Private Equity FunCast Private Equity Predictions 2026
4 snips
Feb 25, 2026 They trade hot takes on fundraising trends and which firms will thrive in a barbell market. Liquidity forecasts focus on secondaries and continuation vehicles. Deep discussion on why AGI is still years away and how LLMs serve as interfaces. Predictions that domain-specific AI models and embedded AI features will dominate product strategy. Cybersecurity and cloud costs get special attention for due diligence.
AI Snips
Chapters
Transcript
Episode notes
AGI Is Still Years Away
- AGI remains far off; current LLMs are powerful conversational predictors but not general thinkers.
- Devin argues AGI is 5–10 years away because LLMs must be integrated with many tighter, non-text prediction systems to reach human-like reasoning.
Domain Models Beat Public LLMs For Products
- Larger general-purpose LLMs will lose appeal inside products; domain-specific, tightly trained models on a company's own data will win.
- Devin cites image-recognition for auction company Hybid as an example where a custom model outperforms a generic LLM.
Replace Pilot LLMs With Controlled Models
- Move from public LLM pilots to your own open or narrow models for latency, cost, and predictability.
- Devin warns token costs, slow API latency, and hallucination risk make vendor APIs impractical for production customer-facing features.