AI-Driven CRMs: The Future of Forecasting and Revenue Modeling

Sep 21, 2026 · Topline
🎧 PodShort 33 min squeezed to 3 AI SprinklerAS Sales Tech New
Episode artwork
Keith Paris
Co-founder and CEO at Lightfield
Sam Jacobs
Co-host, CEO and Founder at Pavilion
Aasad Zaman
CEO at Sales Talent Agency
AJ Bruno
CEO at QuotaPath
Topline
33 min squeezed to 3
Full episode from Topline
Quotable Moments

Forecasting now, prediction now is really hard. People are putting out new products every couple of weeks. Model costs are dropping. Segments are being written and rewritten.

Keith ParisCo-founder and CEO at Lightfield
01:45

I see that deterministic problem as the hardest problem. The models are always going to be non-deterministic.

Keith ParisCo-founder and CEO at Lightfield
03:20

I think the honest answer is we started small. We built the system for two years and when we launched it, it was really only ready for a seed Series A company.

Keith ParisCo-founder and CEO at Lightfield
06:50
Key Insights
  • Forecasting and prediction are increasingly difficult due to rapid product development, dropping model costs, and constant rewriting of segments, making traditional pipeline analysis insufficient.
  • The core problem with deterministic forecasting in an AI-driven world is that models will always be non-deterministic, requiring human oversight and a shift in how we approach data and analysis.
  • A key insight for Lightfield was realizing that to truly do forecasting in today's world, you need to build something larger than a traditional CRM, incorporating signals and understanding the broader business context.
  • The belief that modeling your entire business makes every part of your revenue funnel better is a strong motivator for adopting a comprehensive CRM solution like Lightfield.
  • The ability to run scenario planning and answer strategic questions like 'Should we build this product feature next quarter?' or 'How should we change our sales process?' is a powerful driver for adopting advanced CRM with robust modeling capabilities.
  • The idea of AI telling sales reps exactly what to say or do next is a misunderstanding of the craft of sales, especially for complex products, and will likely lead to failure.
  • The best-of-breed approach in high-growth tech companies means wanting the best solution for each part of the revenue operation, but the definition of 'best' is shifting towards interconnectedness and shared data rather than isolated features.
  • The increasing ease of building software means that companies will build more in-house, leaving less room for traditional partnerships and making it harder for ecosystem players to thrive.
Metrics Mentioned
  • $47 million (Lightfield's Series A funding round, led by Andreessen Horowitz.)
  • 5,000+ companies (Number of companies already using Lightfield's CRM.)
  • 500 million monthly users (Instagram Direct's user base when Keith Paris grew it at Meta.)
  • 10 reps going on 30 (The ideal customer size for Lightfield currently.)
  • 200-person company (Lightfield's deal size for a company of this size looks like a Salesforce deal for 5,000 people.)
  • 5% error rate (A common error rate for top AI models, which is unacceptable for sales forecasting.)

RevBots.ai View:

  • AI Sprinkler stage companies bolt on AI features without transforming core processes.
  • ARM stage companies integrate AI deeply, modeling entire businesses for unified insights.
  • Tab Hopper and SaaS Hoarder stages struggle with traditional forecasting and isolated tools.
  • ARM adoption requires a shift from deterministic to probabilistic forecasting models.
🎧Full Episode:Topline →