The enterprise AI landscape - from an independent perspective.

Microsoft 365 Copilot is the flagship enterprise product - AI embedded across the M365 suite with enterprise security, compliance and admin controls. Copilot Chat provides a side-by-side AI experience in Teams and Outlook. Copilot Studio lets organisations build and deploy custom agents connected to enterprise data and line-of-business systems.

Microsoft's current direction is strongly agentic. Agent 365 gives organisations a governed framework for deploying, observing and securing AI agents at enterprise scale. Copilot Tuning allows organisations to fine-tune agents on proprietary data for specific workflows. Work IQ personalises Copilot behaviour using employee context. Azure AI Foundry and GitHub Copilot extend the platform into application development and software engineering.

Microsoft 365 E7 - the "Frontier Suite" - reached general availability in May 2026. It consolidates M365 E5, Microsoft 365 Copilot, Agent 365 and Work IQ into a single licence, including advanced Entra, Defender, Intune and Purview capabilities. E7 is the logical procurement vehicle for any E5-anchored organisation ready to move AI from pilot to production. Agent 365 is also available as a standalone add-on for organisations not yet at E5.

• Productivity automation within Microsoft 365 - meeting summaries, document drafting, email triage, spreadsheet analysis via Microsoft 365 Copilot
• Enterprise agent deployment - governed, auditable AI agents running business processes via Agent 365 and Copilot Studio
• Custom agent development tuned on proprietary organisational data via Copilot Tuning
• Regulated-industry deployments where Azure's compliance posture (IRAP, ISO 27001, SOC 2, APRA alignment) is a prerequisite - and where E7 bundles the compliance tooling alongside the AI capability

Select when: Your organisation is on M365 E5 and ready to move AI into production - Microsoft 365 E7 is the natural procurement step. Also the default choice where Azure's compliance certifications are a non-negotiable procurement requirement.

ChatGPT Enterprise provides access to GPT-5.4 and GPT-5.5 Pro with unlimited usage, SOC 2 compliance, SSO, audit dashboards, and data privacy assurances - user data is not used to train OpenAI models. Enterprise Key Management allows organisations to control encryption and data access at the workspace level. Workspace agents extend ChatGPT into connected applications including Slack and, via ChatGPT for Excel and Google Sheets, directly into spreadsheet environments.

OpenAI's strategic direction is increasingly agentic - moving from a tool organisations query to an agent that takes multi-step actions across systems. Deep research agents, persistent memory, and the Responses API for custom application development sit alongside the enterprise chat product as the platform expands its scope.

The AWS partnership materially changes OpenAI's enterprise reach. GPT-5.5 is now available on Amazon Bedrock, Codex (OpenAI's coding agent) runs on Bedrock with access via CLI, desktop app and VS Code, and Bedrock Managed Agents powered by OpenAI handles deployment, tool use, orchestration and governance within AWS's security and compliance controls. For organisations running workloads on AWS, this removes the previous tension between cloud infrastructure loyalty and access to frontier models.

• Knowledge work acceleration - research synthesis, complex analysis, report drafting and document-intensive workflows running on GPT-5.4 or GPT-5.5 Pro
• Workspace agent deployment connected to Slack and productivity tools for repeatable business workflows
• AWS-native AI deployment - GPT-5.5 and Bedrock Managed Agents for organisations that need frontier models inside their existing AWS security and compliance perimeter
• Custom AI application development using the Responses API - cloud-neutral when accessed via ChatGPT Enterprise, or AWS-native via Bedrock

Select when: You want frontier GPT-5 capability, need workspace agents integrated with existing tools, or - critically - are an AWS-native organisation that can now access OpenAI models and managed agents through Bedrock without leaving your cloud perimeter.

Anthropic ships across three commercial models: Claude Enterprise (per-seat, workforce-wide deployment with identity management, data controls and audit infrastructure), Claude for Work (per-seat for teams), and API access consuming Claude Opus, Sonnet and Haiku models per token. Claude Enterprise is purpose-built for enterprise IT, security and procurement requirements - not a consumer product adapted for business use.

Anthropic's current enterprise push includes Dispatch for building and managing multi-agent Claude workflows at scale - with scheduling, error handling and audit trails - and native integrations into Microsoft Word, PowerPoint, Excel and Outlook (public beta). A strategic partnership with SAP embeds Claude directly into SAP S/4HANA, SuccessFactors and Ariba. Anthropic's Responsible Scaling Policy and Constitutional AI approach remain public, auditable commitments that regulated-industry procurement teams can reference directly.

