Ask any sales team which software they hate most, and CRM is almost always the answer. Companies spend hundreds of billions on it, yet it remains the most despised tool in the stack. The reason is simple: someone has to manually enter all the data — and that's been true for 40 years.

Keith Peiris and Henri Liriani built Tome to 25 million users, then voluntarily shut it down. After a year in stealth, they launched Lightfield — a CRM where AI extracts data from conversations instead of waiting for humans to type it in.

30-second summary
Tome shutdown 1-year stealth AI converts conversations to CRM 2,500 companies in 3 months Taking on Salesforce head-on

Why did they walk away from a winning app?

Tome raised $81M from Coatue, Greylock, and Lightspeed, hit a $300M valuation, and made the Forbes AI 50. The kind of thing most founders would keep scaling indefinitely.

But while working with customers, the founders spotted a deeper problem: the real issue was a lack of customer intelligence — sales teams simply didn't know their customers. The root cause? CRM data was always incomplete, because nobody was entering it.

Keith Peiris grew Instagram DM to 500 million MAU at Meta. Henri Liriani was a Facebook Messenger engineer. They rebuilt CRM from scratch around one premise: "Eliminate human data entry from CRM entirely."

25M
Tome users they walked away from
$81M
Funding they left behind
2,500+
Companies using Lightfield in 3 months

Lightfield's core architecture is schema-less. Traditional CRMs predefine columns — Name, Company, Deal Size — and wait for humans to fill them in. Lightfield flips this: it captures all conversations, emails, and call recordings first, then derives structure from those. Fields are created automatically by AI and applied retroactively to historical records. With 95%+ recall accuracy at scale.

"The most important yet most hated category in SaaS" — that's CRM. In an $80B+ market, nobody has fundamentally solved the manual entry problem in 40 years.

— SaaStr, 2026

What actually changes?

Skip the tech jargon. What's the biggest barrier to data quality in any Salesforce or HubSpot deployment? Sales reps don't log calls. They're too busy. It's too tedious. CRM admins spend their energy begging people to fill things in.

Legacy CRM (Salesforce / HubSpot) Lightfield
Data entry Manual, human-driven AI auto-extracts from conversations
Data model Fixed schema (predefined columns) Schema-less (auto-generated from data)
Call recording Requires separate tool (Gong, etc.) Built-in call intelligence
Meeting prep Manual research and compilation AI auto-generates briefings
Agent workflows Limited, add-on cost Runs from a single prompt
Startup pricing Complex tiers + onboarding fees $36/user/month

2,500 companies adopted it in the first three months, including 100+ YC startups. One user revived 40+ stalled opportunities in a single 2-hour session. Power users interact with the platform 400+ times per week — that's not a tool, that's a teammate.

Salesforce and HubSpot can't easily replicate this. Not because of tech limitations. Billions in revenue sit on the assumption that humans will enter data. Changing that assumption at the architecture level is structurally impossible for them.

Why YC founders are flocking to Lightfield

Most recent YC startups aren't using Salesforce or HubSpot. In the founder-led sales phase, an AI teammate with perfect customer memory beats a bloated platform. That's the exact wedge Lightfield is driving into.

How to get started

  1. Connect your email (5 minutes)
    Connect Gmail or Outlook and CRM records start generating from existing conversations automatically. No configuration needed.
  2. Connect calls and meetings
    Enable Google Meet, Zoom, or Teams. Meetings are auto-recorded, summarized, and linked to the right contacts. No Gong required.
  3. Launch an AI agent
    Run outbound agents with one prompt: "Find 50 companies matching these criteria and draft personalized emails for each."
  4. Query your pipeline in plain English
    Ask: "Show me all opportunities with no follow-up in the past two weeks." No SQL, no filter setup.
  5. Choose your plan
    Startup plan at $36/user/month (up to 10,000 records). Pro plan at $99/user/month (up to 50,000 records, dedicated CSM included).

Want to go deeper?

SaaStr: Lightfield AI App of the Week The original deep-dive — founders' story, product analysis, early metrics. saastr.com

Lightfield vs HubSpot comparison Feature-by-feature breakdown from Lightfield's own team. lightfield.app

Folk CRM's Lightfield review An honest take from a competitor — great for understanding tradeoffs. folk.app

SaaStr: Which CRM in 2026/2027? The broader agentic CRM landscape, Lightfield included. saastr.com

Lightfield official site Free to start — connect your email and see live records in minutes. lightfield.app

VentureBeat: Why Tome's founders pivoted Full backstory on walking away from 25M users. venturebeat.com