Founder @ KaaS.team, Agentic AI educator

Most agentic AI courses teach one shape: a person asks, an agent retrieves, an answer comes back. Real systems come in three shapes. In a workflow, code decides each step. In an agent, the model decides. In a team of agents, work is shared, starts from an inbox or a schedule, and runs for hours. The wrong shape costs tokens, time and trust.
In four weeks you build two production systems. Project 1 is a Healthcare Agentic AI System, a drug maker's inquiry desk, where an eval gate fails the release on one missed adverse event. Project 2 is your own team of 24/7 AI employees on a mission control board, each hired for one outcome, with a human approving anything risky.
Every week runs I do, We do, You do. Recorded lessons and design walkthroughs explain the idea. On Monday we build it together, live. On Wednesday you start the assignment in a live studio, as an engineer shipping code or as a PM writing the decisions. Your capstone is your own AI employee, defended in a one-to-one architecture review.
Design, build and deploy two production agentic AI systems in four weeks, with an engineer track and a PM track for every assignment.
Place each job on the map as a workflow, an agent or a team of agents.
Decide what starts it, a person, an event or a schedule, and how long it lives.
Price each shape in tokens before writing code.
Route, ground and check every inquiry, with code deciding each step.
Ground every answer in approved content.
Fail the release when one adverse event is missed.
Give each employee one outcome, one metric and one thing it must never do.
Count work as done only when the system of record agrees.
Show the cost per outcome on the board, failed runs included.
Run a manager that hands off work and blocks any agent approving its own action.
Keep a human in the loop on the board for anything that cannot be undone.
Resume a nightly job killed mid-run without writing a row twice.
Block a prompt injection hidden in a customer email.
Gate every model swap on evals built from real traces.
Onboard a second tenant that cannot see the first.
Every assignment has an engineer deliverable and a PM deliverable on the same system.
Engineers ship code that a test proves.
PMs ship the outcome card, eval labels, approval matrix and launch review.

Founder @ KaaS.team, Engineering leader, Agentic AI educator
Software engineers whose agent demo now has to ship (you build both systems and prove each property with a test)
Product managers who own an agent feature (you write the outcome, the eval labels, the approval rules and the cost model)
Tech leads and architects who sign off agent designs (you learn the map, the patterns and the questions a review asks)
The sessions build in Python, and you can read TypeScript.
You can clone a repository, run containers and call an API.
Any major provider works; budget some API credit for the weekly projects.

Live sessions
Learn directly from Paramanantham Harrison in a real-time, interactive format.
Live build sessions
Every Monday we build the next part of each system together, and you leave with it running.
Live assignment studios
Every Wednesday you start that week's assignment with help in the room, then show it at Friday office hours.
Recorded lessons before each week
Short lessons explain each idea, so the live time goes to building, not lectures.
Two reference systems with source code
The Healthcare Agentic AI System and the 24/7 AI employees, each with its tests and decision records, ready to run from the first session.
Engineer and PM tracks
Each assignment has a code deliverable and a decision deliverable, with templates for the outcome card, eval labels, approval matrix and launch review.
Recorded design walkthroughs
Walkthroughs of agentic systems real companies run and have written about, each covering the job, the shape, the design, the measure and the trade.
Architecture review of your capstone
A 30-minute one-to-one review of your own AI employee's design with me, before demo day.
Capstone demo and live feedback
Demo your AI employee on the board, with its outcome and cost per outcome, on the final Friday.
Recordings you keep
Every live session is recorded, so a missed session or a second look is always there.
Rejoin a future cohort
Join a later live cohort of this buildcamp at no extra cost, as the models and the patterns change.
Maven Guarantee
Your purchase is backed by the Maven Guarantee.
12 live sessions • 5 projects
Live sessions
4-4 hrs / week
Two build sessions a week on Mondays and Wednesdays, recorded in case you miss one
Office hours
1-1 hr / week
Fridays, live Q&A where each week's project is due
Projects
3-5 hrs / week
One part of your AI mission control each week
Nov
20
Nov
16
Nov
18
Nov
27
Nov
23
Nov
25
Maven for Teams
Reimbursement
Get your company to pay
Everything L&D needs: email template, receipts, and certificate of completion.
Get reimbursedTeam discount
Learn with your teammates
Save 20%+ when 2 or more teammates enroll in the same cohort.
Save 20%+ with a teamPrivate cohort
Run a cohort for your org
A dedicated cohort with a custom schedule and curriculum, tailored to your team.
Book a private cohort$1,350
USD
2 cohorts