Enterprise AI Adoption: Why 3-5 Years Before Massive Gains
Winston Weinberg, CEO of Harvey ($190M ARR, 1000+ customers), made a bold prediction: 3-5 years until we see massive enterprise productivity gains from AI. That's a much longer timeline than the "AI is eating software" narrative suggests.
Why Enterprise is Different
Consumer AI = Fast Adoption
- Individual downloads app
- No procurement process
- Immediate value
- No one to blame if it fails
Enterprise AI = Slow Roll
- Requires procurement (often 6-12 months)
- Integration across dozens of systems
- Compliance and security reviews
- Change management across teams
- Measurable ROI expectations
What Holds Enterprises Back
1. Integration Complexity
Enterprises run on thousands of interconnected systems. A consumer AI tool connects to 3-5 apps. Enterprise AI needs to work with legacy CRMs, custom ERPs, data warehouses, and homegrown tools—all with different APIs and security models.
2. Compliance Overhead
Every AI vendor must pass:
- Security questionnaires (often 100+ questions)
- SOC 2 / ISO 27001 certifications
- Data residency requirements
- Legal review of AI-specific terms
3. Organizational Friction
Even when the technology works:
- Training thousands of employees
- Changing established workflows
- Manager skepticism
- Union/worker council concerns
The Numbers
| Stage | Timeline | What's Happening |
|---|---|---|
| Pilot | Months 1-6 | Small team, limited scope, proof of concept |
| Expansion | Months 6-18 | More users, integration testing, ROI measurement |
| Scale | Months 18-36 | Organization-wide rollout |
| Optimization | Months 36+ | Refinement, efficiency gains |
What This Means for AI Startups
If You're Selling to Enterprises
- Plan for 3-5 year sales cycles in your runway
- Enterprise-readiness is the moat: Security, compliance, integrations
- Land and expand: Start with one team, prove ROI, expand slowly
- Hire enterprise talent: People who understand procurement
If You're Building Internal Tools
- Start now: You'll learn what works in your context
- Build for scale: Don't hack together something that can't grow
- Measure everything: Enterprise wants ROI proof
The Counterpoint
Some argue:
- Consumer-grade tools will emerge for enterprise "edge" cases
- Vertical AI will solve integration by owning the whole workflow
- AI-native companies (founded after 2023) will adopt faster
Bottom Line
Enterprise AI isn't dead—it's just slower. The teams that win will be patient, enterprise-ready, and focused on measurable ROI.
For founders? This means:
- Don't bet on rapid enterprise adoption
- Build for the long sales cycle
- Focus on clear ROI stories
Winston Weinberg shared this prediction on the 20VC podcast. tldl brings you concise summaries of the best AI podcasts for founders and operators.