AI-Powered Procurement: Sample Prompts You Can Use Today

You know AI can help with procurement. You've seen the headlines, maybe tried a tool once or twice. But you're at your desk staring at a blank chat window, and the question is: what do I actually type?

This post gives you ready-to-use prompts organized around four stages of the procurement lifecycle: researching the market, sourcing vendors, checking for existing contracts, and writing your scope of work. Each one follows a prompting framework called CRAFT:

• Context – What's the situation?
• Role – Who should the AI act as?
• Actions – What do you want it to do?
• Format – How should the output be structured?
• Target –Who is the audience?

Copy these prompts, paste them into your AI generator, swap in any relevant personalized details, and you will  start getting useful output from your tool in minutes.

Before You Start: A Note on Data Security

If you're using a free, consumer AI tool (the free tier of ChatGPT, the free version of Google Gemini, etc.), always assume that anything you type could be used to train the model. That means your data is no longer just yours.

If your agency doesn't have a formal AI use policy yet, talk to your IT team. More organizations are putting these in place, and knowing what's allowed (and what isn't) before you start will save you headaches later.

AI jumped to the number one priority for state technology offices in NASCIO's latest survey, beating out cybersecurity for the first time. Most states are already running pilot projects and training staff on AI tools. But the policies haven't kept up. Only about half of states have even created safe testing environments for AI, and dedicated funding is rare. The tools are moving faster than the rules. If your agency hasn't figured out its AI guardrails yet, you've got plenty of company, but the window for getting ahead of it is closing.

Here are a few notes of practical guidance:

• Don't paste in confidential documents, internal pricing, vendor proposals, evaluation scores, or anything you wouldn't put on a public website.
• General descriptions of your needs are fine. "We're a county government in Texas looking to replace our fleet management software" is safe. Pasting in your entire internal needs assessment with budget numbers and stakeholder names is not.
• If your organization has access to an enterprise AI tool with a data processing agreement in place, you have more room. But when in doubt, keep it general.

The prompts below are written with this in mind. They use general descriptions rather than sensitive specifics. Where a prompt would benefit from more detail that could be identifying or confidential, you'll see a ⚠️ callout.

Research: Understanding the Market

Procurement professionals get asked to buy everything from janitorial supplies to enterprise software. You can't be an expert in every category, and nobody expects you to be. AI can get you up to speed fast: who the major players are, what questions to ask, and what to look for before you write a solicitation. You'll still need your subject matter experts, but these prompts help you make better use of their time by doing your homework first.

Prompt 1: General Market Landscape

Note: This prompt works best with web search enabled to ensure current, cited information.

Context: I'm a procurement professional at a [TYPE OF AGENCY, e.g., "mid-size county government"] in [STATE]. We're beginning to explore options for [GENERAL DESCRIPTION OF NEED, e.g., "a new permitting and land management software system"]. We have not written a solicitation yet and are in the research phase.

Role: Act as a public sector technology research analyst.

Actions:
1. Identify the main categories of solutions available for this need.
2. List 8-10 vendors that are commonly used by government agencies for this type of solution, and briefly describe what differentiates each one.
3. Summarize the key features and capabilities I should expect from a modern solution in this space.
4. Identify 3-5 questions I should ask vendors during market research calls.
5. Note any industry trends or recent developments that could affect our decision.
6. Include a "Sources" section at the end listing any references used.

Format: Organize the output with clear headings for each section. Use a table for the vendor comparison. Where specific claims are made about vendors or market data, cite your sources inline with the URL or publication name. Keep the full response under 1,500 words.

Target: This is for internal use by procurement and project stakeholders who are not deeply technical but need to get smart on this market quickly.

Prompt 2: Pricing and Cost Structure Research

Accuracy note: AI pricing data is often outdated, incomplete, or based on a mix of public and private sector numbers that may not reflect what your agency would like to pay. Use this as a starting point to understand general pricing models and cost structures, but always verify numbers directly with vendors through recent contract data before making budget decisions.

