How to Build AI Startups That Cut Through the Noise — Insights from Varun Nair, CTO at That Works

In a world where anyone can launch an AI startup in a weekend, how do you actually build something useful? Varun Nair, Co-Founder and CTO at That Works, has a few sharp answers.

On The Lucent Perspective, Varun joins Rebecca Hastings to unpack what it really takes to go from “good idea” to a company that wins attention, traction, and time back for customers. With a background that spans scrappy startup exits and high-scale innovation at Meta, Varun’s perspective is uniquely relevant for today’s AI product builders—and for anyone trying to do more with less noise.

Listen to the episode HERE.

Topics Covered:

• Why “move fast” is essential for AI founders—but so is product quality

• What Varun learned going from startup founder to Meta engineering leader

• How to balance data-driven thinking with human intuition

• The myth of perfect positioning—and how to actually stand out to investors

• Why most team tools create noise, not clarity

• Frameworks for testing fast, shipping faster, and learning quickly

• What to do when traction is “mid” and VCs want top-tier growth

• How to know if your co-founder dynamic is your greatest strength

Episode Highlights:

00:00 – Intro to Varun Nair & That Works 

02:00 – From Two Big Ears to Meta: scaling product & mindset 

04:37 – Startup vs. Big Tech: how priorities shift 

08:09 – Lessons from Meta: data, positioning & product clarity 

11:54 – Startup #2: prioritization over hustle 

16:56 – The real problem with dashboards, meetings & Slack 

25:35 – Why shipping fast is still the smartest move 

30:34 – Launching AI products: speed vs. substance 

33:48 – What investors actually look for in AI startups

41:09 – Fundraising: qualifying VCs & avoiding pitch fatigue

48:43 – Varun’s advice for founders: detail, gut-feel & staying close to users

From Meta to Minimal: What Varun Learned About Velocity

After selling his first company (Two Big Ears) to Meta, Varun stepped into a very different product development world—one with billions of users and systems that already scaled. But rather than slowing him down, it gave him a clearer framework for how to move fast with purpose.

At Meta, Varun saw firsthand the power of data-driven iteration, but he also noticed how companies—even large ones—can drown in noise when systems aren’t aligned. That tension became a core insight for his next move.

“Startups need to move quickly, but they also need to be pointed. Speed only matters if it gets you closer to the right answer.”

The Idea Behind That Works

Most collaboration tools generate more work than they save. Varun’s latest venture, That Works, addresses that head-on by tackling what he calls the “information fragmentation” problem. With so many SaaS tools, Slack channels, and standups, it’s harder than ever to see what’s actually getting done.

That Works builds a live, AI-powered change log across a company’s workflows—giving leaders instant visibility into what’s happening, what’s stuck, and what matters most. It’s not just summaries; it’s contextual, actionable insight.

“We don’t treat all information equally. We rank and relate every update to help teams focus on what moves the needle.”

Why AI Startups Must Ship Faster (and Smarter)

Varun’s core thesis: in AI, velocity is everything—but polish still matters. “Done is better than perfect” applies, but only if the version you ship is good enough to generate signal. If the MVP is too clunky, users won’t stick around long enough to give you feedback.

So how do you find the right balance?

• Start with a question, not a product. What do you need to learn?

• Build the simplest thing that will get you that answer.

• Listen to behavior, not just interviews.

  
  

Even That Works’ first prototype wasn’t an app—it was a manual email workflow. The goal wasn’t to scale, but to test if the information was truly valuable.

Breaking Through with Investors

For AI founders, the bar is higher than ever. VCs aren’t just looking for “yet another wrapper” around a language model. What stands out today?

• A unique and opinionated thesis

• Proof of traction—whether usage, waitlist, or narrative momentum

• Realistic plans for distribution and retention

  
  

One tip Varun shared? Use ChatGPT to pitch your product. If the output sounds exactly like your deck, your positioning probably isn’t differentiated enough.

Advice for Founders

As a mentor and investor, Varun sees a pattern: founders slowly drift away from user insight and detail as they scale. His biggest recommendation?

“If you’re relying on other people to understand your users, stop doing that—or at least do it less.”

Stay close to the problem. Dive into the details. And when in doubt? Prioritize brutally.

Key Takeaways

• Speed + Direction: Don’t just build fast—build fast toward something valuable.

• Product > Pitch: In AI, it’s not enough to have a good idea. You need proof and polish.

• Information ≠ Insight: True AI leverage comes from structuring, ranking, and relating your data—not just summarizing it.

• Investors want clarity: Opinionated takes, early traction, and strong distribution ideas still win.

  
  

Follow Varun Nair on LinkedIn for weekly insights on AI, product strategy, and productivity systems. Explore That Works to see how they’re redefining the way teams manage information.

🎧 Listen to the full episode of The Lucent Perspective wherever you get your podcasts.