In two months, we submitted 20 public sector and nonprofit proposals, with AI involved at almost every step. It sourced and qualified tenders, analysed long RFP packs, drafted responses, generated clarification questions, and red-teamed our own drafts to list their gaps. One of us built the tooling, using reusable Claude-based skills so the whole team works the same way. The other ran it on live bids every week.
AI made us dramatically faster. It also taught us lessons we didn’t expect.
Speed made it tempting to bid on everything, and volume without selectivity wasted effort. AI missed addenda that changed what we should have proposed. It confidently stated a “fact” about our own partnerships that was wrong, and it skipped a mandatory section entirely. We put live demo links in our PDFs and had no evidence that evaluators ever opened them. At the same time, some uses were clear wins. Red-teaming a draft against the RFP sharpened every submission. Shared skills gave the team consistent first drafts instead of everyone writing their own prompts. And AI-drafted clarification questions surfaced information no RFP stated, including one buyer’s actual grant budget.
In 20 minutes, we’ll share an honest field report: what we’d keep, what we’d change, and where people must stay accountable.
Key takeaways
- Speed raises the bar for qualification. The faster you can write, the stricter your go/no-go needs to be.
- Make AI critique before it creates. Red-teaming drafts against the RFP added more value than faster drafting did.
- Every claim needs a human owner. Facts, screenshots, pricing, and addenda all need someone who verified them.
- Use AI to ask, not just to answer. Better clarification questions often matter more than better prose.
Why this session matters to the community
Every agency is now experimenting with AI in sales, mostly in isolation, and buyers are receiving more polished proposals than ever. This session offers an honest, practical account from both the builder’s and the daily user’s side, so other teams can skip our mistakes.