Note: This prompt works best with web search enabled to ensure current, cited information.

Context: I'm researching [TYPE OF SOLUTION, e.g., "enterprise asset management (EAM) software"] for a [TYPE OF AGENCY] in [STATE]. I need to understand typical pricing models and cost structures so I can set realistic budget expectations before going to solicitation.

Role: Act as a government procurement advisor with experience in software acquisitions.

Actions:
1. Describe the most common pricing models for this type of solution (e.g., per-user, per-asset, flat license, tiered).
2. Identify the typical cost components beyond the base license (implementation, training, data migration, integrations, ongoing support).
3. Provide general cost ranges where publicly available. If exact numbers aren't available, describe what factors most influence the price.
4. List 3-5 cost-related questions I should include in an RFI.
5. Include a "Sources" section at the end listing any references used.

Format: Use a structured outline with brief explanations. Include a summary table of pricing models with pros and cons for each. Where specific claims are made about pricing or cost data, cite your sources inline with the URL or publication name.

Target: Internal budget planning team and project sponsors who need a ballpark before we go to solicitation.

Source: Finding the Right Vendors

Most procurement teams have a handful of vendors they already know for any given category. That's a fine starting point, but it can lead to the same three companies responding to every solicitation. AI is useful here for broadening your search, especially when you're buying something your agency hasn't purchased before or you're working in an unfamiliar market.

This works for projects of all sizes. For a large formal solicitation, these prompts help you build a longer, more diverse vendor list. For smaller purchases that just need a few quotes, they can help you quickly identify firms to call so you're not just Googling and hoping for the best.

Prompt 3: Finding Qualified Vendors

Note: This prompt works best with web search enabled to ensure current, cited information.

Context: I'm a procurement officer at a [TYPE OF AGENCY, e.g., "state agency"] in [STATE]. I need to identify potential vendors for [GENERAL DESCRIPTION OF NEED, e.g., "body-worn camera systems and associated cloud storage and evidence management"]. I want to go beyond the vendors I already know and find firms that are actively serving the public sector.

Role: Act as a procurement sourcing specialist with expertise in government contracts.

Actions:
1. List 10-15 vendors that provide this type of solution to government agencies. Include a mix of large established firms and smaller or regional providers.
2. For each vendor, note: what they're known for, company size (Large / Mid / Small), whether they appear to have existing government contracts, and whether they appear to be local or regional to [STATE] — flag these with "(Local/Regional)" if so, based on your best available information.
3. Organize vendors into three groups: Large, Mid-size, and Small/Regional.
4. Suggest 3-5 sources I can check to verify vendor qualifications (e.g., contract registries, GSA schedules, state contract portals, industry associations).
5. Recommend any relevant industry conferences, associations, or directories where I might discover additional vendors.
6. Include a "Sources" section at the end listing any references used.

Format: Use a numbered list with brief descriptions for each vendor (2-3 sentences max). Where specific claims are made about vendors or their government contracts, cite your sources inline with the URL or publication name. Follow with a separate section for verification sources and discovery channels.

Target: Procurement team building a solicitation distribution list.

Prompt 4: Vendor Background Check Prep

⚠️ Data Security Note: This prompt asks about a specific vendor by name. Vendor names are generally fine to use in AI tools since they're public entities. However, do not paste any internal evaluation notes, pricing proposals, or confidential correspondence you've received from a vendor.

Note: This prompt works best with web search enabled to ensure current, cited information.

Context: I'm evaluating whether to include [VENDOR NAME] in an upcoming solicitation for [GENERAL DESCRIPTION OF NEED] at a [TYPE OF AGENCY] in [STATE]. I want to do some preliminary research before reaching out to them.

Role: Act as a procurement due diligence analyst for public sector acquisitions.

