Dear Readers,

Before we get to today's exclusive interview with the Lightfield CEO, a quick announcement:

We were kindly invited by NVIDIA to report live and in person from GTC. This was a first for us, and we are incredibly grateful that NVIDIA made this possible. Superintelligence started as a small newsletter, and since then we have not only grown but also regularly offer exclusive interviews with CEOs, and we recently even had the opportunity to interview NVIDIA's VP of Research, Ming-Yu, live on stream.

NVIDIA GTC is NVIDIA’s flagship annual conference where developers, researchers, and industry leaders gather to explore the latest breakthroughs in AI, accelerated computing, robotics, and data center technologies.

Therefore, it is one of (if not the) most important event(s) of the year, where the future is presented.

GTC 2026 (March 16–19 in San Jose) will spotlight NVIDIA’s next wave of AI infrastructure: “AI factories,” high-throughput inference, and the shift toward agentic AI (models that plan and act across tools/workflows), alongside major updates across accelerated computing. Expect a heavy focus on physical AI too - robotics, simulation/digital twins, and real-world autonomy - plus deep technical tracks and hands-on training labs that turn those platform announcements into deployable stacks.

In short: we'll be reporting live from the GTC. If you can't attend in person, you can easily join the virtual sessions for free from home using the following link: https://nvda.ws/45LRvQY

We're already incredibly excited and eager – and I hope you are too! March is going to be amazing!

Today, we have something very special: an interview with Lightfield CEO and co-founder Keith Peiris!

Lightfield is an AI-native CRM built around “complete customer memory”: it updates itself after every meeting, keeps your customer truth in one place, and helps drive the next steps automatically.

Before Lightfield, Keith co-founded and led Tome, spent time as an Entrepreneur-in-Residence at Greylock, and now he’s using that builder mindset to redesign CRM for the AI era.

All the best,

Kim Isenberg

Google has launched Gemini 3.1 Pro, a major leap in reasoning power designed for complex, high-stakes tasks across research, design, and engineering. Scoring 77.1% on the ARC-AGI-2 benchmark - more than double its predecessor - it enables advanced capabilities like code-based animations, real-time aerospace dashboards, immersive 3D simulations, and creative web design from simple prompts. Now rolling out across the Gemini API, Vertex AI, Gemini app, and NotebookLM, this preview signals a big push toward smarter, more agentic AI workflows for developers and enterprises alike.

Kim: Today we're talking about Lightfield, an AI-native CRM solving a classic problem: customer interactions are scattered across messy calls, emails, docs, and chats. Lightfield turns that into a clean, shared customer memory to actually do the follow-up work. I'm joined by CEO and co-founder Keith to discuss their startup story and why CRM needs an AI-first foundation. Welcome, Keith.

Keith: Thank you, I'm excited to be here.

Kim: Take us back to the beginning. What was the spark that led you to start Lightfield?

Keith: Before Lightfield, we built an AI presentations product called Tome. It gained millions of users, but we couldn't actually generate the high-quality presentations people wanted for work. To fix this, we asked companies to let us connect directly to their data—Salesforce, call recorders, customer tickets—to see if real business context could beat basic prompt engineering.

When we got the data, two things happened. First, we realized conventional databases lack the context an AI needs to understand why things happen. Second, this information was stored everywhere in a complex, disorganized mess. It was like handing an assistant 2,000 pages of unstructured data and telling them to figure it out. We realized this was a much bigger idea than a presentations tool: we needed to create a structured "context bucket" that an AI agent could actually drink from to do great work.

Kim: Could you summarize in one sentence why your product has to exist?

Keith: We want intelligent models to do work for us, but if you treat them like humans, you can't just hand them a crude database or a mountain of unstructured data. Models need a cohesive story to understand context, and it turns out the best narrative structure for that is time; by reimagining the CRM as a chronological story of your business, we can finally get AI models to do great work for you.

Kim: You've called CRM "the most complex and lowest satisfaction piece of software on Earth." What is fundamentally broken about legacy CRMs, and why hasn't anyone fixed it?

Keith: Legacy CRMs originated from the era of transferring physical index cards into a digital database. Over time, humans were asked to manually update these databases—not to serve customers better, but just so management could track metrics. That’s why people hate them.

Today, things have changed. First, customer interactions are natively digital, so we can capture them automatically. Second, an AI-native CRM acts as a world model of your business—an oracle you can ask, "How would customers respond if I raised prices or shipped this feature?"

