Maybe AI agents can be lawyers after all

2026-02-06 · Source: AI News & Artificial Intelligence | TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Emerging Technologies & Innovation · Depth: Intermediate, quick

Summary

Anthropic's Opus 4.6 model has significantly improved AI agent performance on professional tasks, as measured by Mercor's APEX-Agents benchmark. Initially, AI agents scored under 25% on tasks like law and corporate analysis, leading to conclusions that human professionals were safe from immediate displacement. However, Opus 4.6 achieved nearly 30% in one-shot trials and an average of 45% with multiple attempts, marking a substantial increase from the previous 18.4%. This improvement is attributed partly to new agentic features, including "agent swarms," designed for multistep problem-solving. Mercor CEO Brendan Foody noted the rapid progress, indicating that foundation model development continues at a fast pace.

Key takeaway

For CTOs and VPs of Engineering assessing AI integration into professional workflows, the rapid performance gains of models like Anthropic's Opus 4.6 on benchmarks like APEX-Agents signal a need to re-evaluate AI's potential impact. Your teams should begin exploring advanced agentic features, such as "agent swarms," to understand their applicability to complex, multistep professional tasks within your organization, rather than dismissing AI for such roles based on older benchmarks.

Key insights

AI agent performance on professional tasks is rapidly improving, challenging previous assumptions about human job security.

Principles

• AI capabilities evolve quickly
• Multistep problems benefit from agentic features

Method

Mercor's APEX-Agents benchmark measures AI agent performance on professional tasks like law and corporate analysis using one-shot and multi-attempt trials.

In practice

• Evaluate AI models with agentic features
• Monitor APEX-Agents leaderboard

Topics

• AI Agents
• Anthropic Opus 4.6
• Mercor Benchmark
• Professional Tasks
• Foundation Models

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Product Manager, Tech Journalist, Legal Professional

Related on AIssential

• Why AI Is Replacing Code Reviews Faster Than Anyone Expected — AI models like Opus 4.5 are rapidly shifting from collaborative tools to autonomous code generation agents.
• Agent Building Trends — AI agent development is rapidly evolving from assistants to autonomous employees and specialized solutions for niche problems.
• [AINews] Agents for Everything Else: Codex for Knowledge Work, Claude for Creative Work — AI agents are expanding beyond coding to automate diverse knowledge work and creative tasks, driving significant productivity gains.
• 🧠 Anthropic's 212-page system card for its latest Claude Opus 4.6 has many interesting stuff — Advanced AI models like Claude Opus 4.6 and OpenAI Frontier are transforming complex tasks and enterprise operations.
• Claude Agents Just Built a Fully Functioning C Compiler With Zero Human Input — AI agents can autonomously build complex software systems through coordinated, iterative development.
• How We Bet Against the Bitter Lesson — Agent Skills integrate human expert judgment and deterministic tools to create a new, efficient AI-human knowledge economy.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI News & Artificial Intelligence | TechCrunch.