Between the model and the product
The AI Runtime is the publication where AI engineers learn the production craft. We cover the operational layer where AI systems actually run — evals, agents, inference, reliability, cost — and the lessons from people shipping serious AI in production.
No vendor pitches. No idea-stage tourists. No "future of AI" speculation. The test for every piece: would a senior engineer at Anthropic, OpenAI, or DeepMind forward this to their team?
What we cover
- 01
Evals & Observability
How engineers know if AI works in production.
- 02
Agents in Production
Patterns that survive contact with real users.
- 03
Inference & Serving
What actually runs in production at scale.
- 04
Reliability & Incidents
Postmortems, drift, recovery patterns.
- 05
Cost & Performance
The economics of running AI.
Kranthi Manchikanti
Kranthi is founder and editor-in-chief of The AI Runtime. He's an AI Architect at Microsoft based in Boston, building the publication and the Boston meetup as a long-term home for engineers shipping AI to production.