We are introducing a new colleague that can take over and execute a significant amount of administrative tasks. Built on AutoMeta's resources and the intelligence of large language models, collAIgue provides dedicated workforce capacity for business administration.
Processes incoming documents and emails, identifies their category and type.
Extracts relevant data, auto-files, categorizes, and routes.
Executes complete business processes without human intervention.
Performs every task regardless of language.
The collAIgue colleague brings experience and references from processing hundreds of thousands of documents, ready to take on new challenges.
Extraction of invoice data, determination of period and validity, line-item invoice identification.
Processing of incoming email text and attachments within a unified context.
Analysis of health and laboratory data, extraction of personal document information.
Automatic image categorization, summary generation, and tag assignment.
Extraction and linking of data from contracts, purchase orders, and delivery confirmations.
Search across document repositories, result evaluation, and report generation.
Determination of filing metadata: type, sender, recipient, subject, date, identifier, Dublin Core data.
Speech-to-text conversion and text summarization.
Joint handling of different document types and files received in a single package, with context-aware processing.
What's the difference between collAIgue and online LLM APIs (ChatGPT, Gemini, Claude, Mistral…)?
| Aspect | collAIgue | Online AI API |
|---|---|---|
| Deployment model | Physical appliance within the client's network | Cloud-hosted service |
| AI execution location | On-premise, on dedicated hardware | On external provider infrastructure |
| Internet connection | Not required | Mandatory |
| Access model | Autonomous processing component | Service accessible via API calls |
| Operating model | "AI worker" in workflow model | Stateless API endpoint |
| Aspect | collAIgue | Online AI API |
|---|---|---|
| Data leaves the organization | No | Yes |
| Internet connection | Not required | Mandatory |
| Data storage | Does not store data during operation | On provider infrastructure |
| Network protocol | Connects as client to repositories | Public API endpoint |
| Data control | Entirely with the client | Shared between provider and client |
| Interface | Function | collAIgue Support |
|---|---|---|
| CMIS, REST API | Document repository integration | Alfresco, SharePoint, FileNet |
| WebDAV | Document retrieval and write-back | Fully supported |
| Microsoft Graph, IMAP | Email processing | Fully supported |
| SQL connection | Database processing | Fully supported |
| File system | Folder and file processing | Fully supported |
| Aspect | collAIgue | Online AI API |
|---|---|---|
| IT operations | Minimal, no supervision required | None |
| Hardware | Required – runs on dedicated appliance | None |
| Updates | Package-based, full control | Provider-managed |
| Skill requirement | No AI expertise needed | Required: models, prompts |
| Model selection | Not required – provider chooses | Required – among many models |
| Model changes | None – guaranteed determinism | Ongoing – model updates |
| Training | Included with the appliance | Must be assembled individually |
Activities of colleagues created in the collAIgue system can be organized into workflows, breaking down complex processing into simple steps.
A business-level task description is created, defining the agent's responsibilities.
Input document pre-processing: OCR, barcode recognition, image transformation.
Selection of data source (file system, SQL, CMIS…) and target storage protocol.
Assembly of a reference package of 10–100 documents matching the task types.
Model selection, fine-tuning, and testing on reference documents.
The tested agent is deployed to the collAIgue appliance, with input/output data sources configured.
AI is "locked in a box," performing entrusted tasks under full control. No confidential information can leave it.
Not a server, not a service accessible by others. Does not reach out to the Internet.
Does not connect to the Internet.
Does not store data during operation; client data is kept completely secure.
All configurations reside on removable media or the client's Git repository.
Operations are governed by the AutoMeta framework, ensuring full traceability.
Continuous logging with timesheets showing completed tasks and time spent.
Request a demo and upload documents to our demo server, then see how collAIgue processes them in real time.
Request Demo → info@broadbit.hu