Nexosis

Ryan C.B. and co-founder Jason Montgomery started their journey into machine learning around 2012 with cybersecurity backgrounds. They initially envisioned Nexosis as a predictive security platform that would analyze firewall data and other security events to predict potential breaches. The pair officially founded the company in 2015 and spent years building proprietary AI/ML capabilities entirely in-house, serving large retailers with time series forecasting solutions.

Nexosis operated in stealth mode for years, developing deep expertise in machine learning models that could handle complex forecasting problems. They built everything internally to own all their intellectual property. By early in the company's lifecycle, Ryan and the team had accumulated years of transaction-level data, weather patterns, social media sentiment, and marketing calendars that fed into increasingly accurate predictions.

The breakthrough came when the team decided to open up their platform to developers. Rather than keeping their proprietary solution locked behind closed doors for enterprise retail clients, they recognized the value of democratizing access. They began offering their API to developers at a minimal cost: just $0.10 per 1,000 predictions. This freemium approach allowed developers to build prototypes and proof-of-concepts without friction. Once these developers demonstrated value to their employers—say, building a forecasting tool for a restaurant chain—Nexosis would engage directly with the enterprise to structure a consumption-based plan.

The developer-first, bottom-up motion proved remarkably effective. Rather than chasing enterprise deals directly, Nexosis leveraged internal champions—engineers already at large companies who had built working prototypes on the platform. This gave them credibility and a clear picture of consumption patterns before negotiating enterprise terms. The team observed that early developer users typically created one or two projects but then invited colleagues, expanding adoption organically within organizations. By the time they went into a formal sales conversation, the enterprise already had skin in the game.

The company modeled itself after Twilio's playbook: low-friction developer pricing subsidized by VC capital, with revenue ultimately driven by enterprise consumption at scale. Rather than obsessing over churn or MRR (metrics that don't fit consumption-based APIs), they focused relentlessly on net expansion rate—whether consumption grew month-over-month.

As of the interview, Nexosis had raised just under $7 million in venture funding and employed 15 people based in Columbus, Ohio. They were serving approximately 100 enterprise customers across a much larger developer base, with typical enterprise plans starting around $10,000 per month and scaling based on prediction volume and data scope. The company was executing at an impressive clip: over 1 million predictions per month with month-over-month doubling growth. They were preparing for a public API launch on June 30, moving beyond their closed "friends and family" phase. Use cases ranged from restaurant chains optimizing inventory and labor to sports bars predicting foot traffic around games to mosquito activity forecasting for malaria prevention—proof that the platform solved a fundamental problem across industries.