AI Is Rebuilding the Reporting Cycle
Andrew Lokenauth explains how AI transforms FP&A by automating monthly reporting while keeping human judgment essential for better financial decisions.
Andrew Lokenauth
Founder of The Finance Newsletter and Be Fluent In Finance
Career Shift: Andrew Lokenauth transitioned from finance roles at major banks to founding finance-focused platforms.
AI Impact: AI has transformed financial reporting, cutting cycles and reducing errors by automating data processes.
Workflow Efficiency: Implementing AI in finance compresses reporting timelines from days to hours with fewer mistakes.
Human Oversight: AI handles data tasks, but critical decisions still require human judgment and contextual understanding.
Data Challenges: Successful AI integration starts with addressing data quality and structuring before technology deployment.
Before founding TheFinanceNewsletter.com and BeFluentInFinance.com, Andrew Lokenauth held director and VP roles in finance at several well-known banks. Today, he is a thought leader in financial disciplines with 3M+ followers.
We caught up with Andrew to get a sense of what AI is doing to the industry — the risks and the rewards. Here’s what he told us.
Automating Before AI Was a Buzzword
I started my career at Citi, then spent time at Goldman Sachs, AIG, JPMorgan, and Signature Bank in roles spanning FP&A, strategic finance, investor relations, and corporate development. I was automating financial workflows with Python, SQL, and VBA long before AI became a topic. At Goldman, I cut days off P&L reporting cycles. At Signature Bank, I built systems that brought executive time-to-insight from days to under 24 hours. When generative AI emerged, it didn't feel disruptive. It felt like the natural next step in work I'd already been doing for close to two decades.
Continue Reading for Free
Create a free account to finish this article, plus get ongoing access to timely insights and practical resources.
Have an account? Log In
After leaving institutional finance, I founded TheFinanceNewsletter.com and BeFluentInFinance.com, where I write about finance and tech for 100,000+ newsletter subscribers and 3+ million social media followers. My insights on this topic have been featured in Forbes, CNBC, the Wall Street Journal, and Business Insider. I keep writing about it for the same reason I got interested in it: most of the conversation is still too theoretical, too vendor-driven, and too far removed from how finance actually works on the ground.
I also run a fractional advisory practice, where I serve as Fractional CFO and COO for high-growth startups and mid-market companies in the $1M to $5M revenue range.
How AI Changes Monthly Reporting
…a finance team would spend the first five to seven business days of every month pulling data, reconciling it, and formatting a deck that was already outdated by the time leadership saw it. Now that same insight is available within hours…The technology isn’t complicated. The willingness to redesign the workflow is what most teams are missing.
The biggest change in my work over the last 12 months has been replacing static monthly reporting packages with an AI-powered FP&A layer that sits on top of a client's core financial systems. It pulls from accounting software, CRM, and payroll, and runs automated variance analysis on a rolling basis.
Before this, a finance team would spend the first five to seven business days of every month pulling data, reconciling it, and formatting a deck that was already outdated by the time leadership saw it. Now that same insight is available within hours, with AI handling data aggregation and variance detection, and a human reviewing the output for context and judgment calls.
The result has been consistent across engagements:
- Monthly close cycles drop by roughly 30 to 40%.
- Board deck production goes from about 3 days to under 6 hours.
- Error rates fall because fewer manual steps mean fewer manual mistakes.
- Finance teams shift their energy from data gathering to actual business analysis.
- The quality of decisions across the whole business improves because the CFO can interrogate their financials in real time.
The technology isn't complicated. The willingness to redesign the workflow is what most teams are missing.
An AI-powered Close and Reporting Workflow
Here's the step-by-step breakdown of the close and reporting workflow:
- Automated data pull via Fivetran: A scheduled pipeline pulls data from accounting software, CRM, and payroll into a clean, centralized database every 24 hours.
- AI-powered reconciliation and variance flagging via BlackLine: The AI layer compares actuals to budget across every major line item, flags variances above defined thresholds, and generates a plain-language explanation of each flag.
- Human review: The finance lead reviews every flagged item, adds business context, and decides what needs escalation. They are assisted by FloQast AI for flux commentary.
- Dashboard and narrative output via Datarails: A formatted board deck and executive summary are auto-generated from the reviewed data, ready for distribution.
What used to take 3 days now takes 4 to 6 hours.
Real-world Examples of How AI Helps and Hurts Monthly Reporting
Here’s an example. I worked with a Series B fintech where we deployed an AI-powered cash flow monitoring system that flagged a client concentration risk six weeks before it became a liquidity problem.
The model picked up a pattern in receivables aging that a human reviewer had missed across the prior monthly closes. That single catch was worth more than the entire cost of the implementation.
On the negative side, I had a client whose AI-generated forecast was off by 15 to 20% in months with high invoice variability. The model was trained on historical patterns that didn't reflect a recent shift in revenue mix. The output looked polished and confident, and nobody challenged it until the CFO caught the variance in a board meeting.
A confident-looking output is not the same thing as a reliable one.
