AI tools I actually use (and how I’m using them right now)

Last week, I wrote about how most of us are still stuck mixing concrete (doing work by hand that AI can now do better, faster, and cheaper). This week, I’m showing you how I’ve put that mindset into action.

If you're building customer or employee experiences that actually work, speed matters. Relevance matters. And your ability to move quickly from insight to output might be the difference between launching something meaningful or missing the moment.

Here’s a breakdown of what’s in my stack, what I’m testing, what’s on deck, and the actual ways I’m using AI-native workflows to create more value for clients and their customer/users/employees faster than ever.

The tools I use weekly (and why)

Lovable – My go-to for building actual working apps with login, user roles/permissions, dynamic data, notifications, etc.. without writing a line of code. I upgraded to the $200/mo plan and have been building internal tools and client prototypes in hours/days vs weeks/months.

Bolt – An alternative to Lovable that’s also great for quickly generating UI from prompts, BUT with the added benefit of importing from Figma. I can pull designs into Bolt, refine, add dynamic data, and get to working proof of concept fast.

Anything.to.Design – I’ve been using this plugin to easily pull pages/screens from a website/app built with Lovable or Bolt into Figma so I can iterate/refine the design. This honestly feels like a cheat code right now and makes the transition from AI prototyping back to what developers/engineers expect (a Figma file) more seamless. I’m currently paying for the monthly html.to.design plugin for unlimited uploads, which is completely worth it IMO.

Fathom – Records and summarizes meetings. I drop the transcript into GPT, combine it with research, and turn it into strategy docs or product requirements. It also automatically assigns actions and ties meeting summaries to deals/contacts in Hubspot. Absolute must have.

ChatGPT – Like most folks, I use ChatGPT dozens of times a day for everything from research, synthesis, ideation, creating docs, drafting emails, etc. Check out my last newsletter for more details on specific workflows and prompts I’ve been suing.

Notion AI – Functions like a connected brain for the whole StealthX team. We store all Fathom transcripts, meeting notes, and project documentation here. Then anyone can ask questions and get insights for everything. It’s like a crazy hive mind for the whole team.

FigJam AI – I use FigJam’s AI capabilities to quickly spin up boards for client workshops and whiteboarding sessions. The team and I typically rough things out here before we dive into design/build, then take the workshop transcript and export from the FigJam board, give it to GPT, and use it to quickly synthesize into a summary doc. This used to take hours before, now it’s practically instant.

Gamma – One of the fastest ways to turn content or research into a quick deck. I use it for pitches, nicer looking summary docs, and decks (although I have come to hate decks). We’ve got a custom StealthX template loaded in and ready to go.

Supabase – My go-to when turning Lovable or Bolt prototypes into functioning tools. It’s an easy one-click setup and requires no technical expertise.

Riverside – I use this almost exclusively for recording and editing the Building Great Experiences podcast. I’ve used it for over 6 months and it’s become core to my workflow. They also keep adding features to make it even better. It easily removes filler words, mixes/masters audio, finds social clips based on duration/topic/speaker, and handles branding and exports in minutes. A must-have for anyone looking to produce great content consistently without spending hours editing or thousands of dollars to pay someone else to do it for you.

What I’m experimenting with right now

Firebase Studio – This week Google entered the no-code app builder (aka “vibe coding”) space with Firebase Studio. It’s still in Preview mode, but I’m really curious where they take it from here. Could become serious competition for Lovable, Bolt, Replit, v0, etc.

Gumloop – Much lighter weight and easier to use than the other automation tools I’ve tried (e.g., Zapier, Make.com, N8N, etc.). I’ve been using it to automatically watch folders, scrape data, run through preset AI prompts (this comes out of the box without any API keys), and load the data into Supabase or Google Sheets for apps I build with Lovable/Bolt. Pretty nifty.

Cove AI – Think visual workspace + generative AI. We used it to build an interactive journey map for a client last week, right in the canvas 🤯. Seems like it will be really useful for workshopping, mind-mapping, and using AI to quickly build a canvas with all docs, images, and assets for a project.

