AI Agents
The module connects any AI agent to the system — cloud-based or deployed in your own infrastructure — and gives it access to documents, reference data and processes under the same rules as a person.
An assistant where the process is fuzzy
Some work in a company is hard to formalize: reading a contract and spotting what has changed; finding a similar project; choosing the right assignee or an equivalent item; pulling a report together from scattered data. These are fuzzy processes, where ordinary rule-based automation is powerless and people spend hours. The AI Agents module covers exactly these: you connect a model — ours, a third-party cloud one or a local one — and assign it a role in the system. The agent sees only what it has permission to see, every action is logged, and the decision always stays with a person: the agent prepares, suggests and highlights — an employee approves.
Connecting a model
Permissions and limits
Task for the agent
Analysis
Suggestion
Human decision
What the module can do
10 feature blocks available once the module is enabled.
Connect any model
Cloud APIs, local models in your own infrastructure, several providers at once. Switch without reworking your processes.
Agent and role registry
An agent is set up as a participant in the system: its own role, access rights, spending limits and permitted actions.
Document pre-analysis
Data extraction, classification, a summary, and highlighting of risks and non-standard terms before the document reaches legal counsel.
Comparison and version control
What changed between revisions, how the document differs from the template, and which clauses dropped out of the approved version.
Matching and semantic search
Similar projects and solutions, equivalent items, suitable assignees by workload and skills, and archive search by meaning rather than keywords.
An agent in the approval route
The agent is added to the route as a separate step: it checks completeness, gives an opinion and flags contentious points for a person.
Drafts and replies
A draft letter, meeting minutes, explanatory note or reply to a counterparty — based on system data, not invention.
Data validation
Inconsistencies in reference data, duplicate counterparties and items, and mismatches between a document and the books.
Explainability and logging
You can see which documents the agent read and what its conclusion is based on. All requests and responses are stored, and the output is marked as machine-generated.
Data privacy
With a local model, data never leaves the perimeter. For cloud providers, you configure which fields and documents they can access.
Modules it works with
The module can be bought on its own, but it shines in combination: data flows between modules automatically.
Feature demo
This section is set up for screen recordings. Videos are added as they are produced.
Contract pre-analysis
Comparing document revisions
An agent as a route step
Semantic search in the archive
How to buy this module
The module is licensed separately or as part of the system, by subscription or perpetually. You can estimate the cost for your setup in the calculator.
With a server
Client-server edition: collaboration, sync, chats and conferencing. The server runs on your side, on ours, or in a hybrid setup.
Local, no server
Installed on a workstation with its own database. Standalone operation, perpetual license; the server side can be added later.
Subscription or perpetual
A monthly fee with updates and support, or a one-time purchase with support under a separate contract.
With implementation
Process setup, data migration from your current systems and user training — as a separate stage.
Let us show you what already works
The module is in development. We'll show you the current build, the roadmap and the terms for joining a pilot project.