Fintok AIFintok AI

Curriculum · 71 lessons

From your first AI chat to a tool your firm can use.

Learn in five clear levels. You will practice on real estate work, check the result, and keep something useful from every lesson.

Showing 71 of 71

Level 1

Talk

Have a useful AI conversation.

You work in
Practice in a chat
You leave with
Prompt patterns and a simple playbook

0 of 10 complete

AI Literacy

Understand basic AI terms and separate facts from predictions.

Working Habits

Put saved context and instructions in the right place.

Level 2

Ground

Use source documents to make reliable work.

You work in
Chat, projects, and practice documents
You leave with
Checked templates and document workflows

0 of 23 complete

20.1Optional clinicCraft4 parts

Fix a Messy Rent Roll

Learn to correct a confident AI answer when a source row proves it is inaccurate.

What you prove: A reconciled rent roll with every change tied to its source row, reason, and corrected value.

Opens after Check a Rent Roll — Level 2.Opens with Pro
20.2Optional clinicJudgment4 parts

Defend an NOI

Separate sourced facts, assumptions, and questions that belong to counsel.

What you prove: A sourced NOI position that survives adversarial review, with its uncertainty named and the boundary decision escalated rather than answered.

Opens after Check a Rent Roll — Level 2 and Fix a Messy Rent Roll.Opens with Pro
20.3Optional clinicDiscovery4 parts

Find Portfolio Document Conflicts

Use rent-roll checks and document controls together to find portfolio problems during diligence.

What you prove: A contradiction register across a property stack, each finding tied to both the roll row and the governing document that disagrees with it.

Opens after Check a Rent Roll — Level 2 and Find Conflicts Across Documents — Level 2.Opens with Pro
22.1Optional clinicCraft4 parts

Recast a T-12

Support every T-12 change with a document and keep unsupported lines open.

What you prove: An adjusted-NOI note listing every reprice with its source document, and the real estate tax line routed to the seller as an open question rather than a number.

Opens after Check an OM Against Source Files — Level 2.Opens with Pro

Level 3

Judge

Make decisions and know what AI cannot decide.

You work in
Chat, evidence, and written rules
You leave with
Decision records another person can check

0 of 9 complete

30.1Optional clinicJudgment4 parts

Compare Loan Quotes

Check the sizing math and the deal plan before recommending a loan.

What you prove: A sourced comparison grid, one invented rate returned to an open question, and a recommendation tied to the deal plan.

Opens after Make the Decision — Level 3.Opens with Pro
30.2Optional clinicJudgment4 parts

Mark Up an LOI Within Your Role

Show exactly where your business judgment ends and counsel’s legal work begins.

What you prove: A marked-up LOI with the indemnification clause removed, an opposing markup, and a pre-counsel packet that names the legal boundary.

Opens after Make the Decision — Level 3.Opens with Pro

Level 4

Build

Build tools for work you do more than once.

You work in
A guided Claude Code workspace
You leave with
A working skill, script, workflow, or local app

0 of 24 complete

39.1Optional clinicJudgment3 parts

Narrow an Overly Broad Rule

Scope a rule by the claim it controls so people do not avoid the instruction file.

What you prove: A versioned CLAUDE.md whose verification gate fires on destination rather than on arithmetic, proven in both directions on the same two figures.

Opens after Save Project Instructions — Level 4.Opens with Pro
39.2Optional clinicSystem3 parts

Put Each Rule in the Right File

Separate a firm standard that applies everywhere from a decision that belongs to one deal.

What you prove: A root CLAUDE.md for firm rules and a deal file for asset rules, with the mixed tenant rule split between them.

Opens after Save Project Instructions — Level 4.Opens with Pro
51.1Optional clinicJudgment4 parts

Decide Whether a Tool Is Worth Building

Measure when a tool saves real time and when the manual process is safer.

What you prove: A cost and time comparison that includes manual work, the tool’s full run, its check, and one documented exception.

Opens after Build a Reusable Claude Code Skill — Level 4.Opens with Pro
55.1Optional clinicSystem5 parts

Check Another AI Tool’s Work

A reliable handoff includes evidence, a decision status, and an owner.

What you prove: An evidence-backed review and a pending approval register that survives a cold refresh.

Opens after Connect Skills into a Pipeline — Level 4.Opens with Pro

Level 5

Ship

Deploy a tool that coworkers can use.

You work in
A hosted application
You leave with
A live app and instructions for running it

0 of 5 complete

Your course tools

Practice and keep your work.

PracticeTry a fresh version of a lesson you finished.Saved asksFind the prompts and tools you chose to keep.GuidesOpen the checklists you earned in lessons.What we gradeSee what each graded request must include.