NotebookLM quietly became a business machine

Turn vetted forecasts into a real plan

July 09, 2026

Sponsored by

For a long time I filed NotebookLM under "handy summarizer." Upload a few PDFs, get a tidy recap, maybe a podcast for the commute. Useful, but passive. I never treated it as something that could do actual work.

Then I watched a walkthrough from Tina Huang, an ex-Meta data scientist turned creator, and by the end my jaw was somewhere near the floor. Her last NotebookLM video, about ten months ago, showed a tool that mostly summarized sources and spun up audio overviews. This new one showed the same product building docs, sheets, slides, infographics, videos, and writing and running its own code. Her line stuck with me: Google basically breathed general intelligence into a research tool.

I read the breakdown twice, because the shift here is bigger than any feature list. So let me walk you through what she found and why it matters for anyone trying to build or decide something.

How Jennifer Aniston’s LolaVie brand grew sales 40% with CTV ads

The DTC beauty category is crowded. To break through, Jennifer Aniston’s brand LolaVie, worked with Roku Ads Manager to easily set up, test, and optimize CTV ad creatives. The campaign helped drive a big lift in sales and customer growth, helping LolaVie break through in the crowded beauty category.

*Ad

The old way versus the new way

The old NotebookLM was a closed loop. You uploaded your sources, it answered questions grounded in them, and that was the ceiling. Helpful, but it waited on you for everything.

The new one, based on what she demonstrated, behaves like a partner instead of a lookup box:

  1. It finds your sources for you. Using the Gemini model, it helps track down and import high-quality sources instead of you hunting them one tab at a time.

  2. It writes and runs code in its own sandbox. So it can actually analyze data, not just describe how it would.

  3. It ships finished formats. Slide decks, videos (anime style included), infographics, quizzes, flashcards, and full reports.

  4. It connects out. It syncs with the Gemini web app and even exposes an MCP server for automations.

The difference is passive lookup versus an active research partner that hands you finished work. Once you see it that way, "note app" feels like calling a smartphone a calculator.

The business-idea demo that got me

Here's where it went from impressive to concrete. She opened a featured notebook called The World Ahead 2026 by The Economist. Featured notebooks matter because the sources are pre-vetted by real experts, so the answers you get are trustworthy by default. That alone is a quiet superpower: no garbage in, so much less garbage out.

Then she gave it a sharp prompt. Roughly: help me find market gaps to start a bootstrapped, automatable, solo B2C business that nets 20k a month within a year, something I can code and build with AI.

It came back with four specific gaps, each pulled from 2026 trends:

  1. An AI daily companion for people on oral GLP-1 weight-loss pills, coaching them through side effects and muscle loss.

  2. A privacy-first log for the "California sober" and microdosing crowd, since wine consumption is in steep decline.

  3. A no-code tool for creators to generate vertical microdramas from text.

  4. A sandbox that teaches Gen Z graduates to build and deploy AI agents, since entry-level hiring is contracting hard.

What struck me is that every idea traced back to a real trend and a real source. This wasn't the usual generic AI brainstorm that could have been written in 2019.

Why owning AI deployment is a career move

Across product, ops, and CX teams, a new role is emerging: the person making AI actually work day to day.

On July 16, three people living this shift, from Yoco, Zumba Fitness, and Fin, share what the job looks like, the skills they'd hire for, and the challenges ahead. Register to save your spot.

*Ad

From a list to a full research report

She didn't stop at four ideas, and this is the part worth stealing. She showed a repeatable loop:

  1. Ask for evaluation frameworks. She had it find sources on how to assess micro-SaaS and solopreneur ideas.

  2. Go deeper per idea. She pulled in industry-specific sources for each of the four.

  3. Score them. She had it rate each idea for viability against those bootstrapping and validation frameworks, saving each as its own output.

One honest catch she flagged: you can't clone a featured notebook the way you clone a Google Doc. So you copy the useful text into your own notebook by hand. Small friction, but real, and worth knowing before you start.

From those saved reports, it spun out a comparison slide deck, an anime-style video walking through a founder's 30-day validation plan, and two infographics. Its verdict: the GLP-1 companion as the top pick, with the "sober" wellness app a strong second. That's the whole arc, from vague ambition to a ranked, illustrated plan, inside one tool.

The part that was flat-out impossible before

Then she switched to investing, and this is where the code execution earns its keep.

She uploaded a snapshot of her real portfolio, exported The Economist's global forecast table to a Google Sheet, and fed that sheet back in as a source. Then she asked NotebookLM to find imbalances and opportunities by cross-referencing her holdings against 2026 predictions.

You could literally watch it run code to do the analysis. It surfaced things like a European defense spending surge, an India decoupling hedge as manufacturing leaves China, and a clean-tech angle. She was clear this is not financial advice, and that your results depend entirely on your own portfolio. But the workflow is the point: your private data, plus vetted forecasts, plus live analysis, all in one place. That combination used to require a spreadsheet, a chatbot, and a lot of copy-paste. Now it's one prompt.

HR and IT need to work as one. Here's how

HR and IT misalignment costs you more than time. Get the guide that closes the gaps across the full employee lifecycle.

*Ad

How to try this yourself

If you want to copy the approach, here's the shape of it:

  1. Start with a featured notebook so your sources are high quality from the first click.

  2. Give it a detailed prompt with real constraints: budget, timeline, skills, target outcome.

  3. Ask it to find and import extra sources to go deeper on each angle.

  4. Have it evaluate and save each result as its own output, so you build a library, not a single answer.

  5. Convert the winners into decks, videos, or infographics you'll actually use.

  6. Jump to the Gemini web app for creative what-ifs that go beyond your sources, since NotebookLM stays strictly grounded in what you gave it.

That last tip is a genuinely useful mental model. Use NotebookLM when you want answers locked to trusted sources. Switch to a broader chatbot when you want it to think freely with you. Knowing which tool you're reaching for, and why, is half the skill.

Where I'd start tonight

The thing I keep chewing on isn't any single feature. It's that the boundary between "research" and "doing the work" just dissolved. I used to gather sources in one place, think in another, and build in a third. This collapses all three into one grounded workspace, and it does the tedious middle for me.

So don't take my word for it. Open NotebookLM, load one featured notebook on a topic you actually care about, and give it a prompt with real constraints attached. Ask it to find gaps, score them, and turn the best one into a one-page plan. Fifteen minutes from now you'll either have a head start on something real, or you'll understand exactly why your old mental model needed the update. Either way you win.