Actions:
1. Summarize what this vendor does and who their typical government clients are.
2. Identify any publicly available government contracts this vendor currently holds (e.g., state contracts, GSA schedule, cooperative contracts).
3. Note any publicly reported issues: lawsuits, contract terminations, protests, or performance complaints involving government clients.
4. List 3-5 questions I should ask this vendor on a pre-solicitation call.
5. Include a "Sources" section at the end listing any references used.

Format: Use clear section headings. Where specific claims are made about the vendor's contracts, clients, or reported issues, cite your sources inline with the URL or publication name. Flag any information you're uncertain about so I know what to verify independently.

Target: Procurement lead preparing for vendor outreach.

Check: Existing Cooperative Contracts

Before you invest weeks writing a solicitation, it's worth checking whether someone else has already done the work. AI is useful here because it can draw connections between what you're buying and how contracts are categorized across different platforms, catching relevant results that a simple keyword search might miss.

Prompt 5: Finding Existing Cooperative Contracts

Note: This prompt works best with web search enabled to ensure current, cited information.

Context: I'm a procurement professional at a [TYPE OF AGENCY, e.g., "county government"] in [STATE]. I need to purchase [WHAT YOU'RE TRYING TO BUY -- describe in plain language, e.g., "janitorial services for several county buildings"]. I'm trying to find existing cooperative contracts I could use for this purchase before deciding whether to run my own solicitation.

My agency already has memberships or established relationships with the following cooperatives (leave blank if none): [LIST ANY PREFERRED OR EXISTING COOPERATIVE MEMBERSHIPS, e.g., "Sourcewell, HGACBuy" or "we are a BuyBoard member"]

Role: Act as a cooperative purchasing research specialist.

Actions:
1. If the agency has listed preferred cooperatives above, search those first. Then search any other major national or regional cooperative purchasing organizations commonly used by [TYPE OF AGENCY] agencies in [STATE].
2. For each cooperative, identify any currently active contracts that cover this need. For each contract found, provide:
- Cooperative name
- Contract number
- Contract title or category
- Awarded vendor(s)
- Contract expiration date (including any exercised renewal options)
- A direct link to the specific contract page. If a direct contract URL is not publicly available (e.g., the cooperative requires a login to view contract details), link to the closest publicly accessible page for that contract and note that login may be required to view full details.
3. If a cooperative has no active contract covering this need, say so briefly and move on. Do not provide general information about how the cooperative works.
4. After listing all contracts found, provide a one-paragraph summary of which contracts appear most relevant and why, and flag any where the scope may not be a clean fit for this need (e.g., supplies-only contracts when services are needed). Include a recommended next step.

Format: Present findings as a table per cooperative, with columns for Contract Number, Title/Category, Vendor(s), Expiration, and Link. If no relevant contract exists for a cooperative, note it in one sentence beneath the cooperative name. End with the summary paragraph.

Target: Procurement staff making a buy vs. bid decision who need to know what's available right now, not how cooperative purchasing works in general.

Write: Building Your Scope of Work

Scopes of work eat up more writing time than almost anything else in procurement. A vague or incomplete scope leads to bad proposals, change orders, and protests. The problem is that writing a good one from scratch requires you to think through dozens of details before you've typed a word.

What AI does well here is give you a solid first draft to react to. It won't write your final scope, but it gets you past the blank page. Most people find it much easier to edit and improve something than to stare at a blank page. Start with an AI draft, then layer in your agency-specific requirements, institutional knowledge, and lessons learned from past contracts.

A couple of pro tips worth trying:

• First, if you have a past solicitation with a scope you liked, upload it and ask the AI to analyze the structure, level of detail, format, and tone. Then use that as a template when drafting your next scope.  That way you're not reinventing the format every time you start a new solicitation.

• Second, if you find solicitations from other agencies for similar projects, upload those too. The AI can pull in relevant requirements and language you might not have thought of. And if you're doing this in the same conversation where you did your market research, all of that context carries forward and directly improves the output.

