A Salesforce Dev Built a $50K Feature in an Afternoon Using Claude Code

Jack Gimbert has shipped Salesforce components worth $50,000 in consulting fees. He's done it the traditional way: spec, dev, test, deploy, iterate. He knows exactly how long it takes and what it costs.

He says he can build the same thing in an afternoon now.

That claim is the opening of Revenue Engineer Episode 1, and the rest of the episode is Jack proving it live on camera.

The starting point: a plain-English user story generated by Claude, describing a feature to be built against a real Salesforce developer edition org. No pre-staged scaffolding. No fake data.

Jack hands it to Claude Code in VS Code and lets it run.

Here's what gets built:

• A ZIP Code custom object with the relevant fields
• An invokable Apex class that calls a free weather API as an external callout, triggered on create and update
• A Record-Triggered Flow to invoke the Apex action
• A Lightning Web Component on the record page to display the weather data

The build covers all the pieces a real feature requires: metadata creation, Apex logic, Flow configuration, and a front-end component. In a typical consulting engagement, those four pieces span multiple tickets and multiple days.

Then it errors during deployment. The deployment fails with a real error. Claude reads it, diagnoses the cause, patches the code, and redeploys. That part is not edited out. It happens on camera.

The other notable moment: when Claude hits a field access error, it doesn't stop and ask Jack what to do. It creates the permission set itself and keeps going. That kind of autonomous unblocking is what makes the "afternoon" claim plausible rather than hyperbolic.

1. The $50,000 claim

Jack is specific: "Components I've built for companies in the past that might have cost like $50,000 in consulting fees, you can build in an afternoon, in four hours." This isn't someone guessing at AI hype. It's a practitioner who has billed for this work before, comparing the before and after directly. The economic implication for Salesforce consultants is obvious and worth sitting with.

2. CLAUDE.md as project memory

One of the more practical takeaways from the episode is how Jack uses CLAUDE.md. This is a markdown file that Claude Code reads at the start of every session. Jack uses it to store org-specific context: naming conventions, project rules, configuration that would otherwise require re-explaining every time. Warren describes it as project configuration; Jack describes it as memory. Both are accurate.

3. Spot-checking best practices is still on you

During the code review at the end of the episode, Jack flags two things explicitly: SOQL in loops and credential storage. These are the places where AI-generated Apex can look clean and still be wrong. Claude's code passed review on most dimensions; these two areas are where Jack recommends always checking manually. The tools are fast. The judgment call on governor limits and security is still the developer's job.

Jack also mentions: "I don't think I've looked at Stack Overflow in at least a few weeks." That's not a dismissal of Stack Overflow. It's a signal about where the friction has moved.

Here's what you need to follow along with what Jack built:

• VS Code (free)
• Salesforce Extension Pack for VS Code
• Salesforce CLI
• Claude Code VS Code extension (the Max plan runs about $150/month)
• A Salesforce Developer Edition org (free) or scratch org

The Salesforce Extension Pack handles project setup, org authorization, and deployment directly from VS Code. Claude Code sits on top of that. You write (or generate) a user story, hand it to Claude Code, and watch it work through the metadata.

A few things to have ready before your first session:

• A CLAUDE.md file in the root of your project with your org context and any naming conventions you care about
• A Remote Site Setting or Named Credential configured if your Apex needs to call out to an external API (Jack addresses this distinction in the episode around the 11-minute mark)
• Plan mode turned on for the first run if you want to review Claude's intended steps before it starts making changes

The weather API Jack uses in this demo is free and requires no account. Jack mentions the specific API during the build; the episode description has the documentation links.

Revenue Engineer Episode 1 is on the walters954 YouTube channel. This is the first in an ongoing series where real Salesforce practitioners show their AI-assisted workflow on camera, against real orgs, with the errors left in.

If this format is useful to you, subscribe on YouTube and sign up for the newsletter at revenueeng.com. New episodes and written breakdowns go there first.

Revenue Engineer is the community for working Salesforce professionals leveling up on AI-augmented development and modern GTM tooling. Weekly drops, live sessions, and a founding cohort opening soon. Join the waitlist and you'll be first in line.

If you have questions about the setup or want to share what you built, drop a comment on the YouTube video. I read every one.