Code with Claude 2026 — By Topic

"Keynotes, demos, and conversations with the teams behind Claude." Day-opener at each city — sets the agenda across three tracks (Research, Claude Platform, Claude Code).

"A twenty-minute summary of what's new in Claude Code: what shipped, why we built it, and how to get started." Both cities ran this as the entry point before deeper CC sessions.

"Frontier models are getting more capable, fast. Where the curve is going, and what it means for developers building on Claude."

"Adaptive thinking and effort controls give developers a new decision: how much should Claude reason for a given task? This session covers thinking budgets, effort levels, and the cost, latency, and quality tradeoffs involved."

"Over the last year, capabilities that used to require heavy scaffolding have moved into the model: reliable tool use, context management, writing and running code, computer use, and more. Walks through what changed between model generations and how these compose into agents that finish work instead of just starting it." Not on the London agenda.

"How to apply core prompting principles to agentic systems that plan, act, and adapt." 45-minute extended session — a new addition for London.

Inside Claude Design, a tool that converts natural-language descriptions into production-ready, on-brand outputs. "Learn how a small team built a design tool that ships in your brand, from prompt to production." New product debut in London.

"Hands-on techniques for testing and comparing models against your use case, so you can make a confident call each time a new release ships." London's new workshop track.

"Building agents used to mean spending development cycles on secure infrastructure, state management, permissioning, and reworking your agent loops for every model upgrade. Managed Agents, on the Claude Platform, now handles that layer for you. This session covers how to build and deploy a production-grade agent at scale."

"How memory and dreaming turn Claude Managed Agents into self-learning systems. This session walks through design considerations for memory architectures and how dreaming verifies and enriches memory between sessions."

"A hands-on build session for Claude Managed Agents. You'll deploy a production-ready agent from scratch, then debug and monitor it live in the developer console." Workshop format introduced in London.

"Most of the AI value in your organization is locked in isolated experiments. That is not the Agentic Enterprise we've been promised. AI can help us ideate, orchestrate, and complete the work. Not just support." Asana's take on enterprise agentic workflows.

"Cut cost, manage context, boost intelligence." Live demos covering prompt caching, keeping context lean for long-running agents with tool search, programmatic tool calling, and compaction, and using the advisor strategy for a cost-effective intelligence boost.

"GitHub's Copilot team ships Claude to millions of developers across chat, CLI, coding agent, and code review, and has become one of the most demanding users of the Claude Platform. GitHub CPO Mario Rodriguez and Anthropic's Brad Abrams break down how the team pushes quality up and costs down at scale, from caching and evaluation to the new Advisor strategy."

"A live build from zero to deployed in thirty minutes. We'll build a feedback app spanning five roles and the full software lifecycle, using Claude and Google Cloud alongside subagents, MCP servers, and custom skills." Attendees test the finished app at the end.

45-minute hands-on. "Stand up Claude Code in Amazon Bedrock, teach it your team's conventions with CLAUDE.md, and turn your everyday workflows into reusable agent skills." Leave with a working, team-tuned agentic dev environment on Bedrock.

"Provision Claude in Microsoft Foundry, connect it to a real MCP server with Claude Code, and build a working agent. You'll leave with running code and a cupcake to prove it."

"The mechanics that separate basic Claude Code use from real leverage: CLAUDE.md done well, wiring tools in with MCP, packaging team knowledge as skills, and using auto mode safely." London workshop only.

"Ever wonder how the Claude Code team uses Claude Code themselves? Watch Boris Cherny, Head of Claude Code, and Jarred Sumner, creator of Bun, livestream their everyday workflows together." SF-only signature session.

"Routines turn Claude Code into a proactive teammate that reads your repo and opens a PR before you've opened your laptop." Covers triggers, context, and steering decisions; ends "one /schedule command away from your first routine."

End-to-end routine construction: trigger, context, steering. Practical implementation via the /schedule command, aimed at developers ready to deploy proactive agent workflows.

"When agentic coding goes from individual tool to org-wide default, the tool isn't the hard part…your processes are." Fiona Fung walks through what broke at Anthropic (review, ownership, hiring) and the norms they had to rewrite to keep shipping.

At Spotify the bottleneck has moved from writing code to orchestrating it — 96% of engineers use AI and PR frequency is up 60%. Covers Honk, a background coding agent on the Agent SDK, integrated with the Fleetshift migration platform and the Backstage catalog. Argument: "the same standardization that makes teams effective makes agents effective too."

Base44's path from one engineer to hypergrowth (including acquisition by Wix). Three challenges Claude Code helped with: "ramping new engineers, compressing the experiment-and-validate cycle, and keeping the lights on as the surface area grew." Plus the "elegant simplicity" principle that kept architecture uncomplicated as they scaled.

Three European companies, three distinct integrations: Delivery Hero's autonomous agent merging 100+ PRs/day, Doctolib governing Claude Code across its healthcare engineering org, and monday.com shipping Claude inside the product to non-developer users.

At Datadog, 90% of engineers adopted AI coding tools in the last four months, with Claude Code driving two-thirds. The reusable tools they produced for verification, debugging, orchestration sprawled into unmaintainable one-offs. Sesh Nalla shares how Datadog built Temper: a constrained framework that produces secure, reusable tools that compound across sessions and teams.

Man Group has trading signals in production "researched, backtested, and proposed by AI." Made possible in a regulated investment environment by two mechanisms: a governed skills framework and a core data layer that let Claude learn decades of quantitative trading methodology. Scaled across ~750 developers and quants with 100+ skills.

"Three teams building AI-native products — Cognition, Gamma, and Harvey — discuss the architectural decisions behind their stacks. The conversation covers multi-agent orchestration, MCP in production, autonomous agent design, and the tradeoffs each team has worked through along the way."

Software built on Lovable serves 600M+ monthly sessions. Three systems: a fleet-learning layer that catches coding mistakes, an eval loop that gates every model release, and continuous improvement mechanisms for Lovable itself. The distinction between rapid prototyping with AI vs. running production at consumer scale.

"Local agents hit a ceiling — they compete for your machine's resources, can't verify their own work, and bottleneck at one or two tasks at a time." Alexi Robbins on giving each agent an isolated VM so agents can write code, spin up browsers, test their own changes, and deliver merge-ready PRs in parallel — now behind 30%+ of Cursor's internal merged PRs.

"Most teams shipping AI products can't build evals that predict how a model will actually perform in production." Michele Catasta on closing that gap with ViBench — a public vibe-coding benchmark that scores whether the generated app works — and the offline/online evaluation loop behind Replit Agent that turns weeks of engineering into compounding overnight gains.

Three architectural patterns shaped Legora's legal-AI agent: "what they could reuse from coding agents, what they had to translate, and what they had to invent." Shows how document editing, linting, and bulk review were rebuilt for legal work, with a live demo.

"A conversation with Dario Amodei & Daniela Amodei moderated by Ami Vora." Anthropic's two co-founders on stage.

"When Opus 4.5 landed, v0 was ready on day one — not by luck, but by design." Guillermo Rauch with Angela Jiang on how Vercel architects for model step-changes: the bets that paid off, the ones that didn't, and what becoming an 'agent-pilled company' actually looks like inside a frontier platform team.