Prompt 6: Analyzing a Past Scope as a Template

⚠️ Data Security Note: This prompt involves uploading a past solicitation document. If the solicitation has already been publicly posted, the risk is lower, but it still may contain internal formatting, contact details, or agency-specific language you'd prefer to keep private. Use an enterprise AI tool with a data processing agreement when possible.

Context: I'm a procurement professional at a [TYPE OF AGENCY] in [STATE]. I'm uploading a solicitation document that contains a scope of work (or similar section -- it may be labeled "Scope of Services," "Statement of Work," "Scope of Work," or something similar). I want to extract the exact structural pattern and tone of that scope section so I can use it as a formatting and voice template for future AI-generated scopes.

Role: Act as a document structure and tone analyst.

Actions:
1. Locate the scope section in the uploaded document. It may be labeled differently -- look for the section that defines what the contractor is required to do. Ignore all other sections (background, deliverables, performance standards, contract administration, etc.) unless they are nested within the scope section itself.
2. Document the exact structure of the scope section as found in this document. Be specific:
* What is the top-level section number and title format?
* How are subsections numbered (e.g., 2.1, 2.2) and titled?
* How many levels of hierarchy exist (section, subsection, sub-subsection)?
* How are requirements written at each level -- as prose paragraphs, bulleted lists, numbered lists, or a mix?
* How many bullets typically appear under each subsection?
* What mandatory language pattern is used (e.g., "The Contractor shall," "Vendor must," "Services include")?
* Does each subsection open with a prose paragraph before the bullets? If so, what does that paragraph typically contain?
* Are there any tables, and if so, where do they appear relative to the bullets and prose?
3. Identify any repeating structural patterns -- subsections that follow the exact same format as each other. Note the pattern and how many times it repeats.
4. Analyze the tone and voice of the scope section specifically. Extract 4-6 concrete "write it this way / not that way" examples pulled directly from the document. Each example should show the actual phrasing style used, contrast it with a weaker or different phrasing that was not used, and explain in one sentence what the difference signals to a vendor. Cover at least: specificity of obligations, treatment of timeframes, how qualifications or exceptions are handled, and level of formality.
5. Flag any structural inconsistencies within the scope section -- places where the format breaks from the dominant pattern.
6. Output two artifacts:
a. A precise structural template that someone could hand to an AI model to replicate this exact format in a new scope. Include placeholder labels (e.g., [SERVICE CATEGORY], [GOVERNING FRAMEWORK]) so the template is immediately usable. The template should be detailed enough that an AI-generated scope using it would be format-identical to the original -- same numbering scheme, same paragraph-before-bullets pattern, same bullet depth, same mandatory language.
b. A tone guide with 4-6 "write it this way / not that way" examples, each drawn directly from the uploaded document. Label this section clearly so it can be pasted alongside the structural template into a scope generation prompt.

Format: Lead with a brief structural summary (5-6 sentences max). Follow with the two labeled artifacts. Do not provide general scope writing advice -- every observation must be tied to something specific in this document.

Target: Procurement professional who will feed both output artifacts directly into a second AI prompt to generate a new scope in the same format and voice.

Prompt 7: Drafting a Scope of Work

⚠️ Data Security Note: To get the best output here, you'll want to include specifics about your project: quantities, locations, timelines, and requirements. If you're using a free AI tool, keep these general. Instead of "450 laptops for Jefferson County School District across 12 campuses," try "approximately 400-500 laptops for a K-12 school district across multiple campuses." Save the precise details for your internal drafting process.

This prompt is designed to work with the output from Prompt 6. Paste the structural template and tone guide you generated there into the placeholders at the bottom of this prompt. If you haven't run Prompt 6 yet, you can still use this prompt, but the output will follow a generic format instead of matching your agency's existing solicitation style.

