ChatGPT Tools and Integrations

Why Tools Matter

A plain language model can only do one thing: predict the next token. What makes ChatGPT genuinely useful in daily work is the layer of tools built on top of that core — tools that let the model search the web, generate images, execute code, and read your uploaded documents. When you know which tool does what, you stop treating ChatGPT as a single black box and start composing it with the rest of your software stack.

This chapter maps every major tool available in ChatGPT as of mid-2026, explains when to use each, and then shows you how to connect ChatGPT to Zapier, Make, Notion, Slack, and VS Code. The final section introduces the broader OpenAI ecosystem for readers who want to go beyond the chat interface.

Built-in Tools in ChatGPT

ChatGPT Plus and Team subscribers have access to a set of built-in tools that GPT-4o can invoke automatically — or that you can request explicitly.

Tool	What it does	When ChatGPT uses it
Web Search	Retrieves live web results	When you ask about recent events, prices, or anything after the knowledge cutoff
DALL-E 3	Generates images from text descriptions	When you ask it to "draw", "create an image", or "generate a picture"
Code Interpreter (Advanced Data Analysis)	Runs Python code in a sandbox	When you upload a file and ask for analysis, or ask it to solve a math problem with code
File Uploads	Lets you provide PDFs, CSVs, images, code files	When you paste or drag a file into the chat
Memory	Remembers facts across conversations	When you explicitly ask it to remember something
Canvas	Opens a side-by-side editing pane	When you ask it to write or refine a long document collaboratively

Enabling Tools

In ChatGPT web and the mobile app, tools are enabled by default for Plus subscribers. To manage them:

Click your profile icon (top-right)
Go to Settings → Personalization
Toggle Memory, or enable/disable specific tools per conversation via the toolbar below the chat input

Web Search

ChatGPT's web search tool lets the model browse live URLs and summarise results. This is essential for anything time-sensitive.

Example prompts that trigger web search:

"What is the current repo rate set by the RBI?"
"What are the top-rated CA coaching institutes in Pune right now?"
"Summarise the latest SEBI circular on mutual fund expense ratios."

What You Get

ChatGPT returns a synthesised answer with numbered citations. Click any citation to open the source. This is significantly more reliable than asking about facts that may have changed since the training cutoff.

Limitations

It cannot log in to paywalled sites (The Economic Times Premium, Bloomberg, etc.)
Search results can still be summarised incorrectly — always verify financial or legal data against the original source
It is not a replacement for a proper research workflow; treat it as a starting point

DALL-E 3 Image Generation

DALL-E 3 is OpenAI's image generation model, integrated directly into ChatGPT. You describe what you want in plain language and it produces a 1024 x 1024 pixel image (or wider/taller variants).

Prompt examples:

"Create a flat-design illustration of a small kirana store in Mumbai
 with a UPI QR code displayed at the counter. Warm evening lighting."

"Generate a professional banner for a Diwali sale — include
 deepams, warm gold tones, and space for the text '40% OFF'."

"Draw a simple flowchart diagram showing how Zomato's food delivery
 process works, from order placement to doorstep."

Tips for Better Images

Be specific about style: "flat design", "photorealistic", "watercolour illustration", "line art"
Mention aspect ratio explicitly: "landscape", "portrait", "square"
Describe what is NOT wanted: "no text overlay", "no cartoon style"
Iterate by saying "same image but change the background to a blue sky"

What DALL-E Cannot Do Well

Render accurate text inside images (letters often distort)
Reproduce real people's faces from descriptions
Match a specific brand's exact logo or colour palette

Code Interpreter (Advanced Data Analysis)

This is arguably the most powerful built-in tool. When you enable Code Interpreter, ChatGPT can write and execute Python code inside a secure sandboxed environment. It can read files you upload, run computations, create charts, and return the output — all within the conversation.

What It Can Do

- Analyse a CSV of sales data and plot monthly trends
- Clean a messy Excel file (remove duplicates, fix date formats)
- Run statistical tests (mean, median, correlation, regression)
- Convert file formats (PDF table → CSV, image → text via OCR)
- Solve complex maths step-by-step with verified numeric output
- Generate a bar chart of your expense categories

Worked Example — Analysing a Flipkart Sales Export

Suppose you download your Flipkart Seller Hub report as a CSV containing columns: order_id, product_name, sale_price, quantity, state, order_date.

