Vendor Selection & Procurement

Many people think AI in procurement is just "help me write an RFP." That's only the shallowest layer. The real value is having AI align vendor information, surface risks early, and standardize the review process.

But this is also where things go wrong easily — if a summary is wrong, a clause gets missed, or a cost judgment is off, the decision-maker gets led astray by bad info.

---

What AI Does Best in Procurement

Bottom line: AI is good at organizing, comparing, questioning, and initial screening. It's not good at making final legal or finance decisions for you.

---

Step 1: Write the RFP Clearly — Otherwise Comparison Is Pointless

Many vendor reviews fail because requirements were too vague from the start. AI is great at turning requirements into a structured RFP.

A Workable RFP Prompt

You are a procurement assistant. Generate a vendor RFP draft.

Background:
- use case: [e.g., AI customer support platform]
- team size: [e.g., 20 agents]
- budget range: [e.g., AUD 30k - 50k / year]
- timeline: [e.g., 6 weeks]

Must-haves:
- SSO
- audit log
- role-based access
- API access

Output:
- requirement list
- evaluation criteria
- submission format
- clarification questions

The biggest value of this output: all vendors answer within the same framework instead of talking past each other.

---

Step 2: Use a Comparison Matrix to Align Information

AI is great at turning multiple vendor proposals into one table. Compare at least these fields:

Recommended output format

Vendor | Pricing | Core features | Gaps | Risks | Questions

Simple format, but excellent for first-round procurement screening.

---

Step 3: Have AI Generate Clarification Questions

Many teams reviewing vendors lose out because they don't ask the right questions. AI is good at auto-generating questions from proposals:

• Where is data stored (which region)?
• What's the default retention policy?
• What's the API rate limit?
• How long after contract termination can you get data back?
• Is there SLA breach credit or compensation?

These questions are worth more than "what other features do you have."

---

Step 4: Clause Review Can Only Be a Checklist — Not a Legal Substitute

This is the most important boundary.

AI can help with:

• Flagging clauses that need close review
• Summarizing termination, renewal, liability, and data processing terms
• Generating internal review notes

AI should NOT:

• Redline contracts directly
• Judge whether clauses are legally valid
• Output a final legal opinion on behalf of the company

A more reliable workflow:

contract -> AI extracts key clauses
         -> internal summary
         -> legal / finance review
         -> final negotiation notes

---

In Procurement, Only Looking at Price Is the Biggest Trap

Cheap doesn't mean cost-effective. Many vendors look affordable upfront but make it up later through:

• Implementation fees
• Seat expansion costs
• Premium support charges
• API overage fees
• Migration / export costs

Break total cost into buckets rather than just reading the first-page quote.

---

A Practical Vendor Review Flow

1. AI generates RFP draft
2. Collect vendor replies
3. AI creates comparison matrix
4. AI generates clarification questions
5. Key clauses go to legal / finance / security review
6. Humans make the final decision

The point of this flow isn't "let AI decide." It's making sure the information is complete and organized before the decision.

---

Common Mistakes

---

Practice

Pick 2 vendors you're currently comparing. Organize this input:

• use case
• budget range
• must-have requirements
• 2 proposal summaries

Then have AI output:

1. Comparison matrix
2. Clarification questions
3. Risk checklist
4. Decision memo draft

Then have procurement, security, finance, or legal do the final review.