Build Your First ChatGPT Plugin: A Reusable Assistant for Work You Repeat
Stop rebuilding the same prompt. In two hands-on hours, build your first ChatGPT plugin — a reusable assistant for a job you do often, such as client onboarding — with its own Skill, reference files and connected apps, ready to share with your team.
You have probably typed the same elaborate prompt a dozen times — to onboard a client, draft a particular kind of email, or check a document against a checklist. A ChatGPT plugin turns that repeated effort into a reusable assistant you configure once.
Plugins are where ChatGPT’s reusable assistants now live. OpenAI is retiring Custom GPTs — they stop running on 11-12-2026 — and plugins replace them: a GPT’s instructions become a Skill, its knowledge files become the plugin’s reference files, and its connected apps come across as apps. Custom actions, the chosen model and sharing settings do not carry over, so a plugin is worth building properly rather than simply converting.
In this two-hour session you build a real one, using a job you actually repeat. We pick the use case together — client onboarding is a common first choice — then write the plugin’s Skill, add the reference files it should rely on, and connect the apps it needs. If you already have Custom GPTs, we migrate one and check what needs rebuilding.
A central part of the session is honest about how these tools really behave. You will draft the Skill with ChatGPT’s help — meta-prompting, asking the model to help write its own brief — and then review, test and tune it. First drafts almost never produce the result you imagined; the skill is in the correction, and in owning the outcome rather than trusting it blindly.
By the end you have a working plugin, tested on your own cases, and you know how to share it privately with colleagues so the whole team works to the same standard. The product is still moving; the session works from whatever is live in your workspace on the day.
This is the ChatGPT edition; the companion session builds the equivalent — a no-code agent — in Microsoft Copilot.
- ■Explain what a ChatGPT plugin is, when it beats a one-off prompt, and where its limits are
- ■Build a working plugin with a clear Skill, reference files and the right connected apps
- ■Draft and refine the Skill by meta-prompting — then review and correct it, because first drafts rarely behave as intended
- ■Share a plugin privately within your team or workspace, and know what stays confidential
- ■Move an existing Custom GPT to a plugin, and know what does not carry over
- 01Plugins explained: a reusable assistant built from Skills, reference files and connected apps
- 02Choosing a good first use case — a repeatable job such as client onboarding
- 03Writing the Skill, adding reference files and connecting apps
- 04Meta-prompting to draft instructions — and why every draft must be reviewed and tuned
- 05Testing on real cases, fixing what misbehaves, and sharing privately within a workspace
- 06Custom GPTs retire on 11-12-2026: migrating one, and what you must rebuild by hand
Ordered directly from Kramer Consulting — no public dates. Pick a time that suits you or your team.
This is a Kramer Consulting intensive, offered directly — not through the Digital Learning Hub. It is delivered one-to-one or to a small team, online or on site, and tailored to your tools and your confidentiality rules. Get in touch to arrange a session.