Turn Days of Proposal Research into Minutes: The 2025 Playbook

What is the difference between a good proposal and a bad proposals? Ok, a lot of things, but my top pick is: hyper-personalization.

For this week’s edition, I spoke with Sean Williams, CEO of AutogenAI about why most proposal AI tech just helps you lose faster. Instead of providing our customers with a proposal that looks almost identical to everyone else’s, our buyers reward proposals that feel tailored to them.

The difference is you can deliver that level of personalization fairly quickly, with citations, and without burning out your team.

In this issue

• A 15-minute customer research setup you can run before 9:00 (it’s so powerful!)

• Stargazer questions on AI Go/No-Go and “faster losing”

• Podcast: AutogenAI on writing that actually scores

• Proposal tech spotlights to help you shortlist smarter

Proposal Research in Minutes

Buyers don’t reward you for effort, unfortunately; they reward you for understanding their world. And yet, they’re flooded with generic, AI-spammed proposals. That’s your opening.

The only way you win is by showing proof you get them, faster than anyone else.

✹ Personalization isn’t optional. McKinsey and HBR link it directly to revenue and win rates. But personalization at human speed is a losing game and impossible with our shortening submission deadlines.

✹ Research agents changed the math. A cited dossier in 15 minutes isn’t that cool anymore? It’s just expected.

✹ Trust is the differentiator. AI lies. If you give yourself a two-minute safety pass, then you can easily turn a risky draft into a credible one. Skip it, and you could 100% embarrass yourself (and lose).

What the AI Is Really Doing

When you use Deep Research to do your customer proposal research, it’s doing very specific things:

1. Plan — Breaks your prompt into queries and a schema.

2. Gather — Pulls from primary sources: sites, filings, portals, credible news.

3. Prove or Flag — Writes claims with evidence links, labels weak/conflicting ones as “uncertain.”

4. Assemble — Produces two outputs: a human-friendly brief and a structured file your platform can ingest.

5. Judgment — You verify anything touching price, scope, compliance, or risk.

6. Hand-off — Your proposal tool (if capable) uses retrieval to draft cited answers and compliance tables.

That’s the engine behind the speed but insane, strategic insights. And it’s why this workflow scales without burning your team.

How you can do all of this in 45 minutes (with a bit of practice)

1. Confirm ICP + risks.

2. Run Deep Research on ChatGPT or Perplexity. It will export a brief + structured file.

3. You remove the weak claims and highlight uncertainty.

4. It generates a draft with citations + compliance table.

5. You edit for tone and edge because no prompt gets it totally right (unless you’ve done it, in which case send it to me.)

Why This Matters Now?

If your process can’t deliver cited insights in minutes and show proof on the page, you’re just adding to the pile of generic AI spam OR spending way more time doing something that will produce worse results…which is nonsense.

The shift is already here. The only question is whether you adapt before your competitors do? Of course you will.

↣ ↠ Get the full playbook with prompts : Proposal Research In Minutes: The 2025 AI Playbook For Bids, Tenders, and RFPs

Join the conversation in Stargazy

Community Go/No-Go with AI
Robert Dickson, RevOps at AutoRFP.ai gave us a video walkthrough of a practical prompt flow plus how people are really using it in the wild. If you ever wanted to know how to set it up, this video is perfect.
Use the Go/No-Go prompt →

Debate: AI in Bid/No-Bid. Help or hype?
Chris Murray (Bid Management Consultant) asked: Does AI raise win strategy or just crank volume targets? Add your take.
Jump into the debate →

Poll + badges
If you need to manage resumes or CVs as part of your proposals, then you already get how tough it can be. Vote in the Flowcase archetype poll and get a community badge.
Choose your CV archetype →

Interview with Sean Williams

Sean Williams on the difference between faster output and higher scores:

• The trap of “faster losing.” Not all AI is equal; some tools just churn out 5/10 answers faster. Sean explains what separates text that wins (9+/10) from text that looks polished but scores low.

• The Cicero test. Why propositional density (at least one strong idea per sentence) is the secret to writing proposals that actually score.

• The Boris Johnson problem. How AutogenAI tackled hallucinations with early RAG, source controls, and even lessons from historiography.

• Bake-off or bust. Forget slick demos. Sean shares the 6-vendor bake-off playbook that reveals which tool really works.

• Behind closed doors. What execs actually say when buying AI: it’s not about saving $150k in headcount, it’s about winning $20M more revenue.

Debate thread: “If AI handles compliance and boilerplate, does that free proposal pros to be more strategic, or just pile on more bids?”  ✸ Is AI giving you space to influence win strategy, or is leadership just increasing volume targets because ‘the machine can do it’?

The Stargazy Buyer’s Field Guide for Architecture, Engineering, and Construction (AEC) Folks

Are you in Architecture, Engineering, and Construction (AEC) and looking for the most innovative tech?

We have so many wonderful proposal tech tools supporting stargazy and its members (you!). But one size is not fits all.

Let’s dive in:

Share this issue

Know a proposal pro who needs faster research and better evidence? Forward this email or share the Playbook link.

Questions or ideas for next week? Reply with one line: “Cover this next: ____.”

✨,

— Chris + The Stargazy Team
www.stargazy.io | stargazy

P.S. Follow us on LinkedIn for launches, playbooks, and upcoming AMAs.

P.P.S. New subscribers this week get an early-access community badge.✨,