AI Products: Augment Workflows, Don't Replace
Augmentation Before ReplacementThe mindset shift for your AI products |
Welcome to the 14 people who have joined us since last week! If you aren’t subscribed, join 3,033 smart, curious, AI folks by subscribing here
I found a way to help my business clients prioritize AI projects. But it was tough, the status quo way of thinking about AI is locked in hype. Crowds of AI builders have hammers, and they’re looking for nails. While pervasive, I can’t put all the blame on the public. AI has an aura of mystique - Maybe the answer to your business is behind one of these 150 prompts shared on LinkedIn?
Hype-thinking causes problems:
Adoption PathsAll technologies go through a period of maturity after they are released. The internet is a great example.
However ChatGPT only took 2 months to go from launch to 100M users. The infrastructure & best practices around AI haven’t had time to build a stable foundation. So how does this lead to hype in the market? The lack of AI’s mature foundation leads to uncertainty and when combined with grand claims, uncertainty leads to hype and confusion. Augmentation Before ReplacementI’ve talked with 100s of companies about how they’re thinking about AI adoption and I see common themes. A mindset shift has helped the top performing businesses stay grounded - Augmentation Before Replacement. They’re focused on applying AI to small work flows rather than get pulled by the allure of replacing jobs. The key point I realized is that businesses and customers still have the same problems as before, they haven’t changed. What’s changed is the types of problems that have been unlocked. The barrier to solving them has decreased. The best AI projects start with a problem a company is trying to solve, both internally and for their customers. The right way to think about this comes in four parts: Ideate, Collaborate, Measure and Govern
Case Study: Mike Knoop @ ZapierA year ago, Mike Knoop, Co-Founder & Head Of AI @ Zapier, gave up his Head Of Products title to go all in on AI - I interviewed him on his approach.
He mentions:
“If I was giving guidance for companies who're just starting to think about how to use [AI]...what you should do is
think about all of the hard problems your organization has encountered that you haven't been able to solve yet.
Maybe ones you've tried to solve but were intractable for whatever reason - Is it too expensive to hire contractors or you couldn't get an engineering grip on the problem itself.
I recommend at least revisiting every single one of those problems now with language models in hand and just see, ‘Can we get a new handle on that problem as a result of having language models?’ As you’re thinking about what to do next with AI, don’t look far. Examine your business and customers through a problem-first lens rather than, “Where can I apply AI?” In case you missed it
Greg Kamradt Twitter / LinkedIn / Youtube / Work With Me |
Greg's Updates & News
AI, Business, and Personal Milestones
Sully Omar Interview 2 years of building with LLMs in 35 minutes Welcome to the 100 people who have joined us since last week! If you aren’t subscribed, join 9,675 AI folks. View this post online. Subscribe Now Sully, CEO Of Otto (Agents in your spreadsheets) came on my new series AI Show & Tell I reached out to him because you can tell he feels the AI. His experience is not only practical, it's battle tested. Sully's literally built a product of autonomous async agents that do research for...
Joining ARC Prize How the cofounder of Zapier recruited me to run a $1M AI competition Welcome to the 2,450 people who have joined us since last post! If you aren’t subscribed, join 9,619 AI folks. View this post online. Subscribe Now "We gotta blow this up." That's what Mike Knoop (co-founder of Zapier) says to me in early 2024. "ARC-AGI, we gotta make it huge. It's too important." "Wait, ARC? What are you talking about?" I quickly reply. "It's the most important benchmark and unsolved...
Building a business around a commodity OpenAI's models are a commodity, now what? Welcome to the 296 people who have joined us since last week! If you aren’t subscribed, join 3,939 AI folks. View this post online. Subscribe Now Large Language Models are becoming a commodity. We all know it. So if you’re a foundational model company, what do you do? You build a defensible business around your model. You build your moat. Google famously said they have no moat, “and neither does OpenAI.” But...