PeerLabs Research
Axis 1

Six capability domains

The landscape divides into six domains. These are not mutually exclusive -- several products span multiple domains -- but they represent distinct problem spaces with different user personas and evaluation criteria. Click any product tag to see details and highlight across all views.

1.1

Code Generation and Editing

Inline code suggestions, completions, and chat-based code generation within an IDE. Reactive. Tightly coupled to editor context. Commoditising.

GitHub Copilot Cursor Tab Windsurf JetBrains AI Amazon Q Gemini Code Assist
1.2

Agentic Coding

Autonomous multi-step code generation, testing, debugging, deployment. Plans, executes, verifies, iterates. Minutes to hours. Current battleground.

Claude Code ($2.5B ARR) OpenAI Codex Cursor Agent Devin Copilot Agent Jules OpenCode Aider
1.3

Knowledge Work Agents

Multi-step non-coding tasks: file organisation, document creation, data analysis, expense processing. The category Cowork created. This is where the "SaaS displacement" thesis lives.

Cowork Claude in Excel Claude in PowerPoint M365 Copilot Gemini Workspace Manus Notion Agent
1.4

Browser and Computer-Use

Navigate web browsers or desktop applications as a human would -- clicking, typing, scrolling, reading screens. Authentication boundary is the critical architectural decision.

Claude in Chrome OpenAI Operator Manus Browser Operator Opera Neon Perplexity Browser
1.5

Research and Analysis

Search, synthesise, and produce structured reports from multiple sources. Evaluated on citation quality, source diversity, and synthesis depth.

ChatGPT Deep Research Gemini Deep Research Perplexity Genspark
1.6

Security Agents

AI-powered vulnerability scanning positioned against traditional SAST/DAST. New competitive axis -- CrowdStrike -8%, Cloudflare -8.1% on Claude Code Security launch.

Claude Code Security OpenAI Aardvark Copilot Autofix
Axis 2

Four-level autonomy spectrum

Adapted from the academic literature (Sapkota et al. 2025) but calibrated to observed product behaviour. Higher autonomy = higher capability and higher risk.

0
Reactive
Responds to immediate context. No planning, no multi-step execution. Sub-second latency.
Copilot inline, Gemini Code Assist tab, Amazon Q suggestions
1
Interactive
Chat-based assistance with short reasoning chains. May use tools but requires human direction for each step.
Claude.ai chat, ChatGPT, Gemini chat, JetBrains AI chat
2
Semi-Autonomous
Plans and executes multi-step tasks with human checkpoints. Pauses for approval on destructive or ambiguous actions.
Cursor Agent, Copilot Agent, Cowork, Claude Code (default)
3
Fully Autonomous
Extended autonomous operation (hours). Plans, executes, recovers from failures. "Vibe coding" territory. Highest capability, highest risk.
Claude Code (skip-perms), Devin, Codex (7+ hr), Manus
The Structural Shift

Escape from the IDE

The most significant structural development: the dissolution of the boundary between "coding tool" and "knowledge work tool." The coding copilot was always a trojan horse for a general-purpose agent platform.

The Anthropic progression

Click events to expand details

Mid-2025
Claude Code ships + click to expand
Terminal-based agentic coding for developers. Boris Cherny's "side project" from internal Bell Labs division.
Late 2025
Users repurpose Claude Code + click to expand
Vacation research, slide decks, email cleanup, subscription cancellation, wedding photo recovery, plant monitoring, oven control. "The underlying Claude Agent is the best agent." -- Boris Cherny
Jan 12, 2026
Cowork launches (macOS) + click to expand
Built in ~10 days using Claude Code itself. Desktop agent with file system access. Max subscribers, then Pro (Jan 16). Triggered $285B software stock selloff.
Jan 30, 2026
Plugin system ships + click to expand
11 open-source plugins: sales, legal, finance, marketing, customer support. Skills + slash commands + MCP connectors + sub-agents. Role-specific specialisation.
Feb 2026
Claude in Chrome / Excel / PowerPoint + click to expand
Browser automation using real user sessions (authenticated, not sandboxed). Excel add-in reads entire workbooks with cell-level citations. PowerPoint deck creation.
Feb 10, 2026
Cowork on Windows + click to expand
Full feature parity. 70% of desktop market unlocked. Persistent instructions (role, tone, formatting). Folder-specific instructions.
The key insight

Anthropic now has the broadest deployment surface of any agent vendor: terminal (Claude Code), desktop (Cowork), browser (Chrome), Office (Excel, PowerPoint), Slack, web (claude.ai), mobile (Claude app), and IDE (Xcode). The underlying agent architecture is domain-agnostic; the surfaces are presentation layers. No other vendor spans this range.

