Working Notes on Agent Systems/Brad Zhang

@teach_fireworks / X longform

From demo to production, there is a lack of a production system in the middle. This Dat...

From demo to production, there is a lack of a production system in the middle. This Databricks sharing is very suitable for calibrating "how to move AI demo to production". Many...

June 23, 2026 · 1 min read

From demo to production, there is a lack of a production system in the middle.

This Databricks sharing is very suitable for calibrating "how to move AI demo to production".

Many companies will start with model selection and demo.

Controllable data, limited scenarios, and beautiful demonstrations make the project look ready to go online.

The real problems usually appear after a few weeks: users start asking various boundary questions, the output is inconsistent with the demo expectations, and the team cannot tell at which step the AI ​​deviates.

The judgment given here is very straightforward: from demo to production, what needs to be supplemented are observability, evaluation, and accountability.

Whether you can see inside the system, whether you can continuously measure business results, and whether you can trace specific links after errors occur, determine whether the project can enter the real environment.

Visual summary

Article argument map

Generated from the post's content graph

FORMATTOPICCAPABILITYMARKETcoverssignalssignalsFORMATlongform noteTOPICevaluationCAPABILITYevaluationCAPABILITYproduct surface
Mermaid outline
flowchart LR
  format-long_post["longform note"]
  topic-evaluation["evaluation"]
  capability-evaluation["evaluation"]
  capability-product-surface["product surface"]
  format-long_post -->|covers| topic-evaluation
  format-long_post -->|signals| capability-evaluation
  format-long_post -->|signals| capability-product-surface

Visual structure

Essay structure map

Built from summary and key paragraph positions

From demo to production, there is a lack of a production system in the middle. This D...THESISFrom demo toproduction, there is alack of a productionsystem in the middle.SIGNALFrom demo toproduction, there is alack of a productionsystem in the middle.OPERATORControllable data,limited scenarios, andbeautifuldemonstrations makeIMPLICATIONWhether you can seeinside the system,whether you cancontinuously measure
Mermaid outline
flowchart LR
  thesis["From demo to production, there is a lack of a production system in the middle. This Databricks sharing is v..."]
  signal["From demo to production, there is a lack of a production system in the middle."]
  operator["Controllable data, limited scenarios, and beautiful demonstrations make the project look ready to go online."]
  implication["Whether you can see inside the system, whether you can continuously measure business results, and whether y..."]
  thesis -->|frames| signal
  signal -->|develops| operator
  operator -->|lands in| implication

Source: View the original post