Live LabsEvery week, one system, built end to end.
Tuesdays for labs up to two hours, Saturdays for the longer ones. You watch it built, build it together, then build it on your own code.
Upcoming
Season 1. Times show in your timezone.
- BuildersSystem design
Pick the right architecture: single agent, workflow or multi-agent
A decision table and a measured comparison of two designs for one real task.
2 hours - LeadersTransformation
Write the AI coding playbook for your team
A two-page playbook: where agents may write code, what needs a spec first, and how review changes.
2 hours - BuildersSystem design
Design a production agent from a blank page
A written architecture for one agent: components, failure modes and a cost estimate, reviewed by peers.
3 hours - LeadersBuying AI
Choose AI models and vendors with your own evals
An eval suite built from your own tasks, and a scored comparison of three models on quality and cost.
2 hours - BuildersEvals
Add evals and tracing to an agent
Traces, a dataset built from real failures, and an experiment that proves a prompt change helped.
3 hours - BuildersSystem design
Make agents survive failures with durable workflows
An agent whose runs retry, pause for a human and resume after a crash.
2 hours - BuildersMCP
Build an MCP gateway: one front door for your tools
Several MCP servers behind one gateway, with authentication and an audit log of every tool call.
2 hours - LeadersAdoption
Measure AI coding adoption without surveillance
Five adoption and cost metrics for your team, and a written charter of what you will not track.
1 hour - LeadersUpskilling
Upskill your team for AI-native work in 90 days
A 90-day plan where every engineer pair ships one small internal AI project, with a check-in each fortnight.
1 hour - BuildersCoding agents
Build a coding agent from scratch: the loop, the tools, the sandbox
A coding agent that reads a repo, edits files and runs tests inside a sandbox, with every step logged.
3 hours - BuildersBrowser agents
Build a browser agent that finishes the task
An agent that completes a multi-step web task, and a way to see where it fails.
2 hours - BuildersAI QA
AI QA: an agent that tests your web app
End-to-end tests written in plain English, running against a real page in CI.
2 hours - BuildersEvals
Test an agent like software
A regression suite and a red-team run for one agent, failing the build when quality drops.
2 hours - BuildersCoding agents
Put coding agents in your CI: review, fix and open pull requests
A repo where an agent reviews every pull request and turns a labelled issue into a draft pull request a person merges.
3 hours - BuildersKnowledge
Build a second brain: an LLM wiki that maintains itself
A markdown wiki that an LLM updates from your raw sources, with rules you control.
2 hours - BuildersKnowledge
Company brain end to end: a Slack teammate that remembers
A Slack agent with team memory, connected to two sources, with access scoped by channel and tested for leaks.
5 hours - BuildersAI SDR
Build an AI SDR that drafts outreach and waits for approval
A pipeline that researches a lead and drafts a first message, with nothing sent until a person approves.
2 hours - BuildersAutomation
Automate a business process with a workflow and an agent
One real process automated, such as invoice triage, with an approval step a person can use.
2 hours - LeadersPlatform
Golden paths for AI services: a template every team starts from
A service template with tracing, evals and cost limits built in, and a check that new services use it.
2 hours
Past labs and recordings
Members get every lab replay and repo. Lightning recordings are free for everyone.
The first Live Lab runs on 20 October. Until then, the free Lightning recordings show how Param teaches.
Lightning recordings
- Recording60 min
Design a Software Factory for Your Coding Agents
See what each layer does and which problem it removes from your daily agent work.
Watch the recordingMaven - Recording60 min
Build Your First Software Factory Execution Harness
Give each agent task its own Git worktree that is created, used and removed cleanly.
Watch the recordingMaven