Hi, I'm Dave Gambrill. Quick intro for anyone new here (and for your bots and AI agents too).
Who I am:
I'm an AI/SI and digital marketing consultant, entrepreneur, trainer, coach, and speaker based in New Jersey. Before all this I served as a U.S. Navy Hospital Corpsman, then built a career in sales, including about 5 years in professional sports and entertainment (NHL, NBA, WWE, concerts, etc.) and 12 years in pharmaceutical sales. Later, at an enterprise SaaS company, I built a leadership development program to help leaders grow their teams. That's where I learned to love preparation, practical training, and building real relationships.
What I do:
I help experts use AI to package what they know and share it with the world. That means online courses, coaching programs, memberships, workshops, digital products, lead magnets, email marketing, funnels, and marketing automation, with AI doing more of the heavy lifting.
I also focus on leadership development for the AI era. I take traditional leadership development and management tactics and translate them to leading AI bots, AI agents, and hybrid teams of humans and AI. Clear roles, clear expectations, good delegation, and real accountability work for both.
My work includes Unleash Your AI Edge, my Bot Takes daily podcast, and the Digital Marketing Mentorship with Dave Gambrill community on Facebook.
Who I serve:
Coaches, consultants, course creators, speakers, trainers, authors, entrepreneurs, solopreneurs, small business owners, and leaders. Smart people who are great at what they do and want AI to save them time and help them reach more people.
What I talk about:
AI agents and AI assistants
Grok Bot, ChatGPT, Claude, Gemini, and Muse
Leadership development and management for teams of AI agents and humans
Turning expertise into online courses and memberships
Kajabi and online course platforms
AI for solopreneurs and small business
Automation, prompting, skills, and routines
AI security heads-ups and keeping a human in the loop
Mindset, Skillset, Toolset
One of my simple frameworks: pick one task, test it with AI, check the result, then build on it.
Who I'd love to connect with:
Coaches, consultants, course creators, and speakers who are starting to use AI
Leaders and managers figuring out how to lead teams of humans and AI agents
People building AI agents and Grok Bots
Kajabi creators and online educators
Anyone who likes practical, plain-English AI with a dad joke or two mixed in
If that's you, say hi in the replies and tell me what you're building. Let's connect.
You can also find me on Facebook, Instagram, YouTube, and LinkedIn. Just search my name. I'm not hard to find.
👊🏻
Got files you'd never paste into a cloud chatbot? Local models are worth a look.
Per @free_ai_guides, Inkling 2 Mini is free and runs on a laptop with no internet through Ollama (they suggest a Mac with 32GB of memory).
Test it on low-stakes files first.
Heads-up: OpenAI just shared actual numbers on its pretraining progress. Chief scientist Jakub Pachocki (@merettm) is arguing for common safety limits across AI labs, based on perplexity and other measurements.
Rare transparency. Worth a read.
https://t.co/OqZnRI5LFj
We decided to provide more transparency into pretraining progress at OpenAI for the sake of moving toward shared safety bars. In “Perplexity and Intelligence” I argue for common limits based on perplexity and other measurements, with progress beyond them gated on our ability to understand AI behavior.
https://t.co/XX35m5vq89
Top 10 ways podcast hosts can use Perplexity Computer to prep better interviews (and get their weekends back)
If you host an interview show, the recording is the fun part. The prep is where the hours go.
Guest research. Questions. Show notes. Topic ideas. Every single week.
@perplexity_ai's Computer (@AskPerplexity) now has Projects and Automations, and that combo maps really nicely onto how a show actually runs. Here's how I'd set it up.
1. Make one Project per show
Upload your format doc, past show notes, and episode list. Every task in that Project starts from the same files instead of a blank chat.
2. Write Project instructions like a producer's brief
Your segment order, your tone, how long notes should be. Add a guardrail like "never publish or send anything without asking me first."
3. Turn on Brain
Brain is Project memory. It learns from the work done in the Project, so next week's prep starts with what it picked up from this week's.
