Most people think “free AI tools” means basic chatbots and simple text rewriting. But some of the best new tools do real work for you, like turning a rough slide into a polished design, testing a website automatically, or turning a YouTube video into a clean data table you can export.
Below are seven free-to-try AI tools featured by Rob The AI Guy, along with practical ways to use each one right away.
1. Google Slides “Beautify this slide” (make ugly slides look designed)
Google Slides has a new upgrade that adds a button called “Beautify this slide.” The point is simple: take a basic, text-heavy slide and transform it into something that looks like it came from a designer, in seconds.
You can start in regular Google Slides using Google Slides presentation editor.
Rob shows it working like this: you pick a slide that’s basically just text, click the beautify option, and it generates a new version that’s visually designed (not just rearranged). It can also generate slides, images, and infographics, and you can add guidance on what you want it to do.
For background on Google’s direction here, this is also covered in Google’s Workspace updates, including design and image features in Slides: Google Workspace announcement on Slides design tools.
How to use “Beautify this slide” quickly
- Open your deck in Google Slides.
- Select the slide you want to improve.
- Click “Beautify this slide.”
- Choose what you want (slides, images, infographics), and add instructions if needed.
- Review the generated version, then tweak it.
What stood out in the demo
The generated slide included detailed elements that matched the original slide’s intent, including:
- Logos and icons that fit the words on the slide (like a YouTube logo).
- Highlight colors that matched the original formatting (yellow highlights carried over).
- A layout that looked “finished,” not like a template.
And it stays editable. Rob points out you can click “edit image” and adjust the result, resize elements, and change details.
2. Deep Agent by Abacus AI (automate complex tasks with a free tier)
Deep Agent (by Abacus AI) is positioned as a “do it for me” tool, not just a chat window. Rob calls it the most powerful AI tool he uses because it’s adaptive and can handle many types of outputs, including chatbots, workflows, videos (including lip-sync style videos), Shorts, PowerPoints, apps, and more.
The big update is that Deep Agent now has a free tier. You can try it using Deep Agent free tier from Abacus AI.
If you want a reference for what Deep Agent is designed to do, Abacus also documents it here: Abacus AI Deep Agent documentation.
Example automation: posting to X (Twitter) on a schedule
Rob’s example prompt is specific: read the last 10 tweets from an account, then post similar tweets that promote Deep Agent in an organic way. It should include a hot take to grab attention, and it should post two tweets per day at 2-hour intervals.
Deep Agent then:
- Asks follow-up questions (tone, what capabilities to promote, what hot take should be, hashtags).
- Has you authorize your X account.
- Builds the automation and generates a report showing what it did.
- Posts the tweets on your behalf.
The practical idea here is simple: connect it to competitors or topics you track, and let it stay current with what people are talking about on X, then post at whatever cadence you want.
Example automation: daily website feature testing
Rob also shows a setup that tests a website’s features. Instead of a person manually clicking around every morning, Deep Agent uses “computer actions” to:
- Open the site
- Click buttons
- Check that pages and features work
- Run on a recurring schedule
- Email you a report with the test details and results
The key outcome is you don’t need a human to run the same check every day.
3. NotebookLM “Data Tables” (turn sources into structured tables)
NotebookLM has a new feature called Data Tables. It takes a source you upload (website, PDF, YouTube video, and more) and turns it into a table.
You can access it using Google NotebookLM. Google also explains the feature here: Google announcement of NotebookLM Data Tables.
How it works best
Rob’s tip is important: don’t enable a ton of sources at once. If you want clean results, use one source, or keep it minimal.
Once your source is loaded, you can generate a data table and customize it by describing:
- What columns you want
- How long the table should be
- What kind of information to include
Demo example: turning a YouTube video into a table
Rob uses a YouTube video about making money with cars, focusing on buying exotic cars and not losing money on them. NotebookLM turns the full video into a structured table with columns like:
- Vehicle models
- Miles and modifications
- Performance details
- Owner intent
- Build style or philosophy
It also shows where the info came from, so you can trace it back to the source. Then you can export the table to Google Sheets and edit it.
4. Prompts.ai model comparison tool (test multiple AI models side-by-side)
If you’ve ever wondered why one model gives better output than another, this tool makes it easy to compare them.
