← back

"The biggest challenge in your stack? Evals, Evals, Evals" - 2026 State of AI Engineering results

4.8K views · Jul 21, 2026 · 19:47 min · Watch on YouTube ↗
Takeaway

AI engineering is becoming a multimodel production discipline shaped by quality, agent capabilities, and cost.

Summary

  • The survey gathered 1,048 responses and found experienced software engineers often have only a few years of AI experience.
  • Among respondents not using audio, 56% planned to adopt it; satisfied image-generation usage rose from 18% to 36%.
  • Closed models were used by 94% and open-weight models by 45%, with over 90% of open-weight users also using closed models.
  • Some 87% used multiple models, while 40% said cost regularly limited AI ambition and another 36% said it sometimes did.
  • Reported agent adoption reached 95%, alongside growing standardization of tools and platforms rather than a single model.
ai-surveymodel-adoptionai-costs
Original description
Barr Yaron shares her perspective on the results and emerging state of AI engineering in 2026.

Speaker:
Barr Yaron — Partner, Amplify Partners
Barr backs founders building the AI infrastructure and applications that will shape the future.

Links:
X: https://x.com/barrnanas
LinkedIn:   / barryaron  
Website: https://barrchives.com

Timestamps

0:00 Introduction and Survey Context
2:26 The AI Engineering Workforce
3:21 Current Modalities and Adoption
5:34 Model Strategy: Closed vs. Open-Weight
8:20 Cost as an Engineering Constraint
9:36 The Rise of Agentic Workflows
11:57 Infrastructure Challenges and Evals
12:53 The Build vs. Buy Trade-off
14:09 Impact on Engineering Culture and Teams
16:27 Future Bets and Predictions

Quotes:

"The median new engineer has nearly as much AI experience as the median 10-year software veteran." (3:07)

"Cost is now a first-class engineering constraint." (8:38)

"Agents are no longer reading, summarizing, drafting—they’re taking actions inside of systems." (10:58)

"The biggest challenge in your stack? Every single year, the number one answer is eval." (12:12)

"Shipping software is not gated on being an engineer anymore." (16:17)