AI Tools & Platforms

AI platform pricing changes often, so treat this as a map of how each one charges, and confirm current prices on the vendor's own page.

ChatGPT and Claude business plans

Both sell per-seat business plans for smaller teams, and enterprise plans priced by quote for larger organizations.

Microsoft 365 Copilot

A per-user license on top of Microsoft 365.

Google Gemini

Built into Google Workspace business plans rather than sold only as a separate add-on.

API pricing (for custom)

Per-token pricing that varies by model. Anthropic, OpenAI and Google all publish their rates.

Bottom line

Seat prices at the major platforms are close enough that capability and ecosystem should decide, not price.

Asking right questions reveals vendor reality.

Technical capability

• Show specific examples for our use case (not generic demos)
• Quality metrics on tasks like ours
• What happens when AI fails?
• How do you handle hallucination?

Integration

• How does this connect to our existing systems?
• What's the integration timeline and cost?
• API access and limitations?

Security and compliance

• SOC 2 Type II?
• Data residency options?
• Data used for training?
• GDPR, sector compliance?

Vendor health

• Financial stability?
• Customer references in our industry?
• Roadmap commitments?

Bottom line

Probing questions reveal vendor reality. Diligence prevents bad selections.

Three major clouds compete for AI workloads. Each has strengths.

AWS

Strongest: Bedrock for managed AI, broad service catalog, mature enterprise customers. SageMaker for custom ML.

Azure

Strongest: OpenAI integration (GPT models native), Microsoft ecosystem (M365 alignment), Copilot. Enterprise comfort.

Google Cloud

Strongest: Native AI capabilities (Gemini, Vertex AI), TPUs for training, BigQuery AI integration.

Choosing

Most enterprises multi-cloud. Microsoft for M365-heavy. AWS for breadth. Google for AI-specific workloads.

Bottom line

All three competitive. Multi-cloud common. Choice depends on existing ecosystem.

AI contracts have terms not in standard SaaS contracts. Critical to address.

AI-specific terms

Model training data rights, output ownership, ethics commitments, regulatory cooperation, AI bias liability, model versioning, prompt injection protection.

Data ownership

Customer data: clearly yours. Output: depends on contract — often customer. Training data use: critical to specify.

Model versions

AI models change. Contract should address version updates, capability changes, deprecation.

Bottom line

AI contracts require AI legal expertise. Standard SaaS reviewers miss key considerations.

RFPs essential for substantial AI purchases. Template and process matter.

RFP components

Business context and goals, technical requirements, integration needs, security/compliance, evaluation criteria, response format, timeline, contract terms.

Process

1. Internal alignment on needs
2. RFP creation
3. Vendor distribution
4. Q&A period
5. Vendor responses
6. Initial evaluation
7. Vendor presentations
8. POCs
9. Final selection

Timeline

3-6 months typical for substantial AI RFPs.

Bottom line

Structured RFP process prevents costly mistakes. Worth the time investment.

AI vendor choice is high-stakes. Multi-year commitment, integration dependency.

Selection process

1. Define requirements
2. Research market
3. Create shortlist (3-5)
4. Demo and discovery
5. RFP for finalists
6. POC for top 2
7. Reference checks
8. Contract negotiation
9. Pilot deployment

Evaluation criteria

Technical fit, integration capability, security/compliance, financial stability, support quality, vendor roadmap.

Decision matrix

Weight criteria by importance. Score each vendor. Calculate overall fit.

Bottom line

Structured vendor selection prevents costly mistakes.

AI contracts have unique considerations. Negotiation matters.

Critical terms

Data rights, model training, exit terms, pricing flexibility, SLAs, ethics commitments, security obligations, indemnification.

Pricing protections

Multi-year price caps, usage flexibility, ramp periods. Don't lock into rigid pricing.

Data and IP

Customer data protection, no training use, model output ownership. AI-specific.

Exit terms

Data portability, transition assistance, advance notice on changes. Plan for vendor changes.

Bottom line

AI contracts require AI-specific negotiation. Standard SaaS terms insufficient.

AI search transforms research. Different platforms different approaches.

Perplexity

Dedicated AI search. Citations standard. Real-time information. $20/month Pro.

Google (Gemini search overlay)

AI summaries in regular search results. Integrated with everything Google.

Claude (web search)

AI assistant with web access. Conversational search experience. Part of Claude Pro/Team.

Bottom line

Different paradigms. Perplexity for research. Google for everyday. Claude for conversational. Many use multiple.

Both serve knowledge management; AI capabilities differ.

Notion AI

Built-in AI features, collaborative, web-first, $10/user/month AI add-on.

Obsidian

Local-first, plugin ecosystem (Smart Connections, Copilot plugins), private, free for personal.

Choosing

Team collaboration: Notion. Personal knowledge with privacy: Obsidian. Both work for individual use.

Bottom line

Notion mainstream choice. Obsidian power-user choice.