Devplan raises $2.5M to turn customer feedback into shipped features. Read more on GeekWire.

BlogGuides

Claude Code for product managers: a practical guide, no engineering background needed

What a product manager can do with Claude Code, how to set it up without an engineering background, and how to give it your customer and product context.

Short answer: Yes, a product manager can use Claude Code, and you don't need to write code to get value from it. PMs use it to ask questions about the codebase, check what actually shipped, analyse customer feedback, draft PRDs and build quick prototypes. Its answers are only as good as the context you give it, which is why we connect it to Devplan: Claude Code then works from what customers actually asked for.

A note before we start. We build Devplan, which connects to Claude Code, so this guide includes how to give Claude Code your product and customer context. Everything else applies whether you use Devplan or not.

What Claude Code is, in plain words

Anthropic describes Claude Code as an agentic coding tool that reads your codebase, edits files, runs commands and connects to your other tools. The word "coding" puts many PMs off. In practice it's an assistant that can open the files you point it at, read them, and do work across them, instead of answering from a blank chat.

You don't have to live in a terminal to use it. Claude Code runs in four places:

  • The terminal, for people comfortable with a command line
  • VS Code and JetBrains, inside the code editor
  • A desktop app for macOS and Windows
  • The web, at claude.ai/code, including the Claude app on iOS and Android

It's included in every paid Claude plan. Pro costs $20 a month, or $17 a month billed annually, and Max starts at $100 a month for more usage (pricing). The free plan doesn't include Claude Code.

The real problem: context

Claude Code only knows what's in front of it. Ask it "what should we build next?" with nothing connected, and you get a confident, generic answer. Ask it with your customer requests, your past decisions and your code in reach, and you get an answer about your product.

There are three ways to give it context:

  1. A CLAUDE.md file. A plain-text file in your project that Claude Code reads at the start of every session (docs). Put your product's purpose, your customers, your terminology and your team's conventions in it.
  2. Files you point it at. Interview notes, exported support tickets, last quarter's PRDs.
  3. Connected tools through MCP. MCP (the Model Context Protocol) is an open standard that lets Claude Code connect to your tools and data, for example to update tickets in Jira or pull data from Slack (docs).

This is where Devplan fits. Devplan reads your calls, support tickets, Slack, Jira or Linear and your code, and keeps one up-to-date picture of your product: what customers ask for, what was decided and why, and what's been built. Through MCP, Claude Code reads that context while it plans and writes (integrations). Instead of pasting files in every session, you ask "what have customers asked for about exports?" and Claude Code answers from the evidence.

Setting it up without a technical background

  1. Get a paid Claude plan (Pro is enough to start).
  2. Pick where to run it. If the terminal feels unfamiliar, start with the desktop app or claude.ai/code in your browser.
  3. Open a project. Ask an engineer for read access to the repository, or start with a folder of your own documents.
  4. Start in plan mode. In plan mode, Claude Code researches and proposes changes but doesn't edit anything until you approve the plan (docs). It's the safest way to learn.
  5. Choose how much it can do alone. In manual mode, it asks before most edits, commands and network calls. In auto mode, which recent versions start in, a second model reviews actions instead of you. If you're unsure, use manual mode.

If you're comfortable with a terminal, Anthropic's setup guide has a one-line install for macOS, Linux and Windows.

5 things a PM can do with it this week

1. Ask questions about the codebase. "How does CSV export work today, and which columns does it include?" You get an answer from the actual code, without interrupting an engineer. Plan mode keeps it read-only.

2. Check what actually shipped. "What changed in the export feature since 1 September?" Claude Code can read the code history and summarise it in plain words, which is useful before a customer call or a release note.

3. Analyse customer feedback. Export a batch of support tickets or interview notes into a folder and ask what customers keep asking for, with quotes. Anthropic's own growth marketing team uses Claude Code to process CSV files with hundreds of ads and find the underperformers (Anthropic). The same approach works for feedback.

4. Draft a PRD from real requests. Give it the requests, the customer quotes and the constraints, and ask for a PRD in your template's format. Our guide to writing a PRD from customer feedback and the free PRD template show what good looks like.

5. Build a quick prototype. Describe a flow and ask for a clickable prototype in a new folder. Anthropic says its product design team feeds Figma files to Claude Code, and even its legal team built prototypes with it (Anthropic). Treat the result as a sketch for a conversation, not production code.

Save your best workflows as skills

When you find a prompt that works, save it. A skill is a SKILL.md file with instructions. Claude Code uses it when it's relevant, or you run it directly with a slash command, such as /prd (docs). A PM might keep skills for "summarise this week's feedback", "draft a PRD" and "write release notes from what shipped".

What not to expect

  • It doesn't know your customers unless you tell it. Without context, answers about priorities are guesses.
  • Prototypes aren't products. Hand them to engineering as a starting point, not as code to ship.
  • It needs access to be useful. A codebase it can't read, or feedback that lives in someone's inbox, stays invisible to it.
  • You still decide. It can rank and summarise. Choosing what to build is still your job.

Frequently asked questions

Can a product manager use Claude Code?

Yes. You don't need to write code. Start with the desktop app or the web version, use plan mode, and ask questions about your codebase, your feedback and your documents.

What's the difference between Claude Code and the Claude chat?

The Claude chat answers from what you paste into the conversation. Claude Code works directly with files, repositories and connected tools, can run commands, and keeps project memory in CLAUDE.md, so it suits repeat work on the same product.

Do I need to know how to code?

No. Reading code helps you check its answers, but most PM tasks, such as asking questions, analysing feedback and drafting documents, need none.

How much does Claude Code cost?

It's included in every paid Claude plan. Pro is $20 a month ($17 billed annually), Max starts at $100 a month, and Team plans start at $25 per seat a month ($20 billed annually).

Is it safe to use on our codebase?

You control what it can do. Plan mode lets it read and propose without editing. Manual mode asks before most edits and commands. Agree with your engineering team which repositories you can open and in which mode.


About the author. Chris Bee is the co-founder and CEO of Devplan.

Give Claude Code your customers' context. Start free with $100 in credits and connect Devplan to Claude Code, no credit card needed, or book a demo.

All posts

Put the ideas to work.

Devplan turns customer feedback into specs your team and AI agents can build.