AIA practical internal series for Baims, Orcas, and MedMasters. Learn to use today's AI capabilities, connect agents to company work, automate repeated processes, and build useful internal tools.
One repeated task. Four ways forward.
Move from context to a useful internal tool.
Connect context
Files, apps, and memory
Use tools
Computer and browser
Automate work
Schedule and replay
Build the tool
Codex and Supabase
Working rule
Agent speed. Human approval.
Session 1
The opening session establishes the foundations. The full curriculum continues across the broader series.
Wednesday
15
July 2026
Intro to AI Agents
You'll see how to give agents better context, connect them to the work, and keep their output verifiable. Then we'll turn repeated internal workflows into tools the group can use.
Engineering joins as both participant and collaborator, contributing technical examples and helping the group turn strong ideas into useful classes and tools.
The full series
The series starts with using AI as part of everyday work, then moves into building the internal tools that repeated work deserves.
Move beyond isolated prompts. Give AI the context, tools, connections, and repeatable processes it needs to do useful work.
The tools available now, why app versus browser matters, and how models and thinking effort change the result.
Projects, memory, AGENTS.md, files, and directories that keep the work grounded.
Plugins, MCPs, Computer Use, and Chrome Use across the apps your team already relies on.
Record & Replay, scheduling, automations, mobile control, and dictation everywhere.
Subagents, reusable skills, artifacts, and inline graphs that make the result easier to use and verify.
Turn a shared pain point into a focused tool the team can use, improve, and own together.
Choose repeated work worth improving. Identify the user, the current steps, and where the frustration happens.
Learn the difference between UI and UX, map the user journey, and sketch what the tool should feel like.
Understand how the frontend, backend, database, and APIs work together to make the experience possible.
Turn the plan into a working internal tool using Replit, Codex, or Claude Code—without prior coding experience.
Learn what bugs are, test real scenarios, connect Supabase, improve the tool, and define human approval.
Real work, not demos
Each example begins with repeated work, uses the right capability, and keeps a person accountable for the final result.
Computer Use
Repeated work
Open the conversation, gather the context, and prepare a useful response.
Human check
A person reviews and sends the answer.
Record & Replay
Repeated work
Repeat the Thursday page-filling process without rebuilding it each week.
Human check
The owner verifies every field before completion.
Scheduling
Repeated work
Check sales on schedule and prepare a consistent daily readout.
Human check
Sales reviews exceptions and chooses the follow-up.
Internal tool
Repeated work
Capture a repeated process, connect its data, and give it a clear interface.
Human check
The team owns the rules and approval points.
Join Discord, share the recurring work, and bring it to the live session. We will use real team processes to decide what to automate and what to build.