ai-vendor-evaluation（AI 供应商评估）

AI Vendor Evaluation

Version 1.0 | October 2025 | Based on $1.2M average AI spend analysis

Overview

This skill provides a systematic framework for evaluating AI vendors and solutions to avoid the costly mistakes that plague 95% of AI projects. Use when conducting vendor due diligence, evaluating proposals, negotiating contracts, or making strategic AI purchasing decisions.

Key capabilities:

• Structured evaluation criteria for AI vendors
• Red flag identification in proposals and demos
• Pricing model analysis and fair market rates
• Technical capability assessment
• Contract term evaluation
• Build vs buy decision framework

Quick Decision Tree

Start here to determine which references to read:

What stage are you in?

├─ Early exploration (multiple vendors being considered)
│  └─ Read: evaluation-criteria.md, use-case-fit.md
│     Use: scorecard-template.xlsx
│
├─ Evaluating specific proposal or demo
│  └─ Read: red-flags.md, technical-assessment.md
│     Check: pricing-models.md for pricing reasonableness
│
├─ Contract negotiation
│  └─ Read: contract-checklist.md, pricing-models.md
│     Reference: red-flags.md for problematic terms
│
├─ Build vs Buy decision
│  └─ Read: build-vs-buy.md, use-case-fit.md
│     Consider: Total cost of ownership from pricing-models.md
│
└─ Post-purchase review or audit
   └─ Read: evaluation-criteria.md, technical-assessment.md
      Assess: Whether vendor is delivering on promises

When to Use This Skill

Trigger scenarios:

• "Help me evaluate this AI vendor proposal"
• "What should I look for in AI vendor demos?"
• "Is this pricing reasonable for an AI solution?"
• "Should we build or buy this AI capability?"
• "What questions should I ask this AI vendor?"
• "Help me compare these AI vendors"
• "Review this AI contract for red flags"
• "Conduct due diligence on this AI company"

Core Evaluation Framework

Phase 1: Initial Screening

Goal: Eliminate obviously problematic vendors before deep evaluation

Key questions:

• Does the vendor have relevant domain experience?
• Are there verifiable customer references?
• Is the technology approach sound?
• Are pricing and terms transparent?

Read: references/red-flags.md for disqualifying signals Read: references/use-case-fit.md for domain fit assessment

Phase 2: Deep Evaluation

Goal: Assess vendor capabilities systematically across all dimensions

Evaluation dimensions:

1. Technical capability - Can they actually deliver?
2. Business viability - Will they still exist in 2 years?
3. Pricing fairness - Are costs reasonable for value delivered?
4. Implementation risk - How likely is successful deployment?
5. Contract terms - Are legal terms acceptable?

Read: references/evaluation-criteria.md for comprehensive framework Read: references/technical-assessment.md for technical evaluation Read: references/pricing-models.md for pricing analysis Use: assets/scorecard-template.xlsx to score vendors systematically

Phase 3: Contract Negotiation

Goal: Secure favorable terms and avoid costly traps

Critical areas:

• Performance guarantees and SLAs
• Data ownership and usage rights
• Pricing structure and escalation terms
• Exit clauses and data portability
• Liability and indemnification

Read: references/contract-checklist.md for essential terms Reference: references/red-flags.md for problematic contract patterns

Common Vendor Patterns

The Overpromiser

Characteristics: Claims to solve everything, vague on technical details, aggressive sales tactics Red flag: "Our AI can handle any use case" Response: Demand specific technical explanations and verifiable references

The Feature Dumper

Characteristics: Long feature lists, complex pricing, unclear core value proposition Red flag: Can't explain what problem they actually solve Response: Force clarity on primary use case and success metrics

The Consultant in Disguise

Characteristics: Software license + mandatory professional services Red flag: Professional services cost more than software Response: Assess true cost of ownership, consider if you're buying software or consulting

The Model Wrapper

Characteristics: Thin layer over OpenAI/Anthropic APIs with high markup Red flag: No proprietary technology, just API access + UI Response: Calculate cost of building similar solution in-house

Full pattern library: See references/red-flags.md

Build vs Buy Decision Framework

When to read this section: Before committing to vendor evaluation, determine if building in-house is better option.

Key factors:

1. Capability availability - Does suitable vendor solution exist?
2. Time to value - Buy: weeks-months, Build: months-years
3. Total cost - Consider 3-year TCO for both options
4. Strategic importance - Core competency? Build. Commodity? Buy.
5. Team capability - Do you have talent to build and maintain?

Read: references/build-vs-buy.md for detailed decision framework

Using the Scorecard Template

The vendor scorecard enables structured comparison across vendors.

To use:

1. Open assets/scorecard-template.xlsx
2. List vendors to compare (up to 5)
3. Score each vendor on evaluation criteria (1-5 scale)
4. Review weighted scores and vendor comparison chart
5. Document decision rationale

Customization: Adjust weights based on priorities for your specific use case.

Reference Documents

references/evaluation-criteria.md

Comprehensive scoring framework across all vendor evaluation dimensions. Includes specific questions to ask, what constitutes good/bad answers, and how to weight criteria for different use cases.

Use when: Conducting systematic vendor evaluation

references/red-flags.md

Catalog of warning signs indicating problematic vendors. Organized by category: technical red flags, business red flags, pricing red flags, contract red flags, and behavioral red flags.

Use when: Initial vendor screening or reviewing proposals

references/pricing-models.md

Guide to AI vendor pricing models (per-seat, usage-based, platform fees, etc.), fair market rates, what drives costs, and how to negotiate. Includes pricing red flags and total cost of ownership analysis.

Use when: Evaluating vendor pricing or negotiating contracts

references/technical-assessment.md

Framework for assessing technical capabilities: architecture review, model evaluation, integration complexity, scalability, security, and data handling. Includes specific technical questions to ask.

Use when: Deep technical evaluation of vendor capabilities

references/contract-checklist.md

Essential contract terms for AI vendor agreements: performance guarantees, data rights, pricing protection, exit terms, liability, and support commitments. Includes negotiation guidance.

Use when: Contract review or negotiation

references/use-case-fit.md

Framework for assessing whether vendor solution actually fits your use case. Includes questions to ask yourself, questions to ask vendor, and warning signs of poor fit.

Use when: Initial vendor screening or use case definition

references/build-vs-buy.md

Decision framework for whether to build AI capability in-house vs purchasing vendor solution. Includes total cost analysis, capability assessment, and strategic considerations.

Use when: Before committing to vendor evaluation process

Assets

assets/scorecard-template.xlsx

Structured spreadsheet for vendor comparison with:

• Evaluation criteria organized by category
• Scoring system (1-5 scale) with descriptions
• Weighted scoring based on priorities
• Vendor comparison charts
• Decision documentation section

Customize: Adjust criteria weights and add company-specific requirements