Howdy wizards,
A special welcome to the 435 new subscribers who joined last week.
In this edition:
ChatGPT’s new connectors for Deep Research
Write LinkedIn content faster with Taplio
I created a research tool for you that uses Deep Research
Locate your best chair, lean back, and brew up some of that amazing bean juice.
Here’s what’s brewing in AI.
DARIO’S PICKS
Your business data + Deep Research = less digging
In a series of LinkedIn announcements this week, OpenAI introduced improvements to their Deep Research feature inside ChatGPT, including three connectors: Microsoft Sharepoint, Github and Dropbox.
Here’s the details:
Deep research can now search and analyse your connected data from these sources in real time; this gives you answers grounded in your company’s knowledge.
If you or your company uses Sharepoint and Dropbox, here’s a few powerful examples of what it could let you do:
Ask for a “fast-track onboarding kit” to give your intern immediate context of your business.
Scan all your boring policy and procedure libraries to surface gaps where you might be breaching current regulations.
Build a creepily detailed brief on a customer based on all internal notes + web data.
For your Github repo, here’s some ideas on how you can use the Deep Research connector:
Create an architecture snapshot easy to understand picture of your codebase for anyone new to it
Scan your repos, lists known security holes
Read all the changes since the last version and generate a friendly “What’s new” summary
The Sharepoint and Github connectors respect the users’ existing platform permissions, Deep Research only shows users what they have access to.
You can now also export all your deep research reports as well-formatted PDFs:
All three connectors are available for everyone on ChatGPT Plus, Pro, and Team now, with support for Enterprise coming soon. Exceptions for the Dropbox connector: EEA, Switzerland, and the UK.
To try these new features, inside ChatGPT on web, go to Settings → Connected Apps
Note that these connectors are different than File Uploads, like the ones that already exist for Google Drive and OneDrive, which only let you add files to ChatGPT messages.
Why it matters You can finally just let ChatGPT navigate the labyrinth of your messy company data. The key distinction to similar solutions is that it uses Deep Research, which often takes 10+ minutes and really crawls deep into your sources (and cites them).
To make it easy for companies to adopt, OpenAI emphasises that they don’t train on your data when using these connectors and that permission levels are respected (ie the user only sees what they have access to).
OpenAI is likely to continue adding new connectors to other software holding important company data. Many companies and enterprise platforms have been building dedicated AI tools for data retrieval, so the demand is quite obvious. ChatGPT is now (or could soon be) a viable alternative for this use case.
IN PARTNERSHIP WITH TAPLIO
Taplio’s AI is trained on 500M+ LinkedIn posts to help you write high-performing content in seconds.
✅ Pick your hook & format
🤖 Hit generate → AI writes the post
✍️ Edit, refine, and schedule—all in one place
Perfect for content creators, founders and professionals looking to build a brand and who value their time.
Subscriber-special: Use the code “Brew2” to get one month for $1
UP CLOSE
In this mini-series I share different ways I’m using AI from week to week, as well as practical tips & tricks I discover and actually use.
I created a tool for estimating market size with Deep Research
Note: the tool below isn’t just useful for market sizing—think of it as a workflow that you can learn from about how AI can be your co-pilot when doing research in general.
Many of you have told me you want AI to help make better decisions, but you're (rightfully) nervous about trusting these shiny language models with their frequent... creative interpretations of reality.
So I rolled up my sleeves and built something I think you’re going to like: a market sizing tool in Google Sheets that puts you—the human—in control while letting AI handle the grunt work. I wanted something that’s more than a prompt/black-box solution where you nod along pretending to understand how the numbers materialized.
Why market sizing
Market sizing is often a key part of how companies figure out whether that brilliant new product idea is worth pursuing, or if they should pivot before burning through (even more) VC money. I’ve used my previous experience in market research to break down the process into something that leverages AI's strengths while keeping human judgment at the wheel.
How it works
The first tab gives you an overview of the tool and explains key concepts (TAM/SAM/Market size)
Define the parameters
Tell the tool what you're sizing up
The yellow fields are for entering your key info about the market; the blue fields are the key assumptions of the model. You either research these yourself and use AI to validate your results, or just let AI do everything.
AI generates a tailored research prompt
AI research tab
When you have entered all your key info, there will be a prompt ready for you. Copy this to your clipboard.
Deep Research does the heavy lifting
Pasting the prompt in ChatGPT’s Deep Research
Paste the prompt to ChatGPT Deep Research and let it do its magic. You'll get a table with estimates and/or suggestions for revising your existing assumptions.
You review and finalize the model
Market Size Calculation tab
Go back into the Human research tab and enter your newly AI estimated/validated data. The last sheet will now contain a full breakdown of your market size.
The real beauty here is of course that, despite using AI, assumptions are transparent - you can trace numbers back to their source. And if something looks fishy, you can adjust it based on your domain knowledge instead of blindly trusting the AI.
The bigger picture (it’s not just about market sizing)
While this tool helps you crunch market numbers, it's really showcasing something more fundamental - how to collaborate with AI without surrendering your critical thinking.
When tackling complex problems, breaking them down into key variables accomplishes two things:
It keeps you in control of the research direction
It makes your conclusions explainable to others (and your future self!)
You don't need to build elaborate spreadsheets for every problem, but the skill of decomposing complex questions into verifiable chunks is gold for doing research with AI.
Want the tool?
You can get immediate, free access to the tool by just referring 1 subscriber to this newsletter! Details below.
PS I’m curious about what other research challenges with AI you are facing. Hit reply and let me know - I might feature your question in an upcoming edition.
THAT’S ALL FOLKS!
Was this email forwarded to you? Sign up here.
Want to get in front of 14,000 AI enthusiasts? Work with me.
This newsletter is written & curated by Dario Chincha.
Affiliate disclosure: To cover the cost of my email software and the time I spend writing this newsletter, I sometimes link to products and other newsletters. Please assume these are affiliate links. If you choose to subscribe to a newsletter or buy a product through any of my links then THANK YOU – it will make it possible for me to continue to do this.