Document Generation
Queria lets you generate professional documents directly from chat, leveraging AI to extract data from your files, fill predefined templates and produce ready-to-use DOCX documents. This feature turns a process that normally takes hours of manual work into an interactive dialog of a few minutes.
How it works
Document generation relies on a guided conversational flow. You don't have to fill forms or navigate complex menus: just ask the assistant what you need.
Full flow
1. Natural language request
Write in chat what you want to generate. Examples:
"Generate explanatory notes based on the balance sheet I uploaded" "Create meeting minutes from the attached notes" "Prepare an audit report based on the documents in the Quality topic"
The assistant automatically recognizes the generation intent and starts the guided process.
2. Template proposal
The system analyzes your request and proposes available templates matching the requested document type. For each template you see:
- Name and description of the template.
- Category (balance, construction, legal, audit, public administration).
- Number of sections and final document structure.
- Data required for filling.
Select the most suitable template by clicking the corresponding card. If no template matches, the assistant can generate a free-structure document.
3. Data sources identification
Once the template is chosen, the system automatically identifies the available data sources:
- Documents uploaded in the current conversation (temporary attachments).
- Company Knowledge Base.
- Standard documents assigned to relevant topics.
The assistant shows a summary of the identified sources and asks for confirmation before proceeding to extraction.
4. Hybrid data extraction
This is the phase where Queria analyzes your documents to extract the needed information. The system uses two complementary approaches:
- Structured extraction: for files organized as Excel sheets, CSV or tables, the system reads the data directly with high precision.
- AI extraction: for unstructured documents (narrative PDFs, minutes, emails), the AI interprets the text and identifies relevant information.
At the end of extraction, the system presents a summary of found data with the confidence level for each value.
5. Confidence levels
Every extracted datum comes with a confidence indicator:
| Level | Meaning | Recommended action |
|---|---|---|
| High | The datum was found clearly and unambiguously. | Quick verification, generally reliable. |
| Medium | The datum was interpreted from context but may need confirmation. | Check the value and confirm or correct. |
| Low | The datum was inferred with uncertainty or found in potentially out-of-date sources. | Check carefully before confirming. |
| Missing | The datum was not found in any available source. | Provide the required value manually. |
6. Handling missing data
If the system doesn't find all the necessary information, the assistant asks targeted questions to fill the gaps:
"I couldn't find the company's share capital. Can you tell me?" "The assembly date is not in the documents. When did it take place?"
You can reply in natural language and the assistant will integrate the information into the document under generation. This iterative dialog continues until all required fields are filled.
7. Preview
Before final generation, the system offers two preview modes:
- Per-section preview: view each document section separately, with the option to edit, approve or reject each.
- Full preview: view the whole document in chat for an overall review.
During preview you can request specific changes:
"In the Income Statement section, correct the revenue: the right value is 1,250,000 euros" "Add a paragraph on depreciation policy in the Evaluation Criteria section"
8. Generation and download
Once preview is confirmed, the system generates the final document in DOCX format:
- The file is created with professional formatting.
- A download button appears directly in chat.
- The generated document is also saved in the conversation history for future access.
Template categories
Queria includes templates organized by sector and type:
Balance and Finance
- Explanatory notes (full and abridged).
- Management report.
- Cash flow statement.
- Financial analysis report.
Construction
- Bill of quantities and cost estimate.
- Technical report.
- Special contract specifications.
- Site diary.
Legal
- Legal opinion.
- Defense brief.
- Standard contract.
- Agreement minutes.
Audit and Quality
- Internal audit report.
- Corrective action plan.
- Compliance analysis.
- Verification checklist.
Public Administration
- Managerial determination.
- Council resolution.
- Assembly minutes.
- Preliminary report.
Quick actions
During the generation process, the interface presents quick action buttons to speed up the flow:
- Confirm data: accept all extracted data and proceed.
- Correct: open correction mode for a specific datum.
- Cancel: stop generation and return to normal chat.
- Download: download the final DOCX document.
- Choose sections: select only the sections to include in the final document.
Tips for optimal results
Use structured files when possible: an Excel sheet with balance data produces much more accurate results than a narrative PDF. Structured extraction is nearly error-free.
Provide clear context: when asking to generate a document, specify exact type and context. "Generate the explanatory notes for the 2025 balance of company Alfa Srl" is much better than "Make explanatory notes".
Upload all needed documents before starting: make sure all source files are present (uploaded in conversation or available in KB) before starting generation.
Check medium and low confidence data: the confidence indicator is your best ally. Pay particular attention to non-high confidence values.
Leverage interactive correction: if a datum is wrong, no need to start over. Ask for correction in natural language and the system updates only that value.
Always review the final document: as accurate as it is, AI generation is a support tool. Human review of the final document remains essential, especially for documents with legal or accounting value.
Save customized templates: if you frequently generate the same document type with specific customizations, ask the administrator to create a dedicated template.
Queria v3.5.0 -- Document Intelligence platform