Claude Code Just Got WAY More Powerful

2026-05-07 · Source: How I AI · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, long

Summary

Anthropic's "Code with Claude" developer event introduced several enhancements to Claude Code and Claude API products. Key updates include "Routines" for scheduling tasks in Claude Code via cron, HTTP, or GitHub webhooks, allowing automated execution of code or workflows. The Claude API now features "Outcomes," enabling agents to self-grade against a defined rubric and iterate up to 20 times to achieve a goal. A new multi-agent framework supports defining hierarchical teams of up to 25 agents, each with distinct toolsets, working on a shared file system. "Dreams," currently in research preview, offers a primitive for consolidating agent memory by reviewing past sessions and writing important insights to disk. Additionally, usage limits for Claude Code's 5-hour sessions have doubled across Pro, Max, Team, and enterprise plans, with peak hour restrictions removed for Pro and Max, and increased rate limits for Opus models in the API.

Key takeaway

For AI Architects and Product Managers building agentic applications, these new Claude features offer practical tools to enhance automation and agent intelligence. Your teams can now implement robust self-correcting agents using "Outcomes," orchestrate complex workflows with the multi-agent framework, and automate routine tasks via "Routines." Consider integrating these capabilities to build more autonomous and efficient AI products, leveraging the increased usage limits for greater operational flexibility.

Key insights

Anthropic's new features enhance Claude's automation, self-correction, multi-agent collaboration, and memory capabilities.

Principles

• Define success with rubrics for agent self-correction.
• Orchestrate agents with explicit hierarchies and toolsets.
• Automate tasks via scheduled or event-driven triggers.

Method

Agents can iterate against a markdown rubric up to 20 times to achieve defined outcomes. Multi-agent teams, up to 25, can be programmatically defined with an orchestrator and delegates, each having specific tools.

In practice

• Automate weekly newsletter drafts from changelogs.
• Use agents to self-grade and refine PRDs against rubrics.
• Consolidate agent memories by reviewing past sessions.

Topics

• Claude Code Routines
• Claude API Outcomes
• Multi-Agent Framework
• Agent Memory
• Anthropic API

Best for: AI Architect, AI Product Manager, CTO, AI Engineer, Machine Learning Engineer, AI Student

Related on AIssential

• Code with Claude: The 5 biggest updates explained — Anthropic's new features enhance agent autonomy, collaboration, and memory, alongside increased usage limits.
• Anthropic's Claude Code Artifacts update brings live, shared dashboards and interactive workspaces to enterprises — Claude Code Artifacts transform technical workflows into secure, shareable, real-time interactive web visualizations, bridging engineering and non-technical stakeholders.
• I Watched the Entire Anthropic Workshop and Here Is a Recap — Engineers often underutilize agentic coding tools, treating them as autocomplete instead of autonomous execution engines.
• Live blog: Code w/ Claude 2026 — Anthropic is enhancing developer productivity and model autonomy through advanced agentic capabilities and platform integrations.
• Anthropic Introduces Routines for Claude Code Automation — Anthropic's Routines automate coding workflows via scheduled, API, or event-driven triggers on cloud infrastructure.
• Claude Code, Codex and Agentic Coding #7: Auto Mode — Auto Mode enhances agentic coding safety by selectively requiring human approval for high-risk actions.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by How I AI.