Every CRM on the market has the same original sin: you have to fill it in.
You get home from a client meeting and you're supposed to log the notes, update the deal stage, record the next steps, and set the follow-up task. You don't. Nobody does. Then six months later you need the history of an account and half of it is in your email, half of it is in your head, and the CRM is full of ghost records no one has touched since February.
Lightfield launched on Product Hunt this month with a different premise entirely: the CRM fills itself in.
It's not a chatbot layer on top of a database. It's a schema-less AI system that reads your emails, call transcripts, and meeting recordings - then builds the customer record automatically, updates it continuously, and backfills your history. You don't configure it first. You connect your email and calendar, and it starts learning your customer relationships.
I've been watching this category long enough to be skeptical of that pitch. But the execution here is actually different from the "AI add-on" CRM category, so let me break down what it does and where it fits.
What Makes It Different
Every major CRM player has added AI features in the last two years. Salesforce has Einstein. HubSpot has Breeze. Most of these are bolt-ons: they summarize a record, suggest a follow-up, or draft a template. The underlying structure is still a conventional database that a human has to build and maintain.
Lightfield starts from a different place. The core concept is what they call "customer memory" - a continuously updated understanding of each relationship, built from passive data capture.
Here's how it works in practice:
You connect your email and calendar. Lightfield reads your thread history with every contact. It doesn't just index subjects - it reads the content, extracts relationship context, and starts building a picture of where each conversation stands.
You set up call recording. Lightfield transcribes and analyzes calls. After a sales call, it auto-drafts a follow-up email and suggests next steps. You review and send - or don't. It updates the record either way.
The CRM fills in. Fields appear as Lightfield encounters data that warrants them. No preset schema. If a deal concept comes up in 40 of your conversations but nowhere else, Lightfield creates that field for those 40 records and leaves the rest alone.
You ask questions in plain language. "Which clients mentioned pricing concerns in Q1?" "What did we promise TechCorp about delivery timing?" The answers come back with citations - linked to the actual emails or call moments that surfaced them.
That last part is worth dwelling on. The differentiator isn't just automation - it's that you can verify every answer. It doesn't hallucinate a relationship history. It shows you the source.
What It Costs
Lightfield runs on a freemium model with paid tiers:
Startup: $59/user/month - Free trial, call recording and transcription, automated data enrichment, unlimited agent queries and actions, configurable data model.
Pro: $149/user/month (billed annually) - Higher record and workflow limits, group configuration and permissioning, white-glove migration, dedicated customer success manager.
That's not cheap for a solo operator. For a small B2B team doing significant volume - consultants, agencies, service businesses with long sales cycles - the math is different. If Lightfield saves two hours a week in manual CRM work, it pays for itself at Startup pricing in the first month.
The competitor comparison: HubSpot's Sales Hub Professional runs $100/user/month. Salesforce Starter runs $25/user/month but the meaningful tier is $80. Attio, which is the other AI-native CRM people compare Lightfield to, starts at $34/user/month. Lightfield is more expensive than the budget options. The pitch is that it replaces work, not just software.
Who Should Try It (and Who Shouldn't)
Good fit:
- Small B2B teams where relationship history matters and deal cycles are long
- Consultants, agencies, and service firms where every client has a narrative
- Any team where CRM maintenance is the bottleneck, not CRM adoption
- Technical founders who are already connecting tools via API and want the CRM to participate in their AI workflow
Not the right fit:
- High-volume transactional businesses where speed of data entry is more important than depth of relationship context
- Teams who need HubSpot or Salesforce integrations for downstream reporting - Lightfield is still building that bridge
- Businesses where client conversations happen outside email and scheduled calls (in-person only, cash transactions, etc.)
- Anyone on a tight budget who hasn't validated their CRM problem first
The Honest Take
The "CRM that builds itself" pitch is not new. You can find five startups from 2022 that had the same tagline. Most of them failed because the AI wasn't good enough to extract reliable context from messy email threads, and the database they built was too confidently wrong to be useful.
What's different in 2026 is the underlying model capability. The language models that Lightfield runs on are genuinely better at reading context, disambiguating relationship history, and flagging uncertainty than anything available two years ago. That changes the viability of the premise.
The schema-less approach is also a real differentiator. Most AI CRM tools still ask you to define your pipeline stages and custom fields before you can use them. That setup friction is where adoption dies. Lightfield's model - let the data define the structure - is closer to how a good sales person actually manages relationships in their head.
The product is early. There will be things it gets wrong. The call recording setup requires desktop, and mobile capture is still listed as "coming soon." The migration story is CSV-based, which means it's workable but not seamless if you're coming from a well-populated HubSpot or Salesforce.
But the core concept is sound, the execution is further along than the comparable products that came before it, and the free trial is real. If you're running a service business and your CRM is a ghost town you're embarrassed by, this is worth 20 minutes of your time to set up and see.
Lightfield is at lightfield.app. Product Hunt listing: producthunt.com/products/lightfield.
Sources: Lightfield product documentation (lightfield.app); SaaStr AI App of the Week feature on Lightfield (saastr.com); Product Hunt April 2026 leaderboard (producthunt.com); Comparateur IA Lightfield review (comparateur-ia.com)
Danny Kowalski covers tools and software for small business at The Useful Daily. Published at theusefuldaily.com.