.
This module is part of Suite 366. It works with the other modules, in the same workspace.