Gunly is built on Claude, Anthropic's AI model, combined with agents designed specifically for public procurement. Here is how it works, and where its limits are.
Tender files, annexes and Q&A notes are converted into clean, structured text, page numbers preserved.
Claude reads the whole file at once and extracts requirements, criteria and the evaluation method.
Relevant passages from your own past bids and references are found in your private knowledge base.
Answers are written per criterion, grounded in your sources, with references for every claim.
A separate AI assessor scores the draft against the tender's own method and explains the gaps.
Hundreds of pages in a single pass, so a requirement in an annex is never read out of context.
Strong at following complex rules and explaining why an answer does or does not meet a requirement.
Produces precise, professional prose in Dutch, English and German.
Your data is processed under terms that do not allow it to be used for model training.
Every extracted requirement and every claim in a draft links to its source page or reference.
The AI drafts and advises. Your team reviews, decides and submits. Gunly never submits anything.
Your documents form a private knowledge base for your workspace only, stored in the EU.
AI can make mistakes or miss nuance. That is why sources are shown and every answer needs your review.
It writes strong first drafts per criterion. The final bid is always written, checked and owned by your team.
Yes. Like any AI, it can misread or miss nuance. Every output comes with sources so you can check it quickly.
No. Your documents are used only inside your own workspace and never to train AI models.
Dutch and English today. German follows with our expansion to Germany.
Upload one tender during onboarding and see your compliance matrix in minutes.