Felix Rieseberg

How the engineer behind Claude Cowork actually uses Claude | Felix Rieseberg (Anthropic)

May 25, 202659 minEngineering Lead for Claude Products at Anthropic

AI Summary

→ WHAT IT COVERS Felix Rieseberg, engineering lead for Claude CoWork, Claude Code, and Claude desktop apps at Anthropic, demonstrates practical Claude workflows including a $19 hardware approval button, live artifact dashboards, and using email as personal inventory data — showing how abstraction layers unlock Claude's real productivity potential. → KEY INSIGHTS - **Abstraction layering:** When using Claude, repeatedly ask "how can Claude handle this instead of me?" at each step. Rieseberg caught himself manually entering furniture dimensions mid-task, stopped, and instead told Claude to extract purchase data from his email receipts — eliminating the manual step entirely and arriving at a richer result faster. - **Email as personal inventory:** Connect Gmail to Claude CoWork and query purchase history to build structured inventories of furniture, clothing, or any category of owned items. Rieseberg used this to populate a 3D interactive floor planner with actual furniture dimensions pulled from order confirmation emails, requiring zero manual data entry. - **Model selection heuristic:** Default to Claude Sonnet 4.6 for most tasks. Switch to Opus only when the problem itself is poorly scoped — meaning when you cannot clearly articulate what you actually want. Well-defined tasks like "extract floor plan units from this document" do not require Opus-level reasoning capacity. - **Live artifacts with connectors:** Build auto-refreshing dashboards in Claude CoWork by combining the "live artifact" prompt keyword with pre-authenticated connectors (Gmail, Spotify, Notion, Calendar). The refresh button pulls current data without re-entering API keys. Use this for daily briefings that include meeting prep, recent Slack activity from attendees, and relevant context. - **$19 hardware Claude buddy:** Purchase any Bluetooth-enabled IoT device, connect it via Claude desktop's developer mode hardware buddy feature, and offload Claude permission approvals to a physical button. Claude Code built the full firmware in one shot with zero corrections, enabling background async work without watching the screen. → NOTABLE MOMENT Rieseberg observed that children using Claude never ask "can it do this?" — they simply ask for what they want. Adults, conditioned by decades of software limitations, self-censor requests before making them. This generational difference in assumption-setting may be the largest barrier to effective AI adoption. 💼 SPONSORS [{"name": "Magic Patterns", "url": "https://magicpatterns.com/howiai"}, {"name": "Guru", "url": "https://getguru.com"}] 🏷️ Claude CoWork, AI Productivity Workflows, Live Artifacts, Hardware IoT Integration, Prompt Engineering

Read Full Summary Listen

Why Anthropic Thinks AI Should Have Its Own Computer — Felix Rieseberg of Claude Cowork & Claude Code Desktop

Latent Space

Mar 17, 202687 minMember of Technical Staff at Anthropic, Product Lead for Claude Cowork

AI Summary

→ WHAT IT COVERS Felix Rieseberg, engineer at Anthropic, explains how Claude Cowork evolved from Claude Code into a VM-based knowledge work tool for non-terminal users. The conversation covers architecture decisions around local versus cloud compute, skills portability, sandbox security tradeoffs, and how Anthropic evaluates agent products against knowledge work tasks rather than coding benchmarks. → KEY INSIGHTS - **VM-first architecture:** Claude Cowork runs Claude Code inside a lightweight Linux virtual machine rather than directly on the host machine. This lets Claude install Python, Node.js, and any other dependency without requiring user approval or IT permission, while network ingress/egress controls remain strict. The VM collapses empty disk space so actual storage use is lower than macOS reports, though startup latency remains a real tradeoff versus running Claude Code natively. - **Skills as portable markdown files:** Cowork skills are plain markdown text files, not proprietary database entries. This makes them inherently portable — users can store them in any folder, sync via GitHub repos, or install plugin marketplaces by pointing to a GitHub URL. The unresolved gap is interpolating personal variables (preferred airports, phone numbers, folder paths) into otherwise shareable skill templates, which Rieseberg identifies as an unsolved industry problem worth building toward. - **Eval against knowledge work, not coding benchmarks:** Anthropic evaluates Cowork against knowledge work tasks — personal finance, mortgage management, legal office workflows — rather than standard coding suites used for Claude Code. The system prompt is tuned weekly based on these evals, steering Claude toward heavier use of the planning tool and ask-user-question tool for longer-horizon tasks where ambiguity is higher and wrong outputs are more costly to reverse. - **Scaffolding has diminishing returns as models improve:** Rieseberg argues that investing heavily in task-specific scaffolding carries risk because each model generation reduces the gap between raw capability and scaffolded output. His current preference is to maximize tool access and safety boundaries rather than build elaborate correction layers, then let model improvements handle the rest. He observes this pattern already in MCP servers being partially superseded by skills as models generalize better. - **Local compute retains strategic value:** Rieseberg pushes back on cloud-first assumptions by noting that moving all agent work to the cloud requires solving authentication, data privacy, and permission synchronization simultaneously. A concrete example: reading Chrome cookies to enable cloud-based browser sessions would trigger bank account lockouts from location anomalies. Keeping Claude on the local machine sidesteps these problems while giving it access to the same file system, applications, and credentials the user already has. - **Build all candidates, then pick:** Anthropic's current product development process skips traditional spec-then-build cycles in favor of rapidly constructing multiple prototypes simultaneously and selecting the best performer with a small focus group. Rieseberg credits this shift — enabled by cheap execution — with Cowork's ten-day build timeline, while clarifying that existing primitives like Claude Code, the VM framework, and internal prototypes accumulated over eighteen months provided the actual foundation. → NOTABLE MOMENT Rieseberg describes a workflow where Cowork autonomously scans an internal crash dashboard each morning, separates fixable bugs from OS-level failures, writes a markdown prompt file for each fixable bug, then spawns a separate Claude Code remote instance per bug — effectively building a self-directed multi-agent repair loop without using Claude's native sub-agents feature. 💼 SPONSORS None detected 🏷️ Claude Cowork, AI Agents, Local Compute, Knowledge Work Automation, Sandboxing, Skills Portability

Read Full Summary Listen

Featured On 2 Podcasts

How I AI

Latent Space

All Appearances

How the engineer behind Claude Cowork actually uses Claude | Felix Rieseberg (Anthropic)

AI Summary

Why Anthropic Thinks AI Should Have Its Own Computer — Felix Rieseberg of Claude Cowork & Claude Code Desktop

AI Summary

Explore More

Never miss Felix Rieseberg's insights