Nathan C. Frey, PhD - CTO at Coefficient Bio

Nathan C. Frey, PhD

CTO & Co-Founder at Coefficient Bio. Building for the future in AI-native biotech. Ex-Prescient Design • Genentech.

“Legacy vs. AI-Native” Companies Isn’t the Right Question

4 minute read

Published:

Executive Summary

Companies that scaled before November 2022 are fundamentally disadvantaged competing against AI-native startups. “Legacy” now means pre-ChatGPT era. Organizations with 30+ employees face structural barriers to becoming truly AI-native due to organizational inertia, incompatible systems, and cultural resistance. The competitive advantage goes to companies with value propositions impossible to articulate before 2022.

Key Insights

  • The New Definition of Legacy: “Legacy company” no longer means decades-old—it means pre-November 2022 (pre-ChatGPT). Technological disruption timelines have compressed dramatically.

  • AI-Encumbered Organizations: Companies adopting AI without strategic redesign amplify existing dysfunctions rather than gaining competitive advantage. MIT’s NANDA initiative found 95% of enterprise GenAI pilots failed to increase revenues—not due to AI quality, but organizational learning gaps and flawed integration.

  • The Workslop Problem: Organizations produce AI-generated content nobody reads but everyone feels obligated to create. Walls of memos, slides, and reports create institutional noise rather than value.

  • Three Distinguishing Factors for truly AI-native companies:
    1. Value propositions impossible to articulate before 2022
    2. Organizations small enough to avoid structural transformation barriers (typically <30 employees)
    3. Companies that don’t loudly claim “AI-native” status—they just build differently
  • Structural vs. Cultural Transformation: Organizations with established systems, processes, and hierarchies cannot simply rebrand as AI-native. True transformation requires rebuilding from foundations.

Who Should Read This

  • Prospective employees evaluating company AI claims during interviews
  • Biotech investors assessing competitive positioning and AI strategy authenticity
  • CTOs and executives questioning their organization’s AI transformation approach
  • Founders building AI-native biotech and drug discovery companies
  • Job seekers choosing between established companies and AI-first startups

The AI-Native Biotech Context

In drug discovery specifically, this distinction matters enormously:

AI-Encumbered Biotechs:

  • Add computational biology teams to existing research organizations
  • Integrate ML tools into traditional medicinal chemistry workflows
  • Generate AI predictions that medicinal chemists review and override
  • Treat data infrastructure as IT problem, not core capability
  • Hire ML scientists who report to biology leadership

Truly AI-Native Biotechs:

  • Build lab-in-the-loop systems where experiments directly inform models
  • Design organizational structure around data generation and model iteration
  • Hire ML engineers and computational biologists in equal proportion to experimentalists
  • Make data quality and accessibility foundational, not afterthought
  • Create value propositions centered on autonomous optimization impossible pre-2022

Practical Takeaways

If You’re Evaluating a Company:

  • Scrutinize companies requiring explicit “AI” mentions in their value proposition
  • Ask: “Could you have built this company in 2021?” If yes, it’s not truly AI-native
  • Recognize organizational size (>30 employees) as structural constraint on transformation
  • Evaluate whether they’re building with AI versus bolting it onto existing systems

If You’re Leading AI Transformation:

  • Accept that true AI-native transformation may be structurally impossible for established organizations
  • Consider focused AI initiatives in specific domains rather than enterprise-wide adoption
  • Recognize that organizational learning gaps, not technology limitations, drive failure
  • Question whether forced AI adoption creates “workslop” rather than value

If You’re Building an AI-Native Biotech:

  • Design organizational structure, data systems, and culture around AI from day one
  • Hire for comfort with AI-first workflows before scaling past ~30 people
  • Focus on capabilities literally impossible before foundation models existed
  • Don’t waste energy claiming “AI-native” status—demonstrate it through what you build

The Bottom Line

The question isn’t “legacy vs. AI-native”—it’s whether your organization can fundamentally restructure around AI capabilities, or whether you’re building with AI as the foundation from inception. For most companies that scaled before ChatGPT, the honest answer is structural transformation isn’t feasible. That’s not a failure—it’s recognition that competitive advantages come from focused AI applications in specific domains rather than company-wide rebranding.

Build new. Build different. Or accept that AI augmentation, not transformation, is the realistic path.

Full article: Read on Substack

Last updated: October 2025

Keywords: AI-native companies, AI transformation failure, organizational design AI, biotech AI strategy, pre-ChatGPT legacy companies, AI-encumbered organizations, drug discovery AI adoption, startup competitive advantage, organizational inertia AI, AI-first biotech

You May Also Enjoy

The Nuances of Antibody Structure Prediction

Published:

Protein structure prediction advanced dramatically, but therapeutic antibody design faces a bottleneck: accurately predicting CDR loops, especially CDR H3, for novel sequences.

Getting Started in BioML Research & Engineering

Published:

Breaking into AI for Bio requires 1-5 years of sustained effort: solid fundamentals, meaningful research experience, demonstrated execution ability, and strong engineering practices. What actually matters when hiring managers review candidates.