The AI-Native CRM

The podcast starts with Keith Peiris sharing his career transition from building Tome, a highly popular AI presentation app that quickly scaled to 25 million users before he decided to pivot. Keith reveals that the decision to pivot was driven by a lack of love for the product and the realization that it couldn't effectively support professional use cases, such as investment banking or consulting, due to technical constraints. This led to a strategic shift towards solving more complex problems within the CRM space, emphasizing the importance of building products that resonate deeply with customer needs.

Keith explains that Lightfield aims to create a 'business world model' by making sense of fragmented and conflicting data spread across various company systems like emails, calls, and CRMs. The team's approach is based on the insight that intelligence surpasses rigid schema, allowing a more flexible, context-aware understanding of company relationships and customer interactions. This new paradigm helps companies model their entire business reality in a way AI can reason about, ultimately enabling smarter decision-making and automation.

The discussion delves into the foundational design principles of Lightfield, notably the importance of modeling relationships via an activity log derived from various data sources. Keith highlights how early decisions, such as focusing on a relationship-centric model inspired by Facebook’s timeline, enable real-time, flexible updates and comprehensive context capturing. The architecture allows the CRM to incorporate unstructured data, infer causality, and adapt dynamically, making it significantly more intuitive and powerful than traditional schema-based systems.

Keith describes how Lightfield’s product development was driven by deep customer engagement, starting with top-tier startups and expanding into complex enterprise sectors like healthcare and life science. He emphasizes the value of understanding customer pain points and iterating rapidly based on feedback. Lightfield’s approach combines pragmatic product design with broad applicability—offering both out-of-the-box dashboards and natural language interaction—catering to diverse user preferences and workflows. Keith also emphasizes the importance of referenceability and trust in market expansion.

The conversation shifts to market entry strategies, contrasting greenfield (new markets) and brownfield (existing, entangled systems) approaches. Keith discusses how Lightfield’s flexible data modeling makes it suited for both: disrupting existing systems with superior tech or providing a fresh start for new companies. Initial pricing experiments ranged from free to consumption-based models, ultimately settling on a hybrid platform fee plus usage-based elements. The importance of outcome-based pricing is acknowledged, though current business depends on paying for active work.

Keith emphasizes the need for organizational agility in a fast-moving AI landscape. He advocates for a culture where everyone is a generalist, actively engaged in product and customer success, with continuous, dynamic planning. This approach contrasts sharply with traditional siloed teams, enabling rapid iteration, adaptation, and a focus on solving highest-impact problems. The company's operational practices are designed to foster speed, alignment, and responsiveness to customer and technological changes.

Looking ahead, Keith is excited about Lightfield becoming a 'crystal ball' for companies to make strategic, scenario-based decisions. He stresses the importance of staying ahead in a competitive, quickly evolving market, particularly preventing customer churn by providing an irreplaceable, intuitive product. Keith shares concerns about maintaining speed and innovation, fearing that slow dashboard development could hinder growth. He highlights the role of trust, references, and product simplicity in capturing market share, especially outside Silicon Valley where awareness and adoption are still growing.