Context: I need to draft a scope of work for [GENERAL DESCRIPTION OF PROJECT, e.g., "janitorial and custodial services for multiple government office buildings"]. The contract will be for a [TYPE OF AGENCY, e.g., "city government"] in [STATE]. The anticipated contract term is [DURATION, e.g., "3 years with two 1-year renewal options"]. Key details:
* [NUMBER/SCALE, e.g., "Approximately 8-10 buildings ranging from 5,000 to 50,000 square feet"]
* [KEY REQUIREMENTS, e.g., "Daily cleaning of common areas, restrooms, and break rooms; weekly floor maintenance; quarterly deep cleaning"]
* [ANY SPECIAL CONSIDERATIONS, e.g., "Some facilities require after-hours service only; green cleaning products preferred"]

Role: Act as a senior procurement specialist experienced in writing scopes of work for public sector service contracts.

Actions:
1. Draft a complete scope of services section only -- not a full solicitation document. Use the structural template and tone guide provided below as your exact formatting and voice specification. Do not deviate from the structure, numbering scheme, bullet depth, mandatory language patterns, or tone examples in the guide.
2. Include placeholder text in brackets for anything that would require internal data I haven't provided (e.g., [INSERT BUILDING ADDRESSES], [INSERT SQUARE FOOTAGE SCHEDULE]).
3. After the scope, list 5-10 items I should verify or add based on my agency's specific policies (e.g., insurance requirements, prevailing wage applicability, security clearance needs). Format this as a separate section labeled "Agency Verification Checklist" -- do not mix it into the scope itself.

Format: Follow the structural template and tone guide exactly as provided below. Do not add sections, change numbering schemes, introduce nested bullets, or use formatting not present in the template.

[PASTE STRUCTURAL TEMPLATE HERE]

[PASTE TONE GUIDE HERE]

Target: This scope section will be dropped directly into a formal solicitation (RFP/IFB) reviewed by vendors, evaluation committees, and legal/contracts staff. It must be format-identical to the agency's existing solicitation templates.

Prompt 8: Improving an Existing Scope of Work

⚠️ Data S If you paste in an existing scope of work, make sure it doesn't contain confidential budget information, in-progress evaluation notes or scores, or anything your agency considers non-public. If you're unsure, describe the scope in general terms and ask the AI to draft a new version rather than pasting the original.

You don't have to copy and paste your scope into the prompt. Most AI tools let you upload a file directly. If you'd rather do that, just change the Context line to something like: "I'm uploading a draft solicitation that contains a scope of work. Please review only the scope section."

Context: I have a draft scope of work for [GENERAL DESCRIPTION OF PROJECT] for a [TYPE OF AGENCY] in [STATE]. I want to strengthen it before including it in a formal solicitation. My draft scope is pasted below.

Role: Act as a procurement quality reviewer who specializes in public sector solicitation documents.

Actions:
1. Review the scope for completeness: are there common sections or requirements that appear to be missing?
2. Identify any language that is vague, ambiguous, or could lead to vendor confusion or protests.
3. Suggest specific improvements to make the requirements more measurable and enforceable.
4. Flag any areas where the scope might unintentionally restrict competition or favor a particular vendor.
5. Recommend any standard clauses or requirements I may want to add based on common public sector best practices.
6. Identify any compliance gaps specific to a [TYPE OF AGENCY] in [STATE] -- including applicable state statutes, local government code requirements, agency-type-specific regulations, and any federal requirements that apply based on the subject matter of the scope. Flag these separately from general best practice recommendations, as they represent potential legal exposure rather than quality improvements.

Format: Organize findings in the order they appear in the document, so the output reads as a sequential to-do list. For each finding, use the following structure:

[Section Reference] -- [Short Issue Title]
Severity: [Critical / High / Medium / Low]
* Issue: [What the problem is, referencing exact language from the draft where relevant]
* Impact: [What could go wrong if this isn't addressed]
* Suggested Revision: [Specific recommended change or addition]

Use a horizontal rule between findings to visually separate them. At the top, include a one-paragraph summary of overall scope quality and a count of findings by severity level. Compliance gap findings (from Action 6) should be clearly labeled with a [COMPLIANCE] tag next to the severity rating so they are visually distinct from quality findings.

