Meridian Capital Services · FP&A · Engagement point of view
From Normalized Pain to Sparking Acceleration and Securing Lasting Trust
A better way to do F&P, for analysts
Problem
PE mandate in. Claude licenses for everyone. A one-hour orientation. Adoption near zero, by a measure no one has defined yet.
Plan
Find the tax the team stopped noticing, take it off work they have already proven, and let AI earn trust one Try at a time.
Outcome
A team that does more, better and faster, and an “AI-enabled finance” the sponsor can measure.
01 · The actual problem
The Invisible Tax on Financial Judgment
“I just don’t see how it applies to what we do.”
People get really good at what they do, in part because they develop a world-class tolerance for the tediousness and annoyance of some of the things their work involves.
So the tax stops being visible. It just looks like the job.
02 · Why adoption stalled
Tools without Clarity — No Traction
Experimenting alone doesn’t show us where to start, let alone how to get there.
Teams committed to sustained intensity
- There are things to do
- Barely enough time to do them
- And it doesn’t stop
…are precisely the ones that benefit the most from AI’s
- Assistancecloses a gap of understanding
- Awarenesscloses a gap of situational insight
- Autonomycloses a gap of laborious effort
So the definition writes itself
The measure of that benefit is exactly the measure of an AI-acceleration, and what unmistakably defines AI-enablement.
Show me a team caught up in sustained intensity, and I’ll show you a team that needs AI enablement, and will give you good reasons to be falling behind on that.
03 · Out thereIn here
And Success Is Happening
Especially in finance: from numbers to judgment.
Microsoft Finance · Data reconciliation
a week per analyst, with Copilot for Finance
Endeavour Energy · Forecasting
forecasting time; accuracy up from 92% to 98%
Aurorium · General-ledger review
of GL combinations where AI signals flagged anomalies
Air · Recurring reporting
less time on specific tasks, measured by its Director of Finance
NBIM · Investment management, Claude Enterprise
of time saved weekly per employee on AI-assisted analytical work
Customer results as published by each company or vendor (Microsoft 365 blog; OneStream, Planful, Vena and Claude customer stories). Signal, not benchmarks.
Draft last month’s variance commentary~30–50% less drafting time
Blank-page writing, and they hold the answer key. Writing studies: 40% faster, 18% higher quality.a
Tie-out check on the live closeEarlier catches
Before anything ships, a catch is a save. Errors in 86%+ of audited spreadsheets.b
Draft the questions to budget ownersFewer back-and-forth rounds
Communication, not judgment. The analyst edits before anything is sent.
Reshape exports into the templateThe most raw hours
Pure assembly; totals must re-tie, so it pairs with Try 2. Only 35% of FP&A time goes to high-value work.c
Gains are our suppositions; each Try measures the real number. a: Noy & Zhang, Science 2023 · b: Panko, “Spreadsheet Errors: What We Know” · c: FP&A Trends Survey 2024
Rule for every Try: the first thing AI catches must make the analyst look good, and never expose anyone.
04 · The first 30 days
30 Days — Spark to Trust
A plan doesn’t get attended. A commitment does.
Simple enough to say yes to on Day 0. Fast enough to feel by their next close.
- Present the one-page plan
- Negotiate the calendar
- Secure champions
- Working sessions, one per champion
- Observe a close task
- Sit in on their close
- Workshop: first pass on their current state
- Select one workflow
- The ContAIner: skills + cognitive layers
- Prompt library & reference
- Champions Try it on real work
- Measure: drafts kept, hours saved, use ramp-up
Collected every cycle
- Drafts kept, modified, tossed
- Hours saved per workflow cycle
- Induced use ramp-up
05 · The discovery motion
The Discovery Motion — Spark
The other part: they got good because they push. Same work, new way.
One recent task, step by step: last month’s variance commentary.
Label every step: judgment or assembly.
AI drafts a first pass from their own numbers for that task.
The analyst corrects, rejects or approves. They hold the answer key.
The workflow to accelerate first: lowest risk, highest gain.
Insight worth digging for, in the same session
How is adoption measured today?
Which parts of the work do they enjoy most, and least?
Which results are they proudest of?
Which tools and process details are they sensitive to?
Analysts see a way to improve what they like doing best, and take the weight off what they don’t. That pushes them forward.
06 · What stays
Lasting Trust — Theirs to Own
We leave in 30 days. The acceleration doesn’t.
1 The definition & its significance
- “AI-enabled finance” in their words: their benefit = their AI-acceleration = their AI-enablement
- Measured by what they collect: drafts kept, modified, tossed · hours saved per cycle · induced use ramp-up
- The leader takes it upstairs: a measure, not a slogan.
2 The impact & its scale
- Their map, current to future state: a backlog they grow without us
- The ContAIner: one AIly per analyst, skills and cognitive layers that keep learning their craft, even when people change
- Champions teach the next Try: one workflow, then four Tries, then the whole map
Why they will
- They judged every draft. Trust was earned on their own answer key, not granted.
- They chose the workflow. The map and the backlog are theirs.
- Their wins carry their names upstairs. The leader reports them.
“AI-enabled finance” started wide open. Now it’s written in their numbers, by them.
Thank you. Q&A, then the live walkthrough.