Anthropic's strategy is horizontal: one agent, many surfaces. Microsoft's strategy is vertical: one agent per application. Google's strategy is embedded: AI native to each Workspace app.

Implications for enterprise technology leadership
Category boundaries are unreliable

"Coding copilot" and "productivity suite AI" are converging. Procurement should treat them as a single landscape.

Architecture over domain

The underlying agent loop (plan, execute, verify, iterate) is identical whether refactoring code or reconciling invoices.

The trojan horse is real

Developer-adopted tools (Claude Code, Cursor) are expanding into non-developer workflows. Coding tool contracts will grow into enterprise-wide negotiations.

Axis 5

Five strategic postures

How the major vendors are positioning. Each represents a distinct bet on where agent value concentrates and how it is captured.

Anthropic

Horizontal Agent Platform

One agent architecture (Claude Agent SDK), many surfaces. Extend from developer base into general knowledge work via Cowork. MCP as ecosystem play.

Broadest surface area MCP ownership Audit/compliance gaps

OpenAI

Model Supremacy + Speed

Rapid model iteration (4 Codex variants in 6 weeks). Cerebras partnership for 1000+ tok/s. Codex as general computer operation agent.

Model velocity Cerebras speed App Server fragmentation

Microsoft

Ecosystem Integration

AI embedded in every M365 application. Copilot Studio for custom agents. Graph for data grounding. Now offers Claude models within Copilot.

1.4B+ Office users Enterprise trust High total cost

Google

Model + Workspace + Cloud

Gemini in Workspace at aggressive pricing ($14/user/mo). CLI open-sourced. Vertex AI for custom agents. Gems and Agentspace for no-code agents.

Price advantage 1M+ token context Enterprise share lags

Open Source / Local-First

Privacy and Portability

Model-agnostic, no vendor lock-in. OpenCode (95K stars), Ollama (NVIDIA partnership), LM Studio (headless daemon). 42%+ of developers running LLMs locally.

Data sovereignty API compatibility 175K exposed Ollama instances

Click a vendor to highlight their products across all tabs. Press ESC to clear.

Cross-Reference

Product-to-domain mapping

Major products mapped against six capability domains. Filled circles indicate primary capability; half-filled indicates partial or emerging.

Product Vendor Code Gen Agentic Code Knowledge Work Browser Research Security
Claude CodeAnthropic
CoworkAnthropic
Claude in ChromeAnthropic
Claude in ExcelAnthropic
OpenAI CodexOpenAI
OpenAI OperatorOpenAI
CursorAnysphere
GitHub CopilotMicrosoft
M365 CopilotMicrosoft
Gemini WorkspaceGoogle
Gemini Code AssistGoogle
JulesGoogle
ManusMeta (acq.)
DevinCognition AI
OpenCodeOpen source
OllamaOpen sourceLocal inference runtime -- enables all domains via compatible models
LM StudioLM StudioLocal inference runtime -- enables all domains via compatible models
Peerlabs Framework

Four Axes mapping

How the taxonomy maps to the Peerlabs Four Axes methodology for enterprise technology assessment.

F Functional

Model capabilities converging at code generation level; differentiating on agentic reasoning, multi-step execution, and context window. Autonomy level is the key functional axis -- Level 2-3 agents are where value and risk concentrate.

A Application

Enterprise adoption signals strong (NVIDIA, Dropbox, Cisco, Salesforce). But Cowork's enterprise readiness gaps (no audit logs, local-only plugins) suggest knowledge work agents are 6-12 months from enterprise-grade. M365 Copilot and Gemini are enterprise-ready but less capable.

S Systems

MCP achieving protocol-level ubiquity. ACP and AGENTS.md emerging as complementary standards. OpenAI's App Server is the notable holdout. Integration architecture decisions in the next 12 months will shape vendor lock-in for 5+ years.

P People / Processes

The "vibe coding" debate extends to "vibe working" -- Cowork enables non-developers to execute complex multi-step workflows. Workforce implications are being priced into markets. Security and governance lag capability by 6-12 months.

Related Intelligence
AI Copilots: State of the Practice
Productivity claims vs. measured reality
Agents at the Gate
The dissolution of the coding copilot category
Taalas: Model-Specific Silicon
Inference cost trajectory implications
Vendor Filter Active
Click a product to see details, or switch tabs to see this vendor's products highlighted.