Custom GPTs: build your first in 20 min.
Custom GPTs are ChatGPT's most underused power feature. Every Plus subscriber can build their own — a purpose-built assistant with custom instructions, conversation starters, knowledge files, and even tool integrations. Most users have never made one. This lesson walks 3 real builds: a sales follow-up GPT, a meeting prep GPT, and an executive summary GPT.
The mental model
Custom GPTs are reusable instructions you build once.
Instead of re-prompting Claude with the same setup every time, you bake the setup into a GPT. Open the GPT, type a short query, get high-quality output that matches the patterns you defined. The 20-minute investment compounds across every future use.
Workflow 01 Build a Sales Follow-Up GPT
GPT: "Sales Follow-Up Pro"
A GPT that drafts follow-up emails using the specific-friction pattern and your tone.
The prompt that works
Best use cases
- B2B sales reps
- Account managers
- Anyone doing customer-facing outreach
- Founders writing customer emails
Workflow 02 Build a Meeting Prep GPT
GPT: "Meeting Prep"
A GPT that helps you walk into any meeting prepared. Tell it who/what, get a brief.
The prompt that works
Best use cases
- Anyone with back-to-back meetings
- Sales reps prepping for calls
- Managers prepping for 1:1s and reviews
- Founders prepping for investor or partner meetings
Workflow 03 Build an Executive Summary GPT
GPT: "Exec Summary"
A GPT that turns long content into ruthless executive summaries that respect a CFO's attention.
The prompt that works
Best use cases
- Anyone who has to forward info to time-poor executives
- Investor updates
- Internal team briefings
- Long-meeting recap summaries
Final challenge: build the one GPT that pays off
Pick the workflow you do most often in ChatGPT. Build a Custom GPT for it in 20 minutes. Use it for the next week instead of regular ChatGPT. Most users find the quality jump immediate.
What you can do now
- Build a Custom GPT in under 20 minutes with the 4 essential fields (name, instructions, starters, capabilities)
- Write instructions that constrain output without paralyzing the model
- Decide what to keep in instructions (process) vs. what to put in each conversation (specifics)
- Test a GPT against real use cases before deploying it widely
- Recognize when to update vs. create-new for an evolving workflow