Upload the CSV, then prompt:

"Analyse this file. Show me:
 1. Total revenue by state (top 10)
 2. Month-over-month revenue trend for 2026
 3. The top 5 best-selling products by quantity"

ChatGPT writes Python (pandas, matplotlib), runs it, and returns:

A table of top 10 states by revenue
A line chart of monthly revenue
A ranked list of best sellers

No Python knowledge required on your part — but understanding what the code does helps you catch errors.

File Size Limits

Files uploaded to Code Interpreter are capped at 512 MB per file. For larger datasets, consider splitting the file or using the OpenAI API with your own infrastructure.

File Uploads

Beyond Code Interpreter, you can upload files to any GPT-4o conversation for the model to read, summarise, and reason about.

Supported file types:

Documents:   PDF, DOCX, TXT, MD
Data:        CSV, XLSX, JSON
Code:        PY, JS, TS, SQL, and most programming languages
Images:      PNG, JPG, WEBP, GIF

Practical Use Cases

Upload a 50-page government report (e.g., the Union Budget PDF) and ask for a structured summary
Upload your company's HR policy PDF and ask "What is the leave encashment rule?"
Upload a screenshot of an error message and ask "Why is this error happening?"
Upload a research paper and ask "Explain this in simple terms for a 12th-standard student"

Connecting ChatGPT to External Services

Zapier

Zapier is a no-code automation platform that connects over 7,000 apps. With the ChatGPT + Zapier integration, you can trigger AI actions from any app event, or have ChatGPT's output flow into any app.

How to set up:

Go to zapier.com and create an account
Search for "ChatGPT" in the App directory
Choose a trigger (e.g., "New email in Gmail") and an action (e.g., "Summarise with ChatGPT → Send to Slack")

Example workflows:

Trigger: New row added to Google Sheets (order data from your website)
Action:  ChatGPT drafts a personalised order confirmation email
Action:  Gmail sends the email

---

Trigger: New form submission in Typeform (customer complaint)
Action:  ChatGPT classifies the complaint (billing / delivery / quality)
         and drafts a response
Action:  Response posted to a Notion database for the support team

Make (formerly Integromat)

Make offers a visual drag-and-drop canvas for multi-step automations. It is more powerful than Zapier for complex conditional logic and is popular with Indian freelancers and agencies.

Scenario example:
[Razorpay webhook: payment received]
  → [ChatGPT: generate invoice summary in Hindi + English]
  → [WhatsApp Business API: send to customer]
  → [Google Sheets: log transaction]

ChatGPT in Notion

Notion AI (powered partly by OpenAI models) is built directly into Notion. But you can also use a Zapier or Make integration to pipe ChatGPT into any Notion database.

What Notion AI / ChatGPT can do inside Notion:

Summarise a long meeting note into bullet-point action items
Translate a page from English to Hindi or Tamil
Rewrite a section in a more formal or casual tone
Generate a table of contents from page headings
Draft a new page based on a brief prompt

Direct integration via Zapier:

Trigger: New page created in Notion (tagged "Needs Summary")
Action:  ChatGPT reads the page body and generates a 3-line summary
Action:  Summary written back to the Notion page property "AI Summary"

ChatGPT in Slack

The official ChatGPT app for Slack (available in the Slack App Directory) lets you mention @ChatGPT in any channel or DM and get a response inline.

Common Slack use cases:

@ChatGPT summarise the last 20 messages in this channel
@ChatGPT draft a message declining a meeting politely
@ChatGPT what does ROI mean in simple terms?
@ChatGPT translate this message to Kannada

Teams use it as a shared research assistant — anyone can ask a question and the answer is visible to the whole channel, reducing repeated queries.

ChatGPT in VS Code

The GitHub Copilot extension (powered by GPT-4o) and the ChatGPT extension (by various publishers) bring AI assistance directly into your editor.

What you can do:

- Highlight a function → right-click → "Explain this code"
- Ask "How do I write a binary search in Python?"
- Ask "What's wrong with this SQL query?" and paste the query
- Generate a unit test for the selected function
- Refactor highlighted code to be more readable
- Ask for documentation (docstring) for a function

GitHub Copilot also provides inline ghost-text autocomplete as you type, making it the most seamless coding experience. It is free for students via GitHub Education Pack, which many Indian college students qualify for.

The Broader OpenAI Ecosystem

Beyond the ChatGPT interface, OpenAI provides a suite of models and APIs that developers and businesses use to build custom applications.

