A 3-Part Strategy for Keeping Up with AI

Every single day, my social media feeds are flooded with announcements of new AI breakthroughs that are supposedly "changing everything."

For a while, I found myself in a constant state of AI FOMO — afraid I'd miss some critical development that would leave my company behind. Maybe you've felt this too.

But here's the thing: while trying to stay on top of everything in AI is exhausting and ultimately impossible, completely ignoring these developments is equally dangerous.

So, how do we strike the right balance?

In today's newsletter, I want to share a simple 3-part strategy I've developed for myself and the teams I work with to stay effectively informed without getting overwhelmed.

—Dharmesh

The AI Announcement Overload

Let's preface this 3-part strategy with an observation from the world of AI that I’m sure many of you can relate to:

For varying values of "X" and varying values of "everything." Putting aside the hyperbole, I like to think of this as:

Pick whichever form pleases you. For this exercise, it doesn't matter.

Here are the two interesting things:

1. The frequency with which we get new values of [X]

2. The variety of sources from where [X] originates

Just look at some [X] values we've seen in recent weeks:

• Inference-time compute

• Reasoning models

• DeepSeek R1

• Hybrid reasoning models (Claude 3.7 by Anthropic)

• Operator (OpenAI)

• Deep Research (Google, OpenAI, Perplexity)

• Model Context Protocol (Anthropic)

• Manus Agent (Manus AI)

• Mistral OCR (Mistral AI)

And those are just the ones I could think of off the top of my head, with zero cups of coffee.

What's truly exciting is how the innovation landscape has evolved. For a while, advances were primarily "bigger, better LLMs" from a handful of well-funded companies (because building state-of-the-art models required prohibitive resources).

Now, we're seeing breakthroughs across multiple dimensions from companies of all sizes. This diversification is incredibly healthy for the ecosystem, but it creates a new challenge: how can any busy professional possibly keep up?

That's where my 3-part strategy comes in.

A 3-Part Strategy for Staying Effectively Informed

So what should you do? After much trial and error, I've settled on a 3-part strategy that works for me and might work for you too:

1. Use AI Daily — Don't Just Study It

The most important thing isn't to track every development, but to actually use AI in your day-to-day work. It's plenty good enough now.

I've found that hands-on experience teaches me more about AI's capabilities and limitations than reading a hundred research papers or product announcements. When you're using these tools regularly, you develop an intuitive sense of what's possible.

For example, here are some practical ways to incorporate AI into your routine with real AI agents that exist today:

1. Company Research Agent: Creates comprehensive company reports in seconds instead of hours.

2. Prospect Extractor: Identifies and verifies potential leads from conversations and existing data.

3. Ideal Customer Profile Builder: Analyzes successful customers to reveal who actually buys your product and why.

But you could go any route, it doesn’t need to be agent.ai. Add an AI notetaker to your meetings, have ChatGPT summarize a long email you’ve been dreading to read, the list goes on…

And by the way, this isn't just about productivity gains — it's about developing AI fluency through practice. The more you use these tools, the better you'll understand what's possible and where the opportunities lie in your specific business.

2. Schedule Quarterly AI Opportunity Reviews

Every now and then (maybe once a quarter), gather your team and take a list of your top 3 challenges that are hindering your growth. Then, see if something in the AI landscape has changed that makes these problems now addressable.

It may not solve your challenges completely, but if they're significant enough problems, even partial solutions can yield substantial benefits.

Here's how I structure these reviews:

1. Identify obstacles: What are the 3 biggest problems slowing our growth?

2. Research recent developments: Has anything emerged in the past quarter that might help?

3. Run small experiments: Test promising approaches with minimal investment

4. Scale what works: Double down on successful experiments

The key is making this review a regular, scheduled event, not just something you do when you happen to read about a breakthrough.

3. Try New Tools with Specific Problems in Mind

When you hear about a new AI tool or capability, don't just play with it aimlessly. Instead, try to approach it with a specific problem you're trying to solve.

This problem-first approach accomplishes three things:

• It gives you a concrete way to evaluate the tool's usefulness

• It keeps your exploration focused and productive

• It connects abstract capabilities to practical business outcomes

For example, when Claude 3.7 Sonnet was released, I didn't just test its general capabilities. I tried using it specifically for our challenge of improving code review efficiency. By focusing on this specific use case, I quickly discovered its hybrid reasoning model was remarkably effective at understanding complex codebases.

This approach turns AI exploration from an abstract exercise into practical problem-solving. You'll build a mental library of which tools excel at which types of problems, making future decisions faster and more informed.

The Big Mistake to Avoid (And How to Find Your Balance)

What you shouldn't do is decide at any point that AI is not good enough to tackle [Z] problem and then forget about it. You may be right at that time, but there are new [X]s coming out all the time, and the answer might change faster than you expect.

I've seen this pattern repeatedly:

• Companies evaluate AI for a specific use case

• They determine it's not quite ready

• They don't revisit the decision for a year or more

• By then, they've fallen behind competitors who were more persistent

Instead, keep a running list of "not yet, but soon" opportunities. These are problems where AI shows promise but isn't quite there yet. Review this list regularly as part of your quarterly process.

The right balance of attention to AI developments will differ based on your industry, company size, and competitive landscape. A startup in an AI-adjacent field will need to be more vigilant than a well-established company in a less tech-dependent industry.

But regardless of where you sit, completely ignoring these developments is increasingly risky. The question isn't whether AI will impact your business, but when and how significantly.

This strategy won't make you an AI expert, but it will help ensure you don't miss opportunities that could meaningfully change your business trajectory.

Hope it’s useful!

—Dharmesh (@dharmesh)