While the AI tools of 2026 have made it easier than ever to discover new business opportunities, the space is also the most crowded it’s ever been - making it more difficult to sift through the piles of garbage “INFO ONLY” documents to find actual fits for your organization. This guide will help you build a perfectly tailored prompt for your organization and preferred AI assistant to make sure you can delight your revenue team with that next ideal opportunity.
Step 1. The Base.
The first step of building the ideal prompt is to give the tool some information about your company and what you can actually offer. This portion should go at the very beginning of your prompt to give the model an initial idea of what to look for.
The more information you give here, the better. A quick tip is that you can link your company’s website and tell the model to do its own research on what you do. This will allow it to get a comprehensive overview of your company. *Note: If using this tip, ensure that your assistant has access to the internet. For ChatGPT, enable “Web Search”. For Claude, enable the “Web Search Tool”.
Step 2. Define What Not to Show You.
This is where most prompts fall apart.
If you don’t explicitly tell the model what to ignore, it will happily surface everything, including vendor announcements, expired notices, and the dreaded “INFO ONLY / NO RESPONSE REQUIRED” documents.
You should be very clear about exclusions upfront.
Examples of exclusions to call out:
- “Information only” or “sources sought” notices
- Expired or past-due RFPs
- Market research surveys with no follow-on procurement
- Partner-only or reseller-only opportunities (if not applicable)
- Opportunities outside your minimum contract value
- Industries or verticals you do not serve
Example exclusion language:
Step 3. Add Fit Constraints
Now that the model knows who you are and what to ignore, you can guide it toward ideal-fit opportunities.
Think like a RevOps leader here. What makes an opportunity worth pursuing?
Common fit constraints include:
- Geography (countries, states, municipalities)
- Buyer type (government, education, healthcare, private enterprise)
- Estimated deal size or budget language
- Keywords that indicate urgency or intent (“replacement,” “vendor,” “implementation,” “rollout”)
Step 4. Tell the Model How to Think
Modern AI assistants perform significantly better when you assign them a role.
You’re not asking it to “search the internet.” You’re asking it to behave like an experienced proposal or capture manager.
This seems to elevate the level of results that the assistant returns, encouraging some critical thinking from it.
Step 5. Specify the Output
If you don’t control the output, it’ll spit out walls of text that is a nightmare to read.
Always request a structured format.
Recommended fields:
- Buyer / agency name
- Opportunity summary (plain English)
- Why this is a fit
- Submission deadline
- Source link (Critical to avoid hallucinations)
Step 6. The Full Example Prompt
Here’s what it looks like when everything comes together:
Final Tip: Start Broad, Then Tighten
Your first run doesn’t need to be perfect.
Run the prompt, review the results, then:
- Add exclusions for junk you still see
- Tighten industry or keyword filters
- Adjust how strict the “fit” criteria should be
After 2–3 iterations, you’ll have a reusable, high-signal prompt your revenue team will actually trust.
Bonus Tip: Use RFP Discovery Tools
While general AI tools can be a great starting point, it’s often still a struggle to identify great opportunities without sinking hours into the prompt refinement process. This is where tailor-made RFP discovery software comes in handy: tools like RFP Hunter or Discovery Pro can speed the process up considerably, with huge potential payoffs. If ChatGPT isn’t giving you what you’re looking for, these options are good next steps to turn to.