Assistants API

The Assistants API lets you build a custom ChatGPT-like experience with:

A persistent thread (conversation history stored server-side)
File retrieval (upload documents once; the assistant references them in every conversation)
Code Interpreter (same sandbox as in ChatGPT)
Function calling (the assistant can trigger your own backend functions)

This is how companies build internal knowledge bases — upload all company policy PDFs, create an assistant, and let employees ask questions in plain language.

Whisper (Speech-to-Text)

Whisper is OpenAI's audio transcription model. It converts spoken audio to text in 99 languages with high accuracy.

Use cases:

- Transcribe a 1-hour boardroom meeting recording
- Subtitle a Hindi YouTube tutorial automatically
- Build a voice-enabled customer support bot (voice → Whisper → ChatGPT → response)
- Dictate meeting notes on your phone during a commute

Whisper is open-source and can be run locally, making it useful for privacy-sensitive applications where you do not want audio leaving your infrastructure.

Embeddings

Embeddings convert text into a list of numbers (a "vector") that captures the semantic meaning of the text. Similar meanings produce similar vectors.

Why this matters:

"How do I pay using UPI?"
"Steps to make a payment via PhonePe"

Both sentences have different words but very similar embeddings.
An embedding-based search finds the second even when you search for the first.

This powers semantic search, document retrieval, recommendation systems, and the Retrieval-Augmented Generation (RAG) pattern — where a chatbot first retrieves relevant documents from a database before generating an answer.

OpenAI Product	Use Case
GPT-4o (API)	Custom chatbots, document Q&A, content generation
Assistants API	Stateful agents with file knowledge and tool use
Whisper	Transcription, subtitles, voice interfaces
DALL-E 3 (API)	Programmatic image generation in your app
Embeddings	Semantic search, recommendation, RAG systems
Fine-tuning	Training GPT on your own dataset for specialised tasks

Common Pitfalls

1. Forgetting that tools have to be active Web search does not run on every query — ChatGPT decides when to use it. If you need live data, explicitly say "search the web for the current...".

2. Treating Code Interpreter output as always correct The model can write buggy code. Always inspect the code block, not just the final answer. Ask "show me the Python code you used" to verify.

3. Uploading sensitive data to DALL-E prompts Your image prompt is sent to OpenAI's servers. Do not include employee names, client details, or proprietary product information in image generation prompts.

4. Assuming Zapier automations run in real time Free Zapier plans run on a 15-minute polling interval. For real-time triggers, you need a paid Zapier plan or a webhook-based approach with Make.

5. Ignoring file context limits When you upload a very large PDF, ChatGPT may only process the first several thousand words. For long documents, split into chapters or provide a specific page range to analyse.

6. Mistaking GitHub Copilot for ChatGPT Copilot and ChatGPT are separate products with different pricing and features. Copilot is primarily a code completion tool; ChatGPT is a general-purpose assistant. Both use GPT-4o but with different system prompts and interfaces.

Practice Exercises

Enable the web search tool and ask ChatGPT: "What is the current gold price per gram in India?" Then verify the answer against goodreturns.in. Note any discrepancy.
Upload any CSV file (even a small one with 10 rows) to Code Interpreter and ask ChatGPT to: (a) describe the data structure, (b) calculate the average of the numeric column, and (c) plot a bar chart. Screenshot the output.
Use DALL-E to generate a banner image for a fictional online course called "Master Excel in 30 Days" — use a blue and white colour scheme, include text space at the top, and a clean professional style.
Go to zapier.com and explore the ChatGPT integration page. Find three pre-built "Zap" templates that would be useful for a small e-commerce business. Write a one-paragraph description of each.
Research the OpenAI Assistants API pricing page and compare the cost of running 1,000 API calls using GPT-4o versus GPT-3.5 Turbo. Calculate which is more cost-effective for a business that needs only basic FAQ responses.

Summary

ChatGPT's built-in tools include web search, DALL-E image generation, Code Interpreter, and file uploads — each serves a distinct purpose
Web search provides live data; always verify critical facts at the source
Code Interpreter runs real Python in a sandbox — powerful for data analysis but the code should be inspected for correctness
DALL-E 3 generates high-quality images from text descriptions; specificity in the prompt dramatically improves output quality
File uploads let ChatGPT read PDFs, CSVs, code, and images and reason about their contents
Zapier and Make connect ChatGPT to thousands of apps without writing code, enabling powerful workflow automation
Native integrations in Notion, Slack, and VS Code bring AI assistance directly into tools you already use
The broader OpenAI ecosystem — Assistants API, Whisper, Embeddings — enables developers to build custom AI-powered products on top of the same models