Anthropic Ships Features in Days. Here's What UK Businesses Can Steal.

Anthropic makes Claude—one of the most advanced AI systems in the world. You'd expect them to have some interesting internal workflows. What I didn't expect is how directly applicable their approach is to a 15-person service business in Hampshire.

Jenny Wen, Design Lead at Anthropic, recently showed how she uses AI to go from raw, messy feedback to polished product decisions in hours, not weeks. She has a phrase for it: “garbage in, treasure out.”

Here are three lessons worth stealing.

Lesson 1: Let AI take the first cut at everything

Jenny keeps a folder of personal notes—things from one-to-ones, random thoughts, observations from using her own product. Before a meeting or podcast, she tells the AI: “Read my personal notes and come up with speaking points.”

She doesn't write anything from scratch. The AI produces a first draft. She reacts to it, edits it, improves it. Her words: “I always let AI take a first cut at everything, and then I just react to it.”

This sounds small. It's not. The difference between staring at a blank page and reacting to a draft is the difference between 30 minutes and 5 minutes. Do that ten times a week and you've reclaimed four hours.

How to apply this in your business

Next time you need to write a client proposal, a team update, or a job description—don't start from scratch. Feed your notes, your previous proposals, and your client brief into an AI tool. Let it produce a first draft. Then edit.

It's not about the AI being perfect. It's about skipping the hardest part—the blank page.

Lesson 2: Automate your weekly intelligence

This is the one that made me sit up.

Jenny has set up a scheduled task that runs every Monday morning at 10am. It automatically:

• Reads a folder of user research transcripts and internal feedback.
• Scans social media (X, Reddit) for mentions of their product.
• Extracts the top themes and insights.
• Generates a presentation with three product ideas based on those insights.
• Sends it to her team via Slack.

Every Monday, before her first coffee, the team has a data-driven brief ready to discuss. No human touched it. No analyst spent Friday afternoon building slides.

At Anthropic, features now ship in days instead of weeks. Jenny attributes this partly to collapsing the feedback-to-action cycle. When insights arrive automatically, decisions happen faster.

How to apply this in your business

Think about the reports you create manually every week:

• Accountancy firm: Weekly summary of overdue invoices, upcoming deadlines, and client queries across all staff. Currently someone spends 2 hours compiling it from Xero and email.
• Recruitment agency: Monday morning brief showing new applications per role, pipeline status, and which clients need updates. Currently pulled from Bullhorn and spreadsheets.
• Marketing agency: Weekly performance dashboard across all client campaigns. Currently copy-pasted from Google Ads, Meta, and analytics.

Each of these can be automated. An AI agent connects to your data sources, runs every Monday at 7am, and delivers the report to your inbox or Slack before you arrive. Setup once, runs forever.

Lesson 3: Kill the annual plan. Iterate monthly.

Anthropic doesn't do annual planning. Jenny was blunt about it: “At the most innovative companies, it's less about annual planning theatre and more about iterating and learning from users.”

Her team does monthly planning. Maximum 12 items on the list.

This isn't just a tech company thing. Most UK service businesses I work with have some version of an annual plan—usually created in January, forgotten by March. The world moves too fast for 12-month roadmaps, especially when AI is changing what's possible every few weeks.

How to apply this in your business

• Monthly, not annual. At the start of each month, list your top 5–10 priorities. Review them at the end. Adjust. Repeat.
• Max 12 items. If your priority list has 30 things on it, nothing is a priority. Force yourself to choose.
• Use AI to track it. Feed your monthly plan into an AI tool. At the end of the month, ask it to compare what you planned vs what you delivered. It'll be honest in a way your team might not be.

The real insight: speed comes from removing bottlenecks

Anthropic doesn't ship fast because they have more engineers. They ship fast because they've removed the bottlenecks between “we learned something” and “we built something.”

The bottlenecks in most businesses are the same:

• Feedback lives in inboxes, spreadsheets, and people's heads—never aggregated.
• Reports are compiled manually, so they're always late and always incomplete.
• Decisions wait for the next meeting instead of happening when the data arrives.
• Staff spend hours on formatting and compiling instead of thinking and acting.

AI doesn't fix bad strategy. But it eliminates the grunt work between having information and acting on it. That's where most businesses lose days every week.

What to do this week

• Pick one report your team creates manually every week. Estimate how long it takes to compile.
• Pick one type of document you write from scratch regularly (proposals, emails, summaries). Try letting AI draft the first version.
• Pick one decision that's been waiting for “more data.” Ask yourself: could an AI agent gather that data overnight?

You don't need Anthropic's budget or engineering team. You need one automated report, one AI-drafted document, and one faster decision. Start there.

Sources: Jenny Wen (Design Lead, Anthropic) on Peter Yang's podcast, “How Anthropic Builds Products with Claude Cowork & Claude Code” (2026).