AI features
Lydaro has three AI-powered capabilities:
- Lyra — an in-app support assistant available on all plans. See Lyra Support Assistant.
- AI risk tools — risk generation and improvement for Pro and Team plans (described below).
- Document Extraction — extract risks, issues, and assumptions from uploaded documents for Pro and Team plans (described below).
All are powered by Anthropic's Claude.
AI Risk Generation
Generate risk suggestions based on your project description. Available on Pro and Team plans.
- In a project, click Generate with AI
- Describe your project (what it involves, key challenges)
- Review the suggested risks
- Accept, edit, or discard each suggestion
AI suggests risks with pre-filled cause, event, and effect fields. You own the content — AI suggestions are a starting point, not a final answer.
AI Risk Improvement
Get AI feedback on a specific risk's description. Available on Pro and Team plans.
- Open a risk
- Use the AI improvement option on the risk detail page
- Review suggestions for clearer cause, event, and effect wording
Document Extraction
Extract risks, issues, and assumptions from uploaded project documents. Available on Pro and Team plans.
The extraction workflow has four steps:
1. Upload documents
Upload PDFs, Word documents, or text files to your project's Documents section. Lydaro extracts the text from each file automatically. The raw file is never sent to the AI — only the extracted text, which keeps token costs low.
2. Run extraction
Go to Documents → Extract and select the documents you want to analyse. Choose what you're extracting — Risks, Issues, or Assumptions — then start the run. You can select multiple target types in a single run.
You can also designate documents as approved context (e.g. a project charter or brief). Context documents are not scanned for candidates; instead, their content is given to the AI as official project facts to help it identify relevant risks and issues in your source documents.
3. Review candidates
The AI returns a list of suggested items (candidates), each with a title, rationale, and a direct quote from the source document. Go to Documents → Candidates to review them.
For each candidate you can:
- Accept — creates the item in the risk register, issue log, or assumption log with all AI-suggested fields pre-populated
- Edit & Accept — adjust the title or description before accepting
- Reclassify — change the type (e.g. from Risk to Issue) before accepting
- Reject — decline with an optional reason
- Mark as duplicate — flag if the item already exists
4. Review the created items
Accepted candidates create fully-formed records with cause/event/effect for risks, priority and type for issues, and category for assumptions — all sourced from the AI output. Each record carries a source field showing which document it came from (e.g. Business Case extracted using AI from Project Charter.pdf).
You own the content. The AI's suggestions are a starting point for review, not a final answer.
AI token limits
AI risk tool usage is measured in tokens (input + output combined). Limits by tier:
| Tier | AI tokens/month |
|---|---|
| Free | Not available |
| Pro | 300,000 tokens |
| Team | 700,000 tokens |
| Enterprise | Custom |
Token usage is based on Anthropic's Claude Haiku model. Monitor your current usage on the Billing page.
Privacy
Risk data sent to AI is used only to generate responses for that session. Data is not stored or used for AI training. See our Privacy Policy.