
Lightfield: How Keith Peiris Shut Down a 20M-User Product, Cut 70 People to 7, and Raised a $47M Series A
November 16, 2026
TL;DR: Lightfield is an AI-native CRM built as a system of record for people and AI agents. It was built by Keith Peiris (Co-Founder and CEO) and Henri Liriani, the founders of the AI presentation company Tome. Lightfield announced a $47M Series A led by Andreessen Horowitz (a16z) in September 2026, and said more than 5,000 companies had signed up. As he shared on the show, Tome had about 20 million users and single-digit millions in ARR when the team turned the product off, cut the company from about 70 people to about 7, and started over. Peiris told the full story on The Product Market Fit Show, including how he sold to his first YC customers by giving the software away.
A note on sourcing. Anything labeled "as he shared on the show" comes from the interview and has not been independently confirmed. Funding and company figures labeled with a source and date were checked against Lightfield's own announcement and press coverage in November 2026.
What does Lightfield do?
Lightfield is a CRM designed for a world where AI agents do customer-facing work. In its Series A announcement (September 2026), the company describes it as a CRM built for AI agents and as a "business world model," a record that updates itself from every call, email and meeting. The idea is that agents can only do good work if they can see the full history of a customer, not just what a sales rep typed into a form.
Peiris explained the origin of the idea on the show. While Tome's team was trying to build tools for sales and marketing teams, they kept hitting the same wall: the data agents needed was not in one place.
"And it turns out that what you need is not in the traditional CRM to do great generative AI work. It's sort of spread everywhere. It's in your call recorder. It's in your email. It's in your Slack. It's in your data warehouse." — Keith Peiris, Lightfield
The data also disagreed with itself. As he put it, what the rep wrote in Salesforce did not match the meeting, the email or the warehouse, because each system used different entities. That pointed to a bigger problem than any single agent: a single source of truth on each customer, so that many agents do not drift out of alignment and create bad work.
Key stat: Lightfield said more than 5,000 companies had signed up as of its September 2026 announcement. The company did not say how many of those are paying customers.
Who founded Lightfield?
Lightfield was built by Keith Peiris, Co-Founder and CEO, and Henri Liriani, his co-founder, who also co-founded Tome (press coverage, 2023 to 2026). Peiris refers to him on the show as his co-founder and says Liriani was instrumental in the launch video. Tome began in 2020 and was an AI presentation product. In February 2023 it announced a $43M Series B led by Lightspeed Venture Partners, with Coatue and Greylock also participating (Lightspeed and TheWrap, 2023).
How much has Lightfield raised?
Here is what has been publicly announced, as of November 2026:
- Series A: $47M, announced September 9, 2026, led by Andreessen Horowitz (a16z) (Lightfield's own blog post). Press coverage of the round (Finsmes, Upstarts Media, September 2026) also lists Maverick Capital, Coatue, Audacious, Alumni Ventures, Greylock and Lightspeed as participants.
- Tome, the earlier company: a $43M Series B announced in February 2023, led by Lightspeed (TheWrap and Yahoo Finance, 2023).
Why did Keith Peiris shut down a product with 20 million users?
Tome had real scale. As Peiris shared on the show, it had around 20 million users and single-digit millions of ARR. Customer acquisition cost was very low because demand was organic. The problems sat underneath: inference costs were high, repeat usage was low, and revenue retention was very low. User growth kept going, but revenue was starting to soften.
The harder issue was emotional and strategic. He said the product was missing context and data structure, and that it would never be capable of making decks for the people who are best at making decks.
He also gave a plain test for why the product felt different from what he wanted to build.
"If I asked our customers, what would happen if we took this away? I think they would be like, yeah, be a little sad. I'd go back to Google Slides." — Keith Peiris, Lightfield
Peiris said a soft landing for the team was probably available while Tome was still adding a million or two million users a month. What changed his mind was a board member named Dan, who argued that technology super cycles do not come often and that taking the soft landing would leave the team stuck for years.
His own words on the decision to cut deep before he knew what to build:
"We took it very seriously. We needed to cut burn really aggressively. You know, we spoke to our board and we're like, we're going to turn off the old product. We are going to dramatically cut burn." — Keith Peiris, Lightfield
As he shared on the show, the company went from about 70 people to about 7. The board backed the move, and he said the message from the board was not to be half in, because a side project inside Tome was not going to find product-market fit again.
How did Lightfield find product-market fit?
The pivot was not instant. Peiris said the team first followed the people who used Tome the most, who turned out to be sales and marketing teams. They ran pilots at roughly 10 companies. Within a few weeks, every one of them lost interest in decks and asked for other things, such as account research, proposals, RFP responses and email writing. That was the first sign the presentation product could not be sold into companies.
He is candid that the middle of the pivot was a dark stretch. Go-to-market tooling was crowded and pilot conversion was low. The way out was to step back and ask what they wanted to build for the next decade.
His answer came in three parts: own proprietary data, take on a hard technical problem, and build something that touches the whole company.
"And actually, that difficulty made us excited because we were like, we'd rather tackle a difficult problem than tackle an easy problem and have like a thousand competitors." — Keith Peiris, Lightfield
The signal that the new idea mattered came quickly. The first version of Lightfield was built in about three months and offered to friends and an extended network. As he shared on the show, it was missing about 99% of the features of Salesforce and HubSpot, yet customers were in it every day.
