Lab: The finance opportunity map

First lab, first artifact: the finance opportunity map. It's the automation-scoring discipline (frequency / rules-vs-judgment / error-cost) with the column that makes it finance: control impact — does inserting AI here preserve, strengthen, or threaten an existing control? Work Alder's list below, or better, your own team's month.

WORKFLOW                    HRS/MO  FREQ  RULES?  ERR-COST  CONTROL IMPACT
carrier invoice keying        340   high  yes     medium    neutral (keying
                                                            isn't a control)
3-way match + exceptions      120   high  yes     high      STRENGTHENS -
                                                            AI checks 100%
                                                            vs 5% sample
carrier rate audit             40   med   yes     high      STRENGTHENS
                                                            (new detective
                                                            control)
expense report review          60   high  partly  medium    caution - approver
                                                            independence
13-wk cash forecast assembly   32   med   mostly  medium    neutral
variance commentary            24   med   no      low-med   neutral (verify
                                                            numbers rule)
journal entry drafting         50   med   partly  HIGH      caution - posting
                                                            control is sacred
vendor master maintenance      15   low   partly  high      caution - ghost-
                                                            vendor risk lives
                                                            here

1. 1Score each workflow on the four columns (use Alder's draft as a starting point — two of Priya's scores are debatable on purpose; find and argue them). The AI-sorting prompt from earlier courses works here; add 'consider control impact: preserve / strengthen / threaten' to it.
2. 2Sort into the finance verdicts: automate the assembly (keying, forecast assembly — high frequency, rules-based, control-neutral), AI-as-detective-control (3-way match at 100%, rate audit — where AI adds control coverage humans couldn't afford), assist with gates (commentary, JE drafting — draft-only, named approver), and handle with care (anything touching approver independence or the vendor master).
3. 3Write the materiality line for each candidate: one sentence — 'worst realistic misstatement if this goes wrong unreviewed.' The exercise recalibrates at least one score every time someone does it honestly.
4. 4Pick the course-long candidates. The labs ahead build: the AP document pipeline (from 'invoice keying' + '3-way match'), the cash forecast (from 'assembly'), and the anomaly sweep (from 'rate audit' + 'vendor master'). If you're working your own inventory, pick your equivalents now — the capstone will want one of them.
5. 5Note the Meld moment: Priya's 'vendor master maintenance' problem — duplicate vendors under name variants — is entity resolution, the same discipline whether it's customers or vendors. Flag it; Module 4 shows why duplicate vendors are a fraud surface, not just a hygiene issue.