One-Day Intensive
Production Agent Engineering
09.17.26 | 6:00 PM - 7:30 PM PDT 1 hours in one focused day
Ship agent swarms that survive real load, eval gates, tracing, guardrails, and cost control.
with Liz Howard
Part of the Agent Engineering Stack: From Zero to Production in 3 Days track, 8 classes, take them in any order.
See the whole trackOverview
This is the production tier: not toy agents, but swarms that survive real load. You'll gate quality with generator-critic loops that evaluate, refine, and refuse to ship below your bar; orchestrate long-running task agents by goal, polling for stalls, capping turns, and killing the infinite loops that wreck real systems. You'll trace every hop of a multi-agent run (who did what, where it broke, what it cost), guardrail the swarm by validating inputs and outputs, fencing tool access, and keeping a human in the loop, and finally build an agent that builds agents: a reusable library that loads MCP tools on demand. You leave with the observability, evaluation, and reliability craft that separates a demo from something you'd actually run, and control of every agent in it.
What You'll Master
- Gate quality with a generator-critic loop, refuse to ship below your bar
- Orchestrate long-running agents by goal, poll for stalls, cap turns, kill infinite loops
- Trace every hop of a multi-agent run, who did what, where it broke, what it cost
- Guardrail your swarm, validate I/O, fence tool access, keep a human in the loop
- Build an agent that builds agents, a reusable library that loads MCP tools on demand
Who this is for & what you'll need
Pro engineering
Built for working engineers.
A few installs
You'll install a couple of tools, we'll help you in class.
Where This Class Leads
What you leave able to do
- reconcile agents from different toolchains into one run
- select among agent architectures
- synthesise a memory hierarchy for an agent
- synthesise a retrieval over relationships not just similarity
- synthesise a self extending agent behind a review gate
- synthesise an accountability regime for an unattended agent
- originate an operating model for managed agent teams
How you show it. A written operating model in use by at least two people, with a recorded instance of it catching a failure that the previous arrangement missed.
It sits on one path
Build production AI systems from prompt to deployment.10 of 12 after Advanced Retrieval Augmented Generation · before Agentic SDLCAbout 24 hours of practice sit behind this one. See what your hours buy.
Join Us
One-Day Intensive
- Industry mentors
- Participate in active research projects
- Networking events
- Integrate AI into your workflow from the beginning
- Your seat in the class
- Lifetime access to materials
- Sponsors 1 scholarship seat for someone else
The minimum is $100, it keeps out spam. Need to pay less? Apply for a scholarship, no one is turned away for lack of funds.
Secure payment via Stripe
3 scholarship seats available. Not enrolling, but want to put someone through this class?
Instructor
Want to schedule this class for your team?
Contact us: liz@themultiverse.school