I spent three weeks chasing ChatGPT plugins that promised to "revolutionise my workflow" before realising most of them were solving problems I didn't have. The plugin ecosystem is half gold, half cargo cult. This is how to tell the difference—and actually enable the things that work.
The honest truth about ChatGPT plugins
ChatGPT plugins are API integrations that let the model call external services: fetch real-time data, run code, query databases, interact with third-party platforms. On paper, brilliant. In practice, you'll find a lot of half-baked integrations that add latency without much benefit.
The core limitation: plugins work only in ChatGPT Plus (the £15/month subscription) and ChatGPT Team. If you're on the free tier or using the API directly, you don't get plugin support. That's worth knowing before you invest time learning them.
How to enable ChatGPT plugins
If you're on Plus or Team, enabling plugins takes 90 seconds.
Step 1: Open ChatGPT and navigate to settings.
Click your profile icon (bottom left) → Settings → Features. You'll see a toggle for "Plugins" under Beta features.
Step 2: Enable the toggle.
Once enabled, you'll see a new button in the model selector (where you pick GPT-4, etc.). Click it and select "Plugins".
Step 3: Access the plugin store.
In any conversation, click the three-dot menu at the bottom left, then "Plugins" → "Plugin store". You'll see hundreds of integrations organised by category.
That's it. No API keys to juggle, no deployment. Just browse, install, and use.
Which plugins are worth installing
I've tested roughly 40 plugins over six months. Most are mediocre. Here are the ones I actually return to.
Code Interpreter (built-in, always enabled)
Not technically a plugin—it's a native feature—but it's the closest thing to essential. Lets ChatGPT execute Python, analyse CSVs, generate charts, debug scripts. If you're doing any data work, this is non-negotiable.
Web Browsing (OpenAI, built-in)
Gives ChatGPT access to current information. Useful for:
- Fact-checking recent events
- Pulling live pricing or availability
- Researching newly released tools
Latency is noticeable (adds 5–15 seconds per query), so I only enable it when I actually need current data. Toggle it off by default.
Zapier
Connects ChatGPT to 5,000+ apps (Slack, Gmail, Airtable, Stripe, etc.). Genuinely useful if you're automating workflows. Example: "Create a Slack reminder for tomorrow at 10am." It works, reliably.
Setup requires linking your Zapier account once. After that, it's seamless.
Retrieval (OpenAI, document search)
Let you upload PDFs, Word docs, or text files, then ask ChatGPT questions about them. Works well for:
- Analysing contracts or specifications
- Extracting data from reports
- Summarising lengthy documentation
I use this weekly. Upload a 50-page technical spec, ask "What are the API rate limits?", get the answer in seconds.
Everything else: skip it
Most plugins I've tried fall into one of three categories:
- Redundant. "Summarise YouTube videos" — ChatGPT + Web Browsing already does this.
- Niche. "Fitness tracker integration" — useful only if you're obsessed with that specific tool.
- Broken. Authentication fails, API limits hit, outdated integrations.
Before installing anything, ask: "Can ChatGPT already do this without a plugin?" If yes, skip it.
How to actually use plugins in conversation
Once enabled, plugins work automatically—mostly. ChatGPT decides when to invoke them based on your request.
Here's what that looks like:
Scenario 1: Using Zapier to create a calendar event
You: "Create a Google Calendar event for next Tuesday at 2pm called 'Project review'."
ChatGPT recognises this needs Zapier. It calls the integration, authenticates (if you've linked your account), and executes the action. You see a confirmation message.
Scenario 2: Using Web Browsing for current data
You: "What's the current price of Nvidia stock?"
ChatGPT enables Web Browsing (if you haven't disabled it), fetches the data, and returns the answer with a timestamp.
Scenario 3: Forcing a specific plugin
If ChatGPT doesn't automatically use the plugin you want, you can force it. In the plugin store, click on the plugin name. This "activates" it for your current conversation. ChatGPT will prefer it when relevant.
One caveat: plugin invocation is non-deterministic. Sometimes ChatGPT decides it doesn't need to call a plugin even when you expected it to. This is frustrating but by design—the model is trying to optimise for speed and cost.
Real-world example: analysing a CSV with Code Interpreter
Here's a concrete workflow I use monthly.
The task: I have a CSV of freelance invoices (date, client, amount, status). I want to find which clients are slowest to pay.
The setup:
- Open ChatGPT with Code Interpreter enabled.
- Upload the CSV file.
- Prompt: "Analyse this file. Calculate average payment time (date invoice sent to date paid) grouped by client. Show me the top 5 slowest payers."
What happens:
Code Interpreter runs Python automatically. It:
- Loads the CSV with pandas
- Calculates payment deltas
- Groups by client
- Generates a sorted table
- Optionally creates a bar chart
All in 10 seconds. No manual scripting required.
Output:
Client A: 45 days average
Client B: 38 days average
Client C: 62 days average
Client D: 29 days average
Client E: 51 days average
That's the real power of plugins—they turn ChatGPT from a text engine into a tool that can interact with your actual data and systems.
Common mistakes to avoid
Enabling too many plugins at once. Each one adds latency. I keep 3–4 active and toggle others on only when needed.
Assuming plugins are more reliable than native features. They're not. Web Browsing occasionally fails. Zapier sometimes times out. If reliability matters, build an integration yourself or use the API.
Forgetting to authenticate. Many plugins require you to link your account (Zapier, Slack, etc.). If you don't see an "Authenticate" button, the plugin won't work. Check the plugin details page.
Using plugins for things the API does better. If you're building a production system, don't rely on ChatGPT plugins. Use the API with function calling instead. Plugins are for exploration and ad-hoc tasks.
What I'd actually do
If you're on ChatGPT Plus:
- Enable plugins now (it's free with your subscription).
- Install Code Interpreter and Web Browsing. Use them for one week. They're the only two worth your time.
- If you need to automate workflows, add Zapier and spend 20 minutes linking your accounts.
- Ignore everything else. The plugin store is 80% noise.
- If you find yourself using plugins for the same task repeatedly, consider building a proper integration or using the API instead. Plugins are a convenience layer, not a replacement for real automation.
The honest take: ChatGPT plugins are useful, but only the boring ones. Skip the flashy integrations. Stick with Code Interpreter, Web Browsing, and Zapier. That's where the actual productivity gains are. If you're evaluating other tools to complement your setup, the comparison on techjournaler.com is worth a look for understanding what alternatives exist in the broader developer tooling space.