Never miss a founder's PMF story
Subscribe to The PMF Show"Just to compare with Tome, with Tome, we had lots and lots of users and none of them cared enough about the product to give us feedback. Whereas with Lightfield, our slacks were lit up, you know, 24-7, 365, about stuff that could be better, stuff we need to fix, stuff that could be different." — Keith Peiris, Lightfield
Angry, detailed feedback from a small group was a stronger signal than 20 million quiet users.
How did Lightfield win its first customers?
Mid-market companies liked the idea but could not swap out their system of record. So Peiris went earlier, to YC companies, seed-stage companies and accelerator startups, where there was no incumbent to rip out. He said that sale came with a clear first problem to solve: pipeline generation, since a new company does not need a full system of record on day one.
He described the first step as a trust problem. Complex software is hard to evaluate, so buyers look for references, and the fastest way to get references is to remove the price.
"So I would say our first 10 YC customers were not even customers. We're like, please use our software. Here's a Slack with us. We're going to make it better with you. And we're not going to stop until you're happy." — Keith Peiris, Lightfield
When those founders were happy, they posted about it in chats and on Bookface, referrals followed, and the team began charging a little.
Outreach followed the same logic. He reached out on LinkedIn rather than email, and he never led with "buy my product." Instead he invited founders to dinners and talks. The team invited the CRO of HubSpot to speak at the office and brought in the author of Predictable Revenue, offering to buy the author's book for everyone who attended. At the end of each event he showed the product for two minutes and offered a longer demo at the back of the room.
"We sort of had this realization of cold should be your last resort." — Keith Peiris, Lightfield
As he shared on the show, each event ended with 10, 20 or 30 people asking to see more. His order of operations: reactivate everyone you have spoken to in the past two years, then mine your network and your investors' networks for introductions, then find people talking about the problem on social media, and only then go cold.
For more on how to land early customers, see our guides to founder-led sales and how to get your first 10 customers.
How did Lightfield make launch videos that reached a million views?
Pablo noted on the show that Lightfield had made several launch videos that passed a million views. Peiris said the goal was never a view count. The goal was for one narrow audience, a B2B pre-seed founder near product-market fit, to feel something. The story followed a founder whose customers loved his product but who had not figured out how to sell it. As Peiris shared on the show, the video took two months and cost in the low six figures. For launch day, he said the team asked about 500 people, including angel investors, customers and friends, to like, comment and reshare at the same time.
Key lessons from Keith's playbook
1. Look for care, not scale. Twenty million users were less valuable than a smaller group whose Slack channels were full of angry, detailed feedback. If nobody would miss your product, growth numbers are hiding the real answer. 2. Metrics rarely jump on their own. Peiris argued that a product inherits its retention from its core, and that moving net retention from 20% to 80% with the same product is unrealistic. If the core metrics are structurally weak, the lever is a different product. Our guide on when to pivot goes deeper on that call. 3. Cut the noise before you know the answer. Dropping from about 70 people to about 7 let a small team work on one thing. His rule: get everyone out of the room who does not want to build the new thing. 4. Give the product away to build references. The first 10 YC users were free, with a shared Slack and a promise to keep working until they were happy. Related reading: how to get design partners. 5. Pick a hard problem on purpose. He chose an agentic system of record because fewer teams would try it.
His closing advice to founders was personal rather than tactical:
"I think the most valuable thing I've learned in the past few years is you have to build a company, a product and service that you love and that moves you emotionally so that you can work on it every single waking hour." — Keith Peiris, Lightfield
For more, see startup pivot case studies and product-market fit for AI SaaS.
Listen to the full interview: He turned off a product with 20M users and went from 70 people to 7, then raised a $47M Series A.
FAQ: Lightfield
Q: What is Lightfield? A: Lightfield is an AI-native CRM that the company describes as a "business world model," a customer record that updates itself from calls, emails and meetings so that people and AI agents work from the same source of truth. It was built by the founders of Tome.
Q: Who is the CEO of Lightfield? A: Keith Peiris is Co-Founder and CEO. He co-founded the company with Henri Liriani, who also co-founded Tome.
Q: How much funding has Lightfield raised? A: Lightfield announced a $47M Series A led by Andreessen Horowitz in September 2026, as of November 2026. No valuation has been disclosed. The founders' earlier company, Tome, announced a $43M Series B in February 2023.
Q: Who are Lightfield's investors? A: a16z led the Series A. Press coverage in September 2026 also lists Maverick Capital, Coatue, Audacious, Alumni Ventures, Greylock and Lightspeed as participants.
Q: How many customers does Lightfield have? A: Lightfield said more than 5,000 companies had signed up as of September 2026. It did not say how many are paying customers. On the show, Peiris said his first 10 YC users started on free access before the team began to charge.
Sources: Listen to the Full Founder Story
- Keith Peiris, Lightfield — a second-time founder who turned off Tome, an AI presentation product with about 20 million users, cut the team from about 70 to about 7, and rebuilt the company as an AI-native CRM.
Last updated: November 2026
Want more founder stories like this?
Subscribe to The Product Market Fit Show for weekly episodes.
Subscribe Now