If you want to know how to start automating with AI, you’ve probably already heard it can transform your work. You might have even picked a tool. But there’s a gap between “I want to automate something” and actually having a workflow that runs on its own — and most guides skip straight over it. This one doesn’t. In the next 30 minutes, you’ll build a real, working automation: ChatGPT Deep Research runs a topic, and the results land in your Notion automatically — summarized and organized, without you touching a thing.
📋 Table of Contents
Before You Open Any Tool: The Decision That Changes Everything
The single most common reason first automations fail isn’t a technical problem — it’s a sequencing problem. People open Make or Zapier, look at the blank canvas, and then try to figure out what to build. That’s backwards. The tool should be the last decision you make, not the first.
Start with the task. Specifically, a task that meets three conditions: you do it at least a few times a week, it takes more than 15 minutes each time, and you could write down the steps clearly enough that someone new could follow them. If it clears all three, it’s ready to automate. If you’re still working out which tasks qualify, What to Automate with AI walks through a simple three-question filter that makes it obvious.
For this walkthrough, the task is research. Specifically: running a Deep Research query in ChatGPT, getting a detailed report back, and having that report automatically saved to Notion — organized and summarized — without any manual copying. If you regularly pull together information on topics for work, content, or decisions, this is the kind of automation that pays off within the first week.
How to Start Automating with AI: Picking the Right Tool
There are three realistic starting points for a first AI automation, and the right one depends entirely on what you’re trying to build.
ChatGPT alone (no connector needed)
Best for tasks that live entirely inside ChatGPT — drafting, summarizing, reformatting content you paste in. There’s no automation in the traditional sense; you’re just using AI to handle a task faster. Good starting point if you’ve never used any automation tool, but the output stays in ChatGPT unless you move it manually.
Zapier
The most beginner-friendly connector tool, with the widest app library (6,000+ apps). Best for simple linear workflows: trigger → one or two actions. The free plan is limited to 100 tasks per month, which goes quickly. If your workflow has a clear “when this happens, do that” structure and you don’t need complex logic, Zapier is the fastest way to get something running.
Make (recommended for this workflow)
Make has a visual canvas interface — you literally drag and drop modules and draw connections between them. It’s slightly more involved than Zapier upfront, but the free plan gives you 1,000 operations per month and it handles multi-step workflows with conditional logic far better. For a workflow that involves an AI step plus a Notion action, Make is the right tool. It also has a native OpenAI integration that connects cleanly without workarounds.
If you find yourself needing more power later, n8n is worth learning once you’ve built a few automations — but it requires self-hosting or a paid cloud plan. Start here with Make, then graduate when you hit its limits.
Build It: Deep Research → Notion, Step by Step
This workflow has two stages. Stage A is semi-automatic: you run Deep Research in ChatGPT and send the result to Notion with one click using a browser extension. Stage B is fully automatic: the moment a new page appears in your Notion research database, Make detects it, sends the content to OpenAI for summarizing, and writes a clean summary section back into the same page. Together, they take about 30 minutes to set up and run without you after that.
Stage A — Get Deep Research into Notion (5 minutes)
Step 1: Install the ChatGPT to Notion Chrome extension. Search “ChatGPT to Notion” in the Chrome Web Store and install it. Once installed, open the extension settings and connect it to your Notion workspace — it will ask you to authorize access.
Step 2: Create a Notion database for research. In Notion, create a new page and add a database called “Research Inbox.” Add a Title property (default), a Date property called “Researched On,” and a Text property called “Summary” — you’ll use this last one in Stage B. Leave it empty for now.
Step 3: Point the extension at your database. In the extension settings, select your “Research Inbox” database as the save destination. Test it: open any ChatGPT conversation, click the extension icon, and confirm a new page appears in your Notion database.
Step 4: Run a Deep Research query. In ChatGPT, click the tools menu and select Deep Research. Enter a topic you’d normally spend time researching manually. Let it run — it typically takes 5 to 15 minutes. When the report is ready, click the extension icon to save it to Notion. One click, and it’s there.