Target: Procurement lead finalizing a solicitation for internal review and eventual public posting.

[PASTE DRAFT SCOPE HERE]

Tips for Getting Better Results

1. Iterate, don't expect perfection on the first try. Your first prompt will get you 70% of the way there. Follow up with specific feedback: "Make the vendor list more focused on firms with K-12 experience" or "The performance standards section is too vague, add specific measurable metrics." Treat it like working with a new team member. Direct feedback gets better output.

2. Build your own prompt library. When you get a prompt working well for a specific task, save it. Keep a shared doc or folder where your team collects prompts that have produced good results. Over time, you'll stop starting from scratch and start from something that already works. The prompts in this post are a starting point. The ones you refine for your agency's specific needs, terminology, and formatting preferences will be more useful than anything we can write for a general audience.

3. Use AI to write your prompts. You don't have to nail the CRAFT structure from memory every time. If you know what you need but aren't sure how to phrase it, just describe the task in plain language and ask the AI to turn it into a well-structured prompt. Something like "I need to find vendors for a new phone system for a city government, can you write me a detailed prompt I can use to research this?" works fine. Let the tool help you use the tool.

4. Tell the AI what's wrong, not just what you want. If a response is too generic, say so: "This reads like it could apply to any agency. Make it more specific to a mid-size city government that operates its own utilities." You can also use multipliers as a shorthand: "Make this 5x more detailed" or "Make this 50x more detailed." These aren't literal measurements, but AI tools respond well to them as directional input. The more specific your feedback, the more useful the next draft.

5. Don't trust, verify. AI will confidently present information that isn't accurate. Vendor lists might include companies that don't exist, pricing ranges might be outdated, and legal guidance could be flat wrong. Use AI output as a starting point for your own research, not as a finished product. Every fact that matters should be verified against a real source.

6. Keep your data safe. We said it at the top and we'll say it again: if you wouldn't put it on a public website, don't paste it into a free AI tool. General descriptions and publicly available information are fair game. Internal documents, pricing, and personnel details are not.
   

Your 30-Day Action Plan

• This week: Pick one prompt from this post and try it. Just one. See what you get. Follow up with one round of feedback to improve the output.
• By end of week two: Try a prompt from a different stage of the lifecycle. Start building a feel for where AI saves you the most time in your specific workflow.
• By end of month one: Share what you've learned with a colleague. Show them a before-and-after: your prompt, the AI's first draft, and the final version you actually used. That's a fast way to get buy-in across your team.

AI Inside the Workflow

The prompts in this post work in any AI tool you have access to today. But standalone AI is just the starting point. The real efficiency gains come when AI is woven directly into the tools you're already using to manage the full procurement lifecycle.

That's what we're building at PlanetBids. We've embedded AI across several parts of the purchasing lifecycle, including:

• Scope of work drafting: AI-assisted scope writing that helps you build detailed, structured scopes faster.
• Q&A analysis and organization: Automatically analyzing and categorizing vendor questions so you can respond more efficiently.
• Certificate of Insurance processing: Importing, reading, and editing COIs so your team spends less time on manual data entry.
• Evaluation setup: Helping configure evaluation criteria, committee structure, and key dates based on your solicitation requirements.
• Evaluation reporting and summaries: Generating clear summaries of evaluation results across your committee.
• Unbiased AI evaluator: An AI-powered evaluator that scores proposals against your stated criteria, giving your committee a consistent, independent baseline to compare against.

We're building AI into the procurement workflow every day, and we'd love to hear how you're putting these tools to work. Reach out to us at PlanetBids to keep the conversation going.