Course 11 · Firm Operating System
Intro to Codex: Build Your AI Bookkeeper
Learn Codex from zero while building a working AI bookkeeper capstone.
Coming soon
—
This course is in the planned curriculum. Available courses are open now.
Browse available coursesAlready a member? Log in
What this course covers
Learn Codex from zero by building something real: an AI bookkeeper for your firm. No technical background required — from opening Codex the first time to a working agent under your supervision.
You will write the agent's job description and firm rules, connect it to LedgerHQ and the firm's tools, run the first work queue, and install the improvement loop that makes the agent better every week.
What you leave with
A Codex project agent that can run bookkeeping routines with firm rules, tool routing, review, and supervisor oversight.
Included with the course
- Full video lessons taught from real firm workflows
- Module resources, worksheets, and implementation notes
- A concrete next step for your firm after every module
- Lifetime access after you open the course — no subscription required
Syllabus
9 modules
Preview the full module list here. Lessons, implementation notes, and downloads open inside the Academy after you get started.
- 01
Open Codex: Your AI Staff Workstation
Open Codex, understand projects, threads, files, review, modes, and why firm work belongs in a project instead of a disposable chat.
~30 min
- 02
Create The AI Bookkeeper Project
Build the file map that makes the agent useful: instructions, firm rules, routines, templates, maps, memory, outputs, and automations.
~30 min
- 03
Write The Agent's Job Description
Create AGENTS.md as the bookkeeper operating manual: role, source of truth, startup, authority, daily loop, owner briefings, and done-when.
~30 min
- 04
Prompt Like A Firm Owner
Delegate work with goal, context, constraints, and done-when; ask Codex to inspect files, save outputs, and improve instructions.
~30 min
- 05
Connect Codex To The Firm's Tools
Explain MCP and app connectors plainly, then route LedgerHQ, Gmail, Drive, Slack, Calendar, and workflow systems to the right source of truth.
~30 min
- 06
Teach The Agent The Daily Bookkeeper Loop
Build daily queue, bank-feed, statement watch, reconciliation, reporting, owner briefing, and instruction improvement routines.
~30 min
- 07
Run And Review The First Bookkeeper Work Queue
Run the first manual queue, classify work, review the owner briefing, catch assumptions, and update one instruction.
~30 min
- 08
Add Authority, Batches, And Automations
Add delegated authority, client batches, daily automation, nightly batch automation, monthly close cycles, and supervisor review cadence.
~30 min
- 09
Make The Agent Better Over Time
Improve the agent through model choice, planning, review, memory, skills, subagents, instruction updates, and a next-agent backlog.
~30 min
More of the operating system is on the way.
Browse the courses that are available now while this one is Coming Soon.