Stage B — Auto-summarize in Make (20 minutes)
Step 5: Set up your Make account. Go to make.com and sign up for a free account. Once you’re in, click “Create a new scenario.” This is where you’ll build the workflow.
Step 6: Add a Notion trigger. Click the “+” on the canvas and search for Notion. Select “Watch Database Items” as your trigger — this fires every time a new page is created in your database. Connect your Notion account when prompted, then select your “Research Inbox” database. Set it to check every 15 minutes.
Step 7: Add an OpenAI action. Click “+” again and search for OpenAI. Select “Create a Completion.” Connect your OpenAI account using your API key (find this at platform.openai.com under API keys). In the prompt field, enter something like: “Summarize the following research report in 5 bullet points. Each bullet should be one sentence. Focus on the most actionable findings: [map the page content from the Notion trigger here].” Set the model to gpt-4o.
Step 8: Write the summary back to Notion. Add one more “+” and select Notion again, this time choosing “Update a Page.” Map the page ID from Step 6 and the OpenAI output from Step 7 into your “Summary” property. This writes the AI-generated summary directly into the same Notion page that triggered the workflow.
Step 9: Turn it on. Click “Run once” first to test it with the page you saved in Stage A. If the summary appears in your Notion page, everything is working. Click the toggle in the bottom left to activate the scenario. From now on, every time you save a Deep Research report to Notion, Make picks it up within 15 minutes and adds a clean summary automatically.
Test It, Break It, Fix It
Almost every first automation breaks at least once before it runs reliably. That’s normal and expected — the goal of the first test run isn’t perfection, it’s learning where the gaps are. Here are the three issues that come up most often with this specific workflow, and how to fix them.
1. Make can’t find new Notion pages
This usually means the Notion integration doesn’t have access to your database. Go to your Notion database, click the “…” menu, and under “Connections,” make sure Make is listed. If it’s not there, add it. Then go back to Make and click “Determine data structure” in the trigger module — this forces it to re-check. After adding Make as a connection in Notion, wait 30 seconds before retrying the data structure check — Notion’s API takes a moment to register new integrations.
2. The OpenAI step returns an error
Nine times out of ten this is an API key issue — either it was entered with an extra space, or the key doesn’t have billing set up. Go to platform.openai.com, confirm your API key is active, and make sure a payment method is added (even with free credits, billing needs to be enabled). Copy the key fresh and re-enter it in Make. If you see a 429 error specifically, you’ve hit a rate limit — wait a minute and retry before changing anything else.
3. The summary writes to the wrong field
If the summary shows up in the wrong Notion property — or doesn’t appear at all — go back to the “Update a Page” module in Make and check the field mapping. Make sure you’re pointing to the “Summary” text property specifically, not the page title or another field. If the property doesn’t appear in the dropdown, refresh the Notion connection and try again. One specific thing to check: the Summary property in Notion must be type “Text” (not “Rich text” or “Notes”) for Make’s default mapping to work cleanly.
Once it runs cleanly three times in a row without errors, you’re done. Don’t optimize further at this stage — resist the urge to add more steps or tweak the prompt. Let it run for a week, see what you actually want to change based on real use, then adjust.
What to Build Next
Once this workflow is running reliably, you’ve crossed the most important threshold: you’ve built something that works. The next step isn’t to make it more complex — it’s to notice other tasks in your day that have the same pattern. Frequent, time-consuming, rule-based.
A few natural expansions from here:
- Add a Slack or email notification. Add one more step to your Make scenario — after the summary is written to Notion, send yourself a Slack message or email with the title and summary. Now you don’t even have to open Notion to know a report is ready.
- Tag by topic automatically. Add a second OpenAI step that reads the summary and outputs a category tag (e.g., “Competitor Research,” “Industry News,” “Product Ideas”). Map that tag to a Select property in your Notion database. Now your research inbox auto-organizes itself.
