# AI in Production
Ashwin Bilgi
*NEC Software Solutions — May 2026*
---
**46%** of AI proof-of-concepts scrapped before production.
— S&P Global, 2025
**6%** of companies fully trust AI agents for core processes.
— HBR, 2025
**89%** observe. **32%** still blocked on quality.
— LangChain, 2026
--- ## A letter
A
B
C
D
E
--- ## Lost
A
B
C
✗
D
E
--- ## A copy at every hop
A
B
C
D
E
--- ## Traditional code Gets **more predictable** in production. --- ## AI Gets **less predictable** in production. --- ## Our tools
Aggregates. Percentiles. Long-tail latencies.
Built for the first kind of system.
--- ## AI inverts it
One wrong answer can be the whole story.
Aggregates hide it.
--- ## And the input is lossy
Customers don't write prompts the way developers do. Especially on mobile. Especially on WhatsApp.
--- ## The one place AI works
Developers with coding agents.
The developer steers, retries, wants it to work.
Your customers won't.
--- ## Where the work is
Context engineering. Long-running agents. Tool curation. Human-in-the-loop, by design.
--- ## An infrastructure problem
The right context for the AI. The right moment for the human. Without overwhelming them.