Agentic AI and vibe coding workshops

From a real HR problem to a clickable prototype in one session

In fall 2025, Fountain Labs ran hands-on AI workshops on site with HR teams. In a single session, leaders learned what agentic AI is, picked a real problem from their own work and used an AI app builder to turn it into a clickable prototype. No coding was needed, and the ground rules called for sample data only. After the first session we built Agent Lab, a web hub that takes a team from a problem to a build prompt.

01 / 05

Why we ran them

HR and talent teams know where their workflows break, but testing a fix usually means waiting on an engineering team. AI coding tools can shorten that path, but a blank prompt box is a hard place to start, and most people don't know what to ask for.

02 / 05

What we built

A repeatable agenda: a short primer on agentic AI, a look at how Fountain uses agents in its own products, an optional live demo, then 60 to 90 minutes of building and a share-out.
A problem-finding exercise, added as the series went on, with four questions: what only works because someone knows the unwritten rules, what your team would do with 10 hours back each week, what question keeps coming up, and what you could do if the problem were solved.
Agent Lab, the workshop hub we built after the first session: a prompt builder that turns a brief into a detailed prompt for an AI app builder, and later a 30-minute worksheet that takes a team from its pain point to a short, build-ready brief.
A worked example app with the full prompt behind it, so people could copy and remix it.
Plain-language guardrails: sample or fake data only, nothing connected to company systems, IT sign-off before the day, and prototypes for exploring ideas, not for production.
A facilitator kit (agenda, slides, the exercise and a recorded walkthrough) so more than one Fountain presenter could run the session.

03 / 05

How it works

Learn the basics: what agentic AI is and how it differs from generative AI.
See it in practice: how Fountain uses agents in hiring, with a live demo if the group wants one.
Pick a problem: small teams answer four questions and land on one problem worth solving.
Build: the prompt builder turns the problem into a prompt, and each team builds a clickable prototype.
Share: each team shows the room what it built.

04 / 05

Where it landed

Ran on site with HR teams in October and November 2025, in groups of about 15 people.
Teams built clickable prototypes in a single session, from dashboards to planning tools.
Once the problem-finding exercise was part of the day, builds stayed on real business problems instead of novelty apps.

05 / 05

What we learned

Start with the problem, not the tool. The problem-finding step is what made the builds useful.
Small teams work better than a room of individuals.
Bring IT in early. Clear rules, like sample data only and nothing connected to company systems, help answer the questions that come up.
A clickable prototype is a fast way to explain an idea to your product, ops or IT partners. It is a starting point, not production software.
Keep the day tight. A primer, a demo, a brainstorm and a build is a lot for one session.
Some people wanted to build agents and automations, not only apps. Building agents in Fountain is the focus of CPUC 2026.