As for why it hasn't been fixed: migrating CRMs used to be a nightmare requiring expensive consultants to map data and rebuild archaic workflows. Now, LLMs are incredible at data migration and code generation. What used to take months of painful mapping can now be executed by AI in the background, making it possible to switch systems in a day.

Kim: When did you doubt this idea the most, and when did you realize you had real traction?

Keith: The doubt came early when investors told us we were crazy. They believed new CRMs inevitably die because they can't build integrations or match legacy features fast enough. We also had to convince people to abandon their old, familiar workflows.

But the traction became undeniable when a 200-person company migrated to Lightfield. Their CEO told us, "You only have a fraction of the legacy features, but we see that you can catch up instantly using AI code generation. We're here for the ride." People are highly productive with tools like Claude and Cursor now; they actively want an AI-native way to work.

Kim: Let's get a bit nerdy. Lightfield stores data in its raw form and AI extracts structure on demand. Why does this beat the rigid schema models used since the 90s?

Keith: Legacy relational databases require you to manually fill in rigid fields. The downside is that crucial context slips through the cracks because business nuances rarely fit into neat boxes. Conversely, you can't just use a pure unstructured data lake because you'd need infinite compute to find anything.

Our approach is semi-structured. We build a world model with core objects—like your businesses and customers—and maintain a simple, chronological timeline of everything that happens to them. Once the AI reads that timeline story, it seamlessly connects the unstructured context with the structured data it needs to take action.

Kim: Over 1000 startups are already using Lightfield daily. What's one user story that best captures why teams are making the switch?

Keith: Our growth has been heavily driven by word-of-mouth because users spend more time in our product daily than they do on Instagram.

My favorite example is seeing small, ambitious teams serve highly specific niches. We have a tiny team building agricultural software for Canadian farmers. Because they are a five-person company, the people building the product, doing marketing, and handling support are the same individuals. With Lightfield, they can just ask their CRM's memory, "What should our marketing say based on how our customers talk?" or "What should our roadmap look like based on recent feedback?" The AI instantly drafts the marketing and helps prioritize the roadmap. It empowers a tiny team to build and operate faster than a massive corporate enterprise ever could.

Kim: Keith, it was a pleasure talking to you. Thank you, and best of luck with continuing the project.

Keith: Thank you. It was great to be here.

Why it matters: Lightfield fundamentally transforms the CRM from a universally hated, manual data-entry tool into an intelligent, automated "world model" of a business. This shift matters because it finally gives AI agents the rich narrative context they need to execute real tasks autonomously, empowering small startup teams to operate with the speed and capability of massive enterprises.

Sources:
🔗 lightfield.app

A modern intranet like Haystack streamlines workplace operations by centralizing knowledge, communication, and resources.

Employees will no longer waste time hunting through email chains or scattered folders—they can find what they need in seconds.

With customizable templates, clear layouts, and multimedia capabilities, teams can create and share content that is easy to read, navigate, and reference. Haystack turns your intranet into an interactive, engaging resource hub that supports collaboration and knowledge retention.

Upgrading your intranet boosts efficiency across departments, reduces duplicated work, and ensures consistent, accurate information is accessible to everyone. Employees stay informed, aligned, and empowered, while leadership gains visibility into engagement and usage.

Haystack transforms your intranet from a static repository into a dynamic platform that drives productivity, connection, and culture.

See Haystack In Action

If you have enough money that you think about buckets for your capital…

Ever invest in something you know will have low returns—just for the sake of diversifying?

CDs… Bonds… REITs… :(

Sure, these “boring” investments have some merits. But you probably overlooked one historically exclusive asset class:

It’s been famously leveraged by billionaires like Bezos and Gates, but just never been widely accessible until now.

It outpaced the S&P 500 (!) overall WITH low correlation to stocks, 1995 to 2025.*

It’s not private equity or real estate. Surprisingly, it’s postwar and contemporary art.

And since 2019, over 70,000 people have started investing in SHARES of artworks featuring legends like Banksy, Basquiat, and Picasso through a platform called Masterworks.

My subscribers can SKIP their waitlist and invest in blue-chip art.

Just click this link

_{Investing involves risk. Past performance not indicative of future returns. Reg A disclosures at masterworks.com/cd}