How Agentic Buildcamp actually runs

Week by week: what you build, when you demo it, who pushes on it, and the practices that keep the thing you ship reliable enough for someone else to depend on.

How You Build

The program is delivered using Sherpa-B, an MCP server that acts as your building mentor or thought partner. Concretely, this is a series of workouts you run directly in your agentic coding environment (eg. Claude Code). You learn the methodology by using it in action while building.

1
Day one: you start building
We help you set up a coding agent environment (eg. Claude Code), and you start building right away.
2
Week one: daily stand-ups & first demo
To help you build momentum, every day this week, we do a short group stand-up with your mentor and cohort: feedback on what you're building, answers to your questions. After about one week of building, you close it out with a short demo.
3
First demo feedback session
As a group, we watch and comment on a few of the submitted (recorded) demos, then discuss them together, so you see how other builders approach the problem, what stack they used, and what feedback they get, catching pitfalls before you hit them. Mentors review every submitted demo individually, then share broader guidance based on patterns they've seen across many cohorts.
4
The build cycles
In the weeks between your first demo and the final week, you build mostly self-directed, guided by Sherpa-B, with async mentor support on Slack (or scheduled calls with your mentor if you've added 1-on-1 mentorship). Each cycle ends with another demo, so you keep getting feedback as you go.
5
Final week: daily stand-ups again
To help you with any last minute issues, we do daily stand-ups again, with direct feedback from mentors and peers on your progress as you close out the build.
6
Public demo
You present your shipped agent to our broader community and network, often several hundred people, including experienced practitioners, founders, and sometimes angel investors or corporate directors. It's common for builders to land follow-up meetings with people in that room.

+ If you've added 1-on-1 mentorship

You'll also meet weekly with your mentor to discuss your project one on one, at whatever timing works for both of you: you book directly against your mentor's availability using a link they share with you, for one week or several at a time.

Rigor Is The Point

Shipping an agent quickly is the easy half. The harder half is knowing whether it works, where it breaks, and what happens when it does. The process brings structure, stability, and reliability to agentic builds, so what you ship is something a skeptic can lean on.

Quality and risk, assessed rather than assumed
Agentic systems fail in ways that are easy to miss if you only watch the happy path. The workflows push you to assess, evaluate, and test what you are building, so you have a way to know whether the output is actually good instead of just moving fast and hoping.
Evaluation as a first-class skill
Evaluation and context management are treated as core AI engineering skills, not an afterthought bolted on at the end. You write test cases and evals for your agent the same way you write the agent itself.
Guardrails, safety, and observability
Good practices are built into the process: guardrails on what your agent is allowed to do, and observability so you can see what it actually did. The result is closer to a practical starter kit for serious multi-agent systems than to a class.
Grounded in what the market needs
User interviews and stakeholder conversations are part of the build, so the thing you ship is grounded in what people actually need rather than what you assumed they wanted.
An insight at every step
The workflows sharpen each step as you go. Builders consistently report catching things at a given stage that they would have overlooked entirely without the structure prompting them to look.

What Is On The Calendar

The schedule below is the live calendar for the cohort currently open for signups. Every marker on it is one of a handful of recurring session kinds, and here is what each one is.

Meet & Greet: The opening session of a cohort. You meet your mentor and the other builders, and you leave knowing who you are building alongside for the next several weeks.
Stand-ups: Short live group sessions with your mentor and cohort, daily in the first week and again in the final week. You say what you are building, what is blocking you, and you get feedback and answers on the spot.
Cycle Demos: Each build cycle ends with a recorded demo. Mentors review every submitted demo individually, and the group watches and discusses a few of them together, so you see how other builders approached the same problem and what feedback they got.
Public Demo: The final demo, presented to the broader community and network rather than just your cohort: experienced practitioners, founders, and sometimes angel investors or corporate directors.
Other: Anything on the calendar that does not fall into the four kinds above, typically a one-off or newly named session for the active cohort.

Full schedule

Loading schedule…