SaaSReadyit generates validation reports with large language models routed through the Lovable AI Gateway. The Expert tier states a confidence level for each major claim.
Which model writes which report? openai/gpt-5 writes the Starter and Expert reports. openai/gpt-5-mini writes the Professional report. google/gemini-2.5-flash screens the idea, streams the homepage suggestion and answers report chat. google/gemini-2.5-flash-lite classifies the input, and google/gemini-3-flash-preview runs site search and the simplified report.
Where do the numbers come from? From patterns in the models' training data, not from a database or a live lookup. No web search runs — the code supports one, but no search provider is configured — so nothing is looked up, and no figure is tied to a source you could check.
What happens when generation fails? The generator retries on another model. If every attempt fails, a generic structured template is returned that is not written from your idea, the request is queued for manual fulfilment, and both the user and the team are emailed.
Is anything fact-checked? No. The Professional generator re-checks its own sections and rewrites any that came back empty; nothing verifies a claim against the world, and the Starter and Expert generators do not run even that check.
Where does the idea go? The idea text alone is sent to OpenAI and Google to write the report — no name, email address or account identifier travels with it. Under the privacy policy it is not sold and the providers do not train on it.
Machine-readable fact sheets: /llms.txt and /llms-full.txt.