Claude Cowork - Anthropic's agentic desktop product included in Claude Enterprise - adds a knowledge work layer that sits across local files, connected applications and browser context. Enterprise administrators can govern Cowork via role-based access controls (managed manually or via SCIM from an identity provider), per-team spend budgets set from the admin console, and detailed usage analytics surfaced through the admin dashboard and Analytics API. OpenTelemetry support means Cowork activity - tool calls, file access, skills invoked, and whether each AI-initiated action was approved - feeds directly into SIEM pipelines including Splunk and Cribl.

• Document-intensive workflows - long contracts, policy documents, regulatory submissions and technical standards analysis, leveraging Claude's long-context capability
• Knowledge work automation via Claude Cowork - desktop-connected, file-aware agent tasks that span local systems and connected enterprise applications
• Multi-agent workflow automation via Dispatch for complex, multi-step processes requiring audit trails and error handling
• SAP-integrated AI for organisations running S/4HANA, SuccessFactors or Ariba - Claude is now native to the SAP Business AI Platform

Select when: Your organisation operates in a regulated industry, needs enterprise-grade audit infrastructure (including SIEM-compatible telemetry via Cowork) alongside AI capability, or runs SAP and wants AI embedded directly into core business systems.

Google's enterprise AI strategy runs across two tracks: Gemini for Google Workspace (AI embedded into Gmail, Docs, Sheets, Slides and Meet, including the Gemini Enterprise app with Agent Gallery for deploying prebuilt agents) and the Gemini Enterprise Agent Platform for organisations building and orchestrating custom AI applications on GCP. Gemini 3.1 Pro is the current flagship model for enterprise workloads. Model Garden provides access to over 200 curated models, including open-source alternatives alongside Gemini.

Google's current strategic direction is strongly agentic. Agent Studio lets organisations build custom AI agents connected to enterprise data and APIs. Agent-to-Agent Orchestration enables complex, multi-step workflows where agents hand off tasks between each other. Agent Registry provides a governed catalogue for deploying and managing agents at enterprise scale. NotebookLM continues to serve as an enterprise knowledge management and deep research layer. For organisations with significant data assets on GCP, the AI and analytics layer - particularly through BigQuery - is increasingly unified.

• Productivity automation for Google Workspace organisations - Gmail, Docs, Sheets and Meet workflows, with prebuilt agents available via the Gemini Enterprise app's Agent Gallery
• Multimodal analysis where the input includes images, video, or audio alongside text - a genuine first-principles capability differentiator
• Custom enterprise agent development and orchestration via Agent Studio and Agent-to-Agent Orchestration on the Gemini Enterprise Agent Platform
• Data-intensive AI use cases where BigQuery and the Gemini Enterprise Agent Platform can be combined - analytics, forecasting and unstructured data processing at GCP scale

Select when: Your organisation runs on Google Workspace, has significant data assets on GCP, or needs enterprise agent orchestration via Agent Studio and Agent-to-Agent Orchestration - particularly where multimodal capability across images, video or audio is a genuine requirement.

Perplexity's strategy is to replace traditional web search for professional knowledge workers - delivering synthesised, cited answers faster than a search engine and more reliably sourced than a general-purpose LLM. The enterprise offering runs on two tiers: Enterprise Pro (team management, advanced model access, SSO, audit logs and API via Sonar) and Enterprise Max (expanded usage limits, priority access, and the Agentic Research API for building custom research automation pipelines).

Computer for Enterprise is Perplexity's push into agentic workflows - a multi-model orchestration layer that can execute multi-step research tasks across connected enterprise systems, with native integrations into Slack and Snowflake. The Sonar API and Agentic Research API allow organisations to embed Perplexity's real-time retrieval and synthesis capability directly into custom applications and internal tooling. Unlike the other platforms on this page, Perplexity is not competing to be an organisation's primary AI platform - it is competing to be the research and retrieval layer that sits alongside one.

• Competitive intelligence, market research and regulatory monitoring where real-time, cited information is required - the core Enterprise Pro use case
• Automated research workflows via Computer for Enterprise, connected to Slack for team distribution and Snowflake for data-layer integration
• Custom research pipelines embedded into internal applications using the Sonar API or Agentic Research API
• Policy and standards tracking for legal, risk, procurement and regulatory affairs functions where source attribution matters for downstream audit

Select when: Your teams need real-time, cited research synthesis rather than generative content, or want to automate research workflows via Computer for Enterprise. Perplexity is best deployed alongside a primary LLM platform - Microsoft Copilot, ChatGPT Enterprise or Claude - not instead of one.