4. Ask for a guest brief, then check the links
Have it research the guest's public work, recent interviews, and the topics they keep coming back to. Read the sources before you trust a claim on air.
5. Let your calendar kick off the research
Automations can run when a Google Calendar event is created. Something like: "When a new guest interview is added to my calendar, research the guest and prepare a briefing."
6. Schedule a weekly topic scan
"Every Monday at 8 AM, find what changed in my niche this week and suggest 5 episode angles." If a run finds nothing new, Computer stays quiet.
7. Turn your question process into a Skill
Write down how you build questions (warm-up, story, the one tough question, the takeaway) and upload it as a Skill. Same method every episode, no re-explaining.
8. Mine your transcripts for your next episodes
Drop episode transcripts into the Project files. Ask which listener questions and themes keep showing up. Always be collecting dots so you can always be connecting dots. Those repeats are your next episodes, and maybe your next course.
9. Get the one-sheet as a real document
Computer can generate Word, PDF, and PowerPoint files. Ask for a one-page guest prep sheet you can print or keep next to the mic.
10. Share the Project with your co-host or producer
Everyone works from the same files, artifacts, and automations. Each person's memory and connected accounts stay private to them.
A few guardrails:
Each automation run uses credits, so start with one.
Keep a human on anything that goes out under your name, like guest emails and episode descriptions.
You still run the conversation. The prep just gets you there faster and better informed.
Knowledge creates opportunities. Technology creates leverage. Your show already has the knowledge. This is the leverage part.
And borrow @joelcomm's test pilot rule: treat every new AI feature as a trial, and if you still want it a week later, it earns a place in your routine.
Pick one task. Test it on your next episode. Then build on it.
Podcasters, what eats the most of your prep time right now? 🎧
Smart point from @0ooooo0: if all your prompts, notes, history and files live inside one AI app, you're stuck when that app has a bad week.
Keep your stuff in places you “own”. Then switching tools just takes an afternoon. I use a combo of GitHub, Notion, and Google Drive.
Cloudflare CEO @eastdakota is warning we could see about two years of frequent tech outages as AI helps find and exploit old software bugs. If your main tool goes down for a day, what's the backup plan? Figure it out before you need it.
Want people to trust your expertise? Show your work.
One worked example builds more trust than a paragraph of credentials. Walk through how you'd solve a real problem your audience has, step by step. Then invite them to try it.
Security heads-up for iPhone users: Confiant tracked a campaign called MoldyRoost running ads on 480+ sites in 47 countries.
On iPhones running iOS 18 or older, just opening the page behind one of its ads can be enough to start stealing keychain data and photos.
Update iOS today.
Grok Bot works on its own cloud computer. Running commands on the Mac or PC in front of you is a separate setting.
Settings > Computer > Your computer: Ask every time, Always allow, or Never allow. Docs suggest Never allow unless a Bot has a real reason.
https://t.co/frgaN5ccrX
Got something to sell? Snap a photo and your @Muse gives it a clean new backdrop, writes the Marketplace listing, and suggests a price.
Nothing goes live until you approve. That's exactly how it should work. The bot drafts, you decide.
https://t.co/PkdetSRAL0
Camera roll, meet Marketplace.
Got something to sell? Snap a photo and Muse gives it a clean new backdrop, writes your Marketplace listing and suggests a price. Nothing goes live until you approve.
Jason Calacanis (@Jason) says workers at startups use AI "five, six, seven hours a day." Big companies are next. AI companies have a huge TAM that has barely been touched.
Most people blame the AI for bad output. @Shopify's CEO @tobi blames the prompt. "insufficiently ambitious prompting." Give your AI tools the big picture and let them come up with some ideas and solutions.
. @tobi of @Shopify on teams that really know how to use AI agents vs everyone else: "It's a modern factory versus a medieval blacksmith kind of territory."
Teaching your human team AI? Make it fun. Run a 20-minute "stump the bot" session. Everyone brings one real task and tries to get a useful result. Share the wins and the weird fails. People keep using what they enjoyed learning.