Agent.ai

I’m selfishly excited if Google Gemini truly is getting good now — mostly because of how much context Google has about me. The next wave of progress is going to be about context. I don’t feel confident about much in terms of AI predictions these days but I feel confident in making that one. But seeding ChatGPT et al with adequate context is SUCH a pain. I even have an ongoing thread whose sole job is to force things into its memory so I don’t have to keep reiterating things so deliberately. Google knows so much about me (especially if they let me connect my personal and work GSuites into a single sense of master context — hint hint Google / Google Workspace ) that they could give me the most valuable AI experience in the world if they made that leap. p.s. years ago I wrote about the difference between data collection being “creepy” vs “awesome”. Tl;dr is that if you: 1) ask permission & 2) *actually use the data to make my experience better not just help you make more money* (novel concepts I know 🙄) that people will willingly share more (and more accurate) information with you. This is a PRIME example of that. I went from being all “wow how does Google have actual terabytes of data about me in my privacy exports” to “wow I really hope Google figures out how to use their terabytes of data about me — I wonder if I can give them even more?” really quickly 🙃 To win the next phase of AI each of us will have to win the trust of our users to gain access to context. Fortunately that’s doable, yet so rarely done I’ll say it again: 1) Ask permission. Clearly, and as often as feasible. Even at the individual task or interaction level. When people feel like they’re in control and informed they trust you more — so WILDLY overcompensate here (without overly cluttering the UX). 2) Actually use the data — in a VERY obvious way — to improve the user experience. Like any interpersonal relationship, reference the data you use in outputs. People do this all the time talking to each other: “I know you like X so I got you Y” will be said by many a gift giver this month 🎁 Go Google Go! (Or go OpenAI / ChatGPT — *IF* you make it easy to import huge amounts of data that you then reference, because this manual memory management stuff is annoying as all heck) The winners of this race DEFINITELY aren’t set yet. But I’m rooting for whoever takes this approach (and I suspect I’m not alone 🫡).

Woo hoo! Launching my latest project, the Company Research agent. Was going to launch it at 1am last night, but decided there was no reason to tempt the fates. It's an agent that lets you...umm...easily do company research. I built it for myself (to lookup startups and other companies of interest). But, it's also useful for sales people, partnership developers, customer success and others. You access it at companyresearch .ai It pulls from a ton of different sources, including several proprietary/paid ones to pull together a pretty robust report that includes: * firmographics * top executives * web traffic * search marketing data * G2 product reviews. For publicly traded companies even shows things like revenue multiple and revenue per employee. Additional features: 1) You can save companies or add them to HubSpot (of course) 2) Customize the report by reordering sections or removing them 3) Add your own custom questions that you're curious about 4) Share links to reports with others (links don't require a login) Currently working on a "Smart Note" feature that will create a customized summary/note for you, and add it to HubSpot on the company object (with a link back to the full report). Lots more to come, but I believe in releasing early and often. Let me know what you think, or if you have ideas for improvement. Thanks for your support.

"What if AI makes a mistake?" Oh, my friends. It's not an "if" it's a "when". So what do you do? You need to plan for this inevitability. I could tell you how to take a systems approach to this, building in human-in-the-loop checks with a plan to measure and reduce hallucinations, errors passed through or deviations. But let's be honest, ain't nobody got time for that. Not most of you anyway. You need a simple framework. I've got you. Check, Cross-check, Confirm. I'll tell you what each of these mean AND how to implement them here. Read the details and get your plan: https://lnkd.in/d4-3FzSc

I love tools, but I'm not a "tools first" guy. I'm a "problem first" guy. This goes for software, woodworking, rock climbing or any of my other hobbies. That's why I really loved Carole Mahoney's episode of Prompted. She started with her own Sales training methodology. She wrote the book about it "Buyer First" and she's a sales coach for the Harvard Entrepreneurial MBA program. She starts with a proven methodology, and then brought in Agent.ai to help. Her belief that "sales skills are life skills" highlights the importance of cross-functional collaboration. This one's a must listen for anyone in sales or interested in innovation.

The next decade of work won’t be won by those who know the most — but by those who ask the most. We’re entering an era where curiosity is the new competitive advantage. The best leaders, salespeople, and operators won’t be defined by their technical depth, but by their ability to learn fast, experiment boldly, and bring others along. AI isn’t here to replace you — it’s here to reward your curiosity. Those who lean in, ask “what if?”, and explore new tools will unlock new ways to create value. Agent.ai just published this idea in more depth here: 👉 “The Future of Work Belongs to the Curious, Not the Technical.” https://lnkd.in/e2_EEetz Apple inspired us to think different. Agent.ai exists to stay curious. Stay curious, my friends — and start with a question. #AI #FutureOfWork #Curiosity #AgentAI

Ethan Mollick -- Laura Fitton and I love your posts on the "evolution" of prompting. Given how fast "best practices" evolve are you "collecting" your insights on your blog or elsewhere?