Restore original stack-section wording, no bold

Author: Jayanaka-98

docs/blog/posts/449-project-repost.md

@@ -22,7 +22,7 @@ draft: true
 
 ## One language for the whole stack
 
-Building a full AI application usually means stitching together separate systems: one framework for the interface, another for the backend, a database underneath, and more tooling to call models and orchestrate workflows, with every seam its own thing to learn, wire up, and keep from breaking. **That glue work eats up much of a semester, and it is exactly what Jac removes.** **Jac and the Jaseci runtime let a team handle the interface, the application logic, the data, and the AI itself in one language, with model calls native to the language rather than a bolted-on service and model-agnostic underneath, so teams could reach for whatever model fit without re-architecting anything.** Less time on plumbing, more on the actual problem, which is most of why a single semester was enough.
+Building a full AI application usually means stitching together separate systems: one framework for the interface, another for the backend, a database layer underneath, and more tooling to call models and orchestrate workflows. Each seam is its own thing to learn, wire up, and keep from breaking, and for a four-person team on a deadline, a lot of the semester can disappear into that glue work instead of the product. The single-language design is a big reason a semester was enough. Jac and the Jaseci runtime let a team handle the interface, the application logic, the data, and the AI itself in one language. AI calls are part of the language rather than a bolted-on service, and Jac stays model-agnostic underneath, so teams could reach for whatever model fit without re-architecting anything. Less time on plumbing, more on the actual problem.
 
 ## A few from the showcase