Building an AI-Native CRM from Scratch
How Lightfield is reimagining customer relationship management with contextual AI
1From presentations to sales workflows
Keith Peiris, previously founder of presentation tool Tome, pivoted to CRM after realizing AI couldn't sufficiently understand presenter-audience relationships. His team followed users from AI presentations into sales processes, uncovering a deeper challenge: making sense of fragmented customer data across emails, calls, and existing CRM systems. This insight led to Lightfield's focus on solving data coherence in sales workflows.
2The limitations of traditional CRM
Traditional CRMs rely on rigid data schemas that struggle with real-world business complexity. Lightfield identified that critical customer context gets lost across disconnected systems, forcing sales teams to manually piece together information. This fragmentation creates inefficiencies and blind spots in customer understanding, which AI-native approaches aim to solve through more flexible data modeling.
3Business world model concept
Lightfield's solution centers on a 'business world model' where AI understands entities and relationships holistically. Instead of forcing data into predefined fields, the system learns organizational context to connect dots across communication channels. This approach allows the CRM to reason about company-customer relationships dynamically, adapting as new information emerges from various touchpoints.
4AI's contextual advantage
Where traditional software requires explicit data entry, Lightfield's AI builds understanding from existing communications. The system analyzes emails, call transcripts, and meeting notes to maintain updated relationship context. This reduces manual data entry while providing sales teams with continuously refreshed customer intelligence drawn from actual interactions.
5Greenfield vs brownfield adoption
The team faces distinct challenges selling to companies with versus without existing CRM systems. Greenfield adopters can implement AI-native workflows from scratch, while brownfield clients require integration with legacy systems. Lightfield must demonstrate enough value to justify displacing entrenched solutions, particularly for sales teams resistant to changing established processes.
6Pricing in the AI era
The company rethinks traditional SaaS pricing models to align with AI's value proposition. Rather than charging per seat or feature tier, Lightfield considers performance-based models where pricing reflects the system's impact on sales productivity. This matches how AI delivers value through outcomes rather than software access alone.
7Generalist team structure
Lightfield operates with a deliberately generalist workforce where employees span multiple functions. This structure aims to maintain flexibility during the company's rapid evolution, avoiding silos that might constrain product direction. The approach reflects lessons from Peiris's previous startup experience and the need for adaptability in AI product development.
8Strategic pivot lessons
Peiris's decision to walk away from Tome's 25 million users demonstrates the discipline required when core technology can't fulfill long-term vision. The pivot to CRM required identifying a problem where AI could deliver transformative value rather than incremental improvement. Lightfield's trajectory shows how founder experience shapes recognition of truly disruptive opportunities in enterprise software.
Original article: a16z Podcast — Building an AI-Native CRM from Scratch