Sheet 04 · Capability matrix

AI Platform Comparison

Route work by sensitivity, capability, cost, and latency.

Comparison of AI platforms by strength, best uses, data class, and deployment
Platform	Core strength	Best for	Max data class	Deployment
Microsoft Copilot	Microsoft 365 productivity	Meetings, email, and documents SharePoint knowledge Power Platform workflows	Internal	Cloud (Microsoft 365 tenant)
ChatGPT	Broad analysis and prototyping	Research and data exploration Multimodal work Rapid prototyping	Public / Internal	Cloud (vendor-hosted)
Claude	Long technical documents	Specifications and contracts Root-cause reports Large software repositories	Public / Internal	Cloud (vendor-hosted)
Local Models	Controlled and offline workloads	Plant-floor and offline use Proprietary content Restricted workloads	Confidential / Restricted	On your infrastructure (Ollama)

No single model family wins everywhere. The comparison below reflects each platform's demonstrated strengths:

Microsoft Copilot — Microsoft 365 productivity, meetings, email, documents, SharePoint knowledge, Power Platform workflows.
ChatGPT — broad analysis, research, multimodal work, data exploration, rapid prototyping.
Claude — long technical documents, specifications, root-cause reports, contracts, and large software repositories.
Local models (Ollama) — controlled, offline, plant-floor, proprietary, or restricted workloads where data must not leave your infrastructure.

Routing principle: classify the data first (Public, Internal, Confidential, Restricted), then pick the platform. Restricted engineering, defense-related, export-controlled, or customer-controlled information must never be sent to public consumer AI services.