Ep. 199 - What AI Taught One Founder About the Future of SaaS

In this episode of SaaS Backwards, founder‑CEO Ivan Lee explains how Datasore evolved from a data‑annotation platform into a provider of private, secure large‑language models for regulated industries such as healthcare, finance, legal and government. He stresses that public models like ChatGPT only see about 2 % of an organization’s knowledge, leaving the remaining 98 %—contracts, internal wikis, proprietary workflows—unavailable for AI assistance. By deploying open‑weight models inside a client’s own infrastructure, Datasore guarantees that sensitive data never leaves the firewall, preserving compliance and competitive advantage.

The launch of ChatGPT in 2022 forced Datasore into a survival mode: existing customers paused projects, churn accelerated, and the company’s SaaS‑licensing playbook collapsed. Lee discovered that enterprise buyers were unwilling to fund a new, high‑ticket LLM platform when internal bandwidth and political capital were scarce. The turning point came when the team abandoned the traditional subscription model, embraced open‑source tooling, and began building custom end‑to‑end solutions that could be sold at three‑to‑five times the platform cost. This pivot restored cash flow and validated a “job‑to‑be‑done” approach.

Looking ahead, Lee notes that only about 15 % of enterprises use AI weekly, but joint ventures between OpenAI, Anthropic and management consultants signal a forthcoming wave of tailored deployments. The next phase will shift from opt‑in chatbots to opt‑out AI agents that operate continuously, surfacing insights without user prompting. For SaaS founders, the lesson is clear: prioritize data sovereignty, adopt open‑source foundations, and design solutions that embed directly into customers’ workflows. Companies that can deliver private, compliant AI agents are poised to capture the bulk of enterprise spend as adoption accelerates toward 2025.