Code with Claude 🤖
Four takeaways from Anthropic's developer conference in San Francisco — rapid growth, context switching as a superpower, customization gaps, and Slack-native AI workflows.
Attended Code with Claude — Anthropic's developer conference in San Francisco — and came away with a lot to think about. 🧠
Here are my four main takeaways:
1. The growth numbers are staggering. Anthropic planned for 10x expansion but hit 80x. That kind of scale creates both enormous excitement and a lot of organizational growing pains. It's a good problem to have, but it's still a problem.
2. Context switching between agent sessions is the real superpower. This is what separates casual Claude users from power users — the ability to operate across 30+ worktrees in parallel, spinning up and down agent sessions fluidly. It's a skill, and it's learnable.
3. Most teams are leaving customization on the table. CLAUDE.md files, MCP tools, and skills are how you bake institutional knowledge into your workflows. Without them, you're using a generic tool. With them, you're building a force multiplier that knows your codebase, your conventions, and your context.
4. Anthropic runs on Slack. And not just Slack the product — SlackBot, Slack MCP, and internal AI agents are all woven into how they work. Seeing a company that builds AI also rely on AI-in-Slack to do their best work was genuinely validating.
My two favorite sessions were "Beyond the Basics with Claude Code" and "Running an AI-native engineering org." Both gave me concrete things to bring back to my team.
Finished the day with a closing dinner alongside Slack colleagues and some Anthropic team members — one of those evenings where the conversation just keeps going. 💛
Organizations 🏢
Slack
