What 75 students shipped in one semester with Jac (#46)

* Add blog post: What 75 students shipped in one semester with Jac

* Update blog post: What 75 students shipped in one semester with Jac

* Style 449 repost with lead, stat grid, pull quote, and CTA

* Revert decorative styling; keep repost concise

* Style 449 repost with lead, pull quote, and CTA

* Make showcase intro a callout and tighten copy

* Tighten showcase intro and merge stack section into one para

* Bold key phrases in the stack section

* Bold the why-Jac-fits sentences in the stack section

* Clarify the glue-work sentence

* Restore original stack-section wording, no bold

* Restore stack section to original two paragraphs

* Retitle stack section to One language, one semester

* Retitle showcase section and clarify its lead-in

* Add transition between glue-work and single-language paragraphs

* Collapse closing section into a single CTA box

* Upgrade CTA to rich card variant and clarify products line

* Bold CTA opener and drop documentation link

* Update blog post: What 75 students shipped in one semester with Jac

* Update blog post: What 75 students shipped in one semester with Jac

Author: Jayanaka-98

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

@@ -0,0 +1,47 @@
+---
+date: 2026-06-29
+authors:
+  - jaseci-team
+categories:
+  - Community
+  - Developers
+slug: 449-project-repost
+repost: true
+repost_url: "https://cse.engin.umich.edu/stories/students-build-agentic-ai-tools-for-work-travel-wellness"
+repost_source: "Students build agentic AI tools for work, travel, wellness"
+draft: true
+---
+
+# What 75 students shipped in one semester with Jac
+
+<p class="lead" markdown="1">The University of Michigan's CSE division just published a [story on the final showcase for **EECS 449: Conversational AI**](https://cse.engin.umich.edu/stories/students-build-agentic-ai-tools-for-work-travel-wellness), where **75+ students**, working in teams, built **17 full-stack AI applications** in a single Winter semester, most of them in [Jac](https://www.jaseci.org/).</p>
+
+<!-- more -->
+
+**These products aren't chatbot demos.** They are live and full-stack: conversational interfaces, visual reasoning, document analysis, multimodal input, and multi-step agentic workflows, built for everything from academic advising and health to travel, finance, and career prep.
+
+## One language, one semester
+
+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.
+
+That is exactly where Jac came in. Its 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.
+
+## From the showcase
+
+The range shows what Jaseci made possible. Here are three standouts:
+
+[**Skooch.ai**](http://skooch.ai) is an adaptive scheduling assistant that learns someone's focus windows and energy patterns, then places flexible tasks around fixed commitments from a plain-language prompt like "build me a study plan."
+
+[**MaizeMind.com**](http://maizemind.com) turns half-formed notes into an interactive argument map, surfacing claims, evidence, and contradictions without ghostwriting, with Jac's graph data model mapping directly onto the argument structure.
+
+[**EdgeCastApp.com**](http://edgecastapp.com) is a research terminal for prediction-market traders that links live Kalshi markets to a real-time newsfeed and an agent that already knows which market you're viewing.
+
+The same story sits underneath all three: Jaseci was the common thread that made them feasible on a student timeline, letting each team keep the interface, the logic, the data, and the AI in one application instead of a stack of integrations.
+
+You can explore the full lineup, with every team, demo videos, live apps, and GitHub repos, on the [EECS 449 projects showcase](https://jaseci-labs.github.io/eecs-449-projects/), and read more in the [U-M story](https://cse.engin.umich.edu/stories/students-build-agentic-ai-tools-for-work-travel-wellness).
+
+<div class="cta cta-rich" markdown="1">
+**Want to build something full-stack in a single language?** Install [Jac](https://www.jaseci.org/) and scaffold your first app in a few minutes, with the backend, frontend, and AI in one codebase, the same setup behind these 17 student projects.
+
+[Try it yourself](https://docs.jaseci.org/){ .cta-button }
+</div>

styles/prose.css

@@ -379,6 +379,27 @@
     transform: translateY(-1px);
 }
 
+/* Rich variant — a self-contained card (full hairline border + soft shadow
+   instead of the left accent bar) with a little more breathing room and a
+   button that lifts off the surface. */
+.prose .cta.cta-rich {
+    margin: 3.5em 0 1em;
+    padding: 2.75rem 2.5rem;
+    border-left: none;
+    border: 1px solid var(--rule);
+    border-radius: 16px;
+    box-shadow: var(--shadow-card);
+}
+.prose .cta.cta-rich a.cta-button {
+    margin-top: 1.4rem;
+    padding: 0.8rem 1.9rem;
+    box-shadow: 0 8px 20px -8px var(--accent);
+}
+.prose .cta.cta-rich a.cta-button:hover {
+    transform: translateY(-2px);
+    box-shadow: 0 12px 26px -8px var(--accent);
+}
+
 /* Lead paragraph — the standfirst right under the title. A soft card on a warm
    neutral fill (not an accent tint, which reads synthetic) with a thin accent
    edge, same serif as the body, justified to match the article. */