Skip to main content

The C.L.E.A.R. Framework

C.L.E.A.R. is the operating model for CFO Stack: Capture, Log, Extract, Automate, Report.

C

Capture

Pull every raw money artifact into one reviewable queue.

Where is every piece of evidence of money right now?

  • /capture
  • /statement-export
  • /statement-export-private
  • /capture-dedupe
  • /doc-preprocess
  • /bank-import
  • /receipt-scan
L

Log

Convert evidence into reviewed Beancount entries.

Do I know where every dollar came from and went?

  • /log
  • /classify
  • /validate
E

Extract

Pull patterns, anomalies, and tax signals out of the books.

What are the numbers telling me to do next?

  • /extract
  • /reconcile
  • /tax-plan
  • /consult
A

Automate

Turn repeated workflows into scripts, pipelines, and checks.

Have I repeated this enough times to automate it?

  • /automate
  • /monthly-close
  • /quarterly-tax
R

Report

Produce decision-ready outputs instead of raw ledger noise.

Can I summarize financial health in one paragraph?

  • /report
  • /fava
  • /advisor

Overview​

CLEAR is a five-step framework for AI-powered financial management. Each step builds on the previous one, creating a complete cycle from raw evidence to structured books, operational insight, automation, and decision-ready reporting.

C — Capture​

Definition: Consolidate every piece of evidence of money into one place.

Typical inputs

  • Bank CSV exports
  • Credit card statements
  • WeChat and Alipay bills
  • Receipt photos
  • Invoice PDFs

Core question: Where is every piece of evidence of my money right now?

Skills: /capture, /statement-export, /statement-export-private, /capture-dedupe, /doc-preprocess, /bank-import, /receipt-scan

L — Log​

Definition: Transform captured data into structured double-entry records.

What happens here

  • Raw CSV becomes cleaned, deduplicated transactions
  • Transactions become Beancount entries
  • Classification is AI-assisted instead of manual copy-paste

Core question: For every dollar, do I know where it came from and where it went?

Skills: /log, /classify, /validate

E — Extract​

Definition: Distill action from the books instead of stopping at raw numbers.

Typical analysis

  • Spending pattern analysis
  • Anomaly detection
  • Trend forecasting
  • Tax preparation summaries

Core question: What are these numbers telling me? What should I do next?

Skills: /extract, /reconcile, /tax-plan, /consult

A — Automate​

Definition: Turn repeated logging and extraction work into scripts and repeatable pipelines.

What automation should target

  • Code-generated processing scripts
  • Scheduled close routines
  • Alerting for validation errors or missing evidence

Core question: Have I done this more than three times? Can a machine do it instead?

Skills: /automate, /monthly-close, /quarterly-tax

R — Report​

Definition: Produce a clear financial picture for decisions, review, and filing prep.

Typical outputs

  • Fava visual reports
  • The three core financial statements
  • GST/VAT filing packets
  • Plain-English summaries of financial health

Core question: Can I describe the current financial health of the business in one paragraph?

Skills: /report, /fava, /advisor

Why the stages matter​

Without capture, the ledger is incomplete. Without logging, the evidence is not queryable. Without extraction, the books stay inert. Without automation, the system does not scale. Without reporting, the numbers never turn into decisions.