Replit 2.0 – It’s in early access, but so far seems like it’s going to be a strong contender for AI-assisted coding. My team has primarily been using Claude Code lately, but Replit may become our go-to.

What’s on deck to test next

General Agents – They claim on their website that they’re “The first realtime computer autopilot.” Essentially, you can create AI-powered workflows that control your computer for you. If this thing does half of what they promise, it’s going to be wild. I’m curious to see how it things take shape during the research preview.

Bubble – I’m returning to this with fresh eyes. One of the biggest challenges (currently) with Lovable, Bolt, etc. is that you can’t easily connect them to a content management system (CMS) where other folks can easily manage/update content. I’m hopeful tools like Bubble, Webflow, Framer, and Elementor will make this better.

Lindy AI – An AI virtual assistant that can schedule meetings, manage inboxes, and execute multi-step tasks. I haven’t set it up yet, but have heard a lot of folks talking about how well it’s done so I’ll be trying out this week. Stay tuned 😉 

Recent workflows that changed everything

This is the circuit I’m running almost daily:

1. Voice Dump: Use Otter.ai or Fathom.video to record a brain storm, workshop, or meeting.

2. Transcript to Strategy: Feed the transcript into GPT and add deep research around the competitors or related topics.

3. Generate Requirements: Ask GPT to write a product requirements doc (PRD) based on the transcript and research.

4. Generate Prompts: Use GPT to convert the PRD into Lovable/Bolt prompts.

5. Build Prototype: Feed the prompts into Lovable/Bolt. Get 80% of a prototype or PoC done in an hour or less.

6. Finish in Figma: Import screens to Figma with Anything.to.Design and polish the design. When ready, import design back into Bolt or upload screenshots to Lovable to update the code to match the changes.

7. Supabase Integration: Ask Lovable/Bolt to scrub out any dummy data and connect to Supabase to start storing real data. Prompt it to build out user roles/permissions and account registration/login.

8. Test & Iterate: Share with a few key stakeholders/users/customers, collect feedback, iterate. Loop in engineers and start collaborating so they can begin architecting a full-scale solution based on the prototype/PoC you built.

Result? From idea to working prototype/proof of concept in a couple hours. Sometimes less than an hour. Insanity! This took weeks/months less than a year ago..

What I’ve been listening to

Why AI Will Take Over the $20T Professional Services Industry – This episode from AI Daily Brief is a must-listen if you're in consulting, services, or any business that runs on brainpower. It breaks down exactly how and why AI is positioned to disrupt knowledge work, fast.

If I Wanted to Build a $1M+ AI Startup in 2025, I'd Do This – A wide-ranging conversation between Greg Isenberg and Andrew Wilkinson. If you’re thinking about product ideas or building anything in the AI space, this is full of gems.

GPT-4o Just Replaced Your Creative Team (For Free) – From the Marketing Against the Grain podcast. Explores how GPT-4o is changing the game for content, branding, and creative work. Less about fear-mongering, more about what’s actually possible right now.

Wrapping up

In last week’s edition (“Stop Mixing Concrete”) I laid out the mental shift that everyone needs to make. This week I wanted to share more examples of how the shift is playing out in my daily work. I’m absolutely loving the process of tinkering and experimenting with tools/ideas daily.

If your job is to improve customer experiences, create better internal tools for your team, or lead new product development.. I have a challenge for you:

• Try a voice dump to GPT to requirements doc to Lovable workflow this week.

• Try Gamma to turn your next outline into slides.

• Start a prompt library for your team with chunked use cases.

• Replace one manual process with an AI-powered automation in Gumloop or Lindy.

• Test a tool from the lists I shared above and let me know what you learn.

If you missed it, I co-authored the AI North Star Playbook to help business leaders identify the right problems to solve with AI, align teams fast, and move from vision to AI-powered execution with speed and clarity. Check it out if you’re trying to help your organization figure out where to start.

Let me know what you're experimenting with, or if there's something I should be testing. I’m hungry for more 🤘

Onward & upward,
Drew

P.s. I publish this newsletter weekly to share ideas about building smarter, faster, and more customer-centered. If this was helpful, forward it to a friend or check out past editions at stealthx.co.