- Schedule recurring research. Instead of triggering on a new Notion page, flip the trigger to a schedule — every Monday morning, Make sends a prompt to OpenAI for a specific topic and creates the Notion page itself. Fully automated from start to finish.
The pattern you’ve just learned — trigger, AI step, action — applies to dozens of other workflows. Every time you catch yourself doing a repetitive, rule-based task, you now have the foundation to automate it.
Pricing and plan limits for tools mentioned in this post — including ChatGPT Plus ($20/month as of May 2026), Make’s free plan (1,000 operations/month as of May 2026), Zapier’s free plan (100 tasks/month as of May 2026), and OpenAI API rates — change regularly. Always check each tool’s official pricing page before subscribing or budgeting.
💬 Quick Answers
What do I do if the Make scenario runs but nothing appears in Notion after 15 minutes?
First, check Make’s scenario history — click the clock icon in the bottom bar. If the scenario ran with no errors but nothing appeared in Notion, the issue is almost always field mapping: the OpenAI output is being written to a field that doesn’t exist or has the wrong type. Open the “Update a Page” module, confirm the Summary property is selected (not the title field), and re-run manually using “Run once.” If the scenario shows it never ran at all, check that the trigger is set to watch the correct database and that Make has Notion connection access.
Does this workflow work with Claude or Gemini Deep Research instead of ChatGPT?
The Chrome extension step (Stage A) works with Claude and Gemini too — the ChatGPT to Notion extension supports multiple AI platforms. For Stage B, you’d swap the OpenAI module in Make for an Anthropic or Google AI module, or keep OpenAI for the summarization step regardless of which tool you used for research. The workflow logic stays exactly the same.
What if I don’t use Notion? Can I save to Google Docs or another tool instead?
Yes. Make has native integrations for Google Docs, Airtable, Obsidian (via webhook), and most other note-taking or database tools. The workflow structure is identical — you just swap the Notion modules for whichever tool you prefer. The trigger changes slightly depending on the tool, but the OpenAI summarization step in the middle stays the same.
My OpenAI API key is active but Make keeps returning a 401 error — what’s missing?
A 401 means Make can’t authenticate with the OpenAI API — almost always because the key was pasted with a leading or trailing space, or because the key was recently regenerated and the old version is still in Make. Go to the OpenAI module in your Make scenario, delete the connection, and create a fresh one by pasting the key directly from your clipboard without any modifications. If you regenerated the key after setting up Make, that’s the cause — the old key is invalid the moment a new one is created.
I set it up but the summary quality isn’t great. How do I improve it?
Tighten the prompt in your OpenAI module. The more specific you are about format and focus, the better the output. Instead of “summarize this,” try: “Summarize the key findings in exactly 5 bullet points. Each bullet must be one sentence and start with an action verb. Focus only on findings I can act on immediately.” If the reports you’re saving tend to cover a specific domain (e.g., competitor analysis, market trends), add that context to the prompt too — it significantly improves relevance.
Related guides on AI Workflows & Automation
📚AI Automation for Beginners: Stop the BusyworkThe complete guide to AI Workflows & Automation — explore all posts in the Categories menu above.→What to Automate with AI (And How to Stop Wasting Time on the Wrong Tasks)Find the right tasks before you touch any tool — a three-question filter that makes it obvious.→How to Connect Your Apps with AI Automation Tools (No Coding Needed)The practical guide to linking your tools together without writing a single line of code.→How to Automate Your Workday with AI (Step-by-Step Guide)Once your first workflow works, here’s how to build out the rest of your day.→n8n vs Make: What Nobody Tells Beginners Before They ChooseReady to go deeper? Here’s how to decide which tool to stick with long-term.→🔍 Everything here is grounded in real use — direct testing in actual workflows, combined with research pulled from real user communities, review platforms, and hands-on reports from people who’ve actually been there. Because one person’s experience only goes so far. Either way, it goes through the same lens: no jargon, no recycled takes, just what actually works for non-technical users. About DailyTechEdge →