Rob uses Prompts.ai text generation chat and model comparison, then opens “compare models.” From there, you can choose multiple models (ChatGPT, Claude, DeepSeek, Grok, and more) and run the same prompt across all of them.
A practical way to use it
Rob’s demo compares “ChatGPT 5” vs “ChatGPT 5.1” using a simple prompt:
- “Give me 10 title ideas for YouTube’s algorithm change for 2026.”
What you get from this setup:
- Side-by-side responses
- A feel for output style differences
- Speed differences (Rob notes 5.1 is much faster in the example)
You can also change settings like max tokens and streaming, which helps when you’re testing longer outputs or trying to control formatting.
5. Wispr Flow (voice-to-text that’s faster than typing)
Wispr Flow is a voice-to-text tool that turns speech into polished writing across apps. Rob highlights the reason people stick with typing even when it’s slower: typing takes more mental effort. Talking is more natural for most people, and it removes the “keyboard friction.”
You can check it out at Wispr Flow voice dictation app.
Rob shares a speed comparison from their site:
- Typical typing: 45 words per minute
- Speaking with Wispr Flow: 220 words per minute
That’s about 4 times faster.
He also points out it’s useful across roles: creators, customer support, sales, students, teams, lawyers, developers.
6. PhotoEditorAI.io (free AI photo editing without sign-up)
A common frustration with AI image tools is limits, credits, or getting throttled after a few edits. Rob highlights a free option: PhotoEditorAI.io.
Try it here: PhotoEditorAI.io free AI photo editor.
How Rob uses it
- Upload a headshot.
- Add a simple prompt like “make my sweatshirt really bright yellow.”
- Choose a model (he mentions ImageEditorAI 2.0, and also tries “Nano Banana” in the demo).
- Choose an aspect ratio.
- Generate the image.
What’s interesting is how targeted the edit can be. The goal is not to “re-imagine” the whole photo, it’s to change one specific part while keeping everything else consistent.
Rob also notes you can upload multiple images and do other edits like changing backgrounds or hair color. He emphasizes that you don’t even need an account, and he’s surprised it’s free given that model usage normally costs money.
7. TripleTen prompt optimizer (turn basic prompts into strong instructions)
If you use AI tools often and the output feels off, it might not be the model. It might be the prompt. TripleTen has a free prompt optimizer that takes a simple request and turns it into a clearer, more detailed instruction set for an LLM.
Use it here: TripleTen prompt optimizer tool.
What it does
- You pick a use case (automatic, coding, reasoning, creative, structured output).
- You can specify the expected output (list, comparison, detailed explanation).
- You can add guidance like tone, goal, audience, style, constraints.
Rob’s demo uses “Creative” and starts with a simple prompt: “Generate image of a sixpack transformation for a man.”
After generating, the tool outputs a more detailed prompt you can copy as plain text or JSON. The practical benefit is that it helps you get better results even when you’re not sure how to “talk to” the model.
What I learned using these AI tools (the “small wins” add up fast)
After seeing tools like these back-to-back, the biggest lesson is that productivity rarely comes from one giant change. It comes from removing tiny blocks that steal time every day.
A few takeaways that stuck:
- Slides are a time trap. If a tool can take a plain slide and produce a clean design, that’s hours saved across a month.
- Automation is only “advanced” until you see it working. Deep Agent’s examples show that you can describe an outcome and let the tool handle the steps.
- Turning messy info into tables is underrated. NotebookLM’s Data Tables feature is the type of thing you don’t miss until you have it, then you start wanting it everywhere.
If you like keeping up with where automation is heading, this ties closely to broader trends toward more independent AI agents: Future of AI in 2026 and agentic automation.
Conclusion
The best AI tools aren’t always the loudest ones. They’re the ones that remove work you hate doing, like slide design, manual checks, rewriting, or repetitive posting. If you pick just two from this list and use them daily, you’ll feel the time savings fast, and the rest become easier to adopt later.
If you’re keeping an eye on where AI is going next, it’s also worth reading about agent-style systems that can work for long stretches without stopping: Microsoft Kosmos AI scientist breakthrough. The more these systems improve, the more “background work” shifts from people to software, and that’s the real productivity change.
0 Comments