AI Owns the "What" — Humans Own the "So What"
The activities most suitable for AI are the ones with clear rules, historical patterns, and high repetition. I use AI for:
- Forecasting model runs
- Variance analysis
- Cash flow scenario modeling
- Anomaly detection
- Routine reporting
Where humans stay in the loop is on capital allocation decisions, risk judgment calls, contract negotiation, investor communications, regulatory interpretations, and anything that requires reading business context that isn't in the data. A model can flag that your burn rate is accelerating. It can't tell you that the reason is a founder relationship issue or a sales rep sandbagging pipeline.
The mental model I use is this: AI owns the "what," humans own the "so what."
This is one of the most practical ways to think about AI integration in a finance function. One without the other is incomplete, and both failure modes (over-automating or under-automating) cost organizations real money.
Why Over-Automation and Over-Confidence in AI Are Big Risks for Finance Teams
AI comes with a lot of wins. But the downside is real, too. AI creates a false sense of confidence when the underlying data isn’t clean. There’s also an over-automation trap…I think the field needs more honest reporting on this, and less vendor hype.
AI comes with a lot of wins. But the downside is real, too.
AI creates a false sense of confidence when the underlying data isn't clean. I've seen finance teams trust AI-generated variance reports without questioning source data quality, and that creates real risk.
There's also an over-automation trap: when you remove human review from too many steps, you lose the institutional knowledge that catches what models can't see.
I think the field needs more honest reporting on this, and less vendor hype.
Where AI Falls Short
When it comes to finance, AI is falling short in three main places.
First, I expected AI to be further along in compliance interpretation and regulatory judgment. In my years at Signature Bank and Amalgamated Bank, so much of financial controls work lived in nuanced judgment calls: reading regulatory guidance, anticipating examiner expectations, knowing when a number technically passes but still warrants a flag. AI tools today can surface relevant rules and summarize guidance documents. But they can't replicate the judgment that comes from years of sitting across the table from regulators.
Second, audit trail generation and explainability inside complex financial workflows. When I managed investor relations at Signature Bank for a $110B+ institution, every number that touched an SEC filing, an earnings call, or a regulatory submission had a documented audit trail and a clear human owner. AI-generated outputs, especially those produced by large language model-based systems, rarely have a chain of reasoning that would satisfy a regulator or external auditor. The output may be correct, but you can't always show your work in the way a compliance environment requires.
And third, early-stage companies with messy data infrastructure. The pitch for AI assumes reasonably clean, connected data. Most startups don't have that. Let's dive deeper into that.
Why Every AI Initiative Should Start with a Data Audit
Data quality is the real bottleneck. Not the model, not the tool, not the vendor. I went into my first AI-enabled finance project expecting the hard part to be the technology. It wasn't. The hard part was discovering that the company's financial data lived in three different systems, none talking to each other, with inconsistent naming conventions and months of missing entries. We spent roughly 60% of the project timeline on data cleanup before the AI tooling could do anything useful.
Every AI initiative should begin with a data audit, not a tool selection. Know what data you have, where it lives, how clean it is, and whether it's structured enough for a model to learn from. That single step would save weeks of rework and reset leadership expectations much earlier.
The old model was to hire more people to produce more insight. The new model is to build better data infrastructure, then let AI multiply your existing team's output. It's a different capital allocation decision, and most finance leaders haven't made it yet.
How Finance Teams Are Changing Due to AI
The biggest shift I've seen in teams is a compression of the junior analyst layer. Tasks that used to require 2 to 3 analysts doing data pulls, reconciliations, and basic variance analysis can now be handled by an AI layer with one person overseeing the output. That doesn't always mean headcount goes down. More often, the same headcount gets redeployed toward higher-value work: business partnering, scenario analysis, investor communication, and strategic planning.
The talent profile of a strong finance hire is also changing. The best people I look for now aren't just strong in accounting or financial modeling. They understand how to work with AI outputs, know when to trust a model and when to challenge it, and can communicate financial insight to non-finance stakeholders with clarity. That combination of financial rigor, AI literacy, and communication skills is rare right now and commands a premium.
What CFOs Must Learn to Thrive in the Next Decade
The CFOs who thrive in the next decade will look very different from the ones who thrived in the last decade. The technical finance skills (modeling, reporting, compliance) will increasingly be table stakes handled by AI. The skills that define great CFOs will be judgment, communication, strategic thinking, and the ability to lead through ambiguity.
My advice is to invest in the skills AI can't replicate. Build stronger relationships with business leaders. Improve your ability to communicate financial insight to non-finance audiences. Develop genuine strategic judgment, not just financial analysis skills. Those are the capabilities that will define elite CFOs in an AI-augmented future.
Four Steps Every CFO Should Take
Four things I'd tell every CFO right now:
First, do a data audit before anything else. Know where your financial data lives, how clean it is, and whether it's structured enough for a model to work with.
Second, start with one workflow. The CFOs I've seen succeed with AI pick the highest-pain, highest-repetition process in their function — usually the monthly close or the board deck — then redesign the workflow before automating it. That sequencing matters. Build the workflow on paper before you build it in a tool. Document every step in the process you want to automate, identify which steps are rule-based (AI-ready) and which require judgment (human-required), then build accordingly.
Third, define human review checkpoints before you go live. Every AI output in a finance workflow should have a human owner who signs off on it.
Fourth, measure the baseline before you start. If you don't know how long your current close takes or what your current forecast error rate is, you can't prove AI made it better.
These four steps sound straightforward, and they are. But most organizations skip two or three of them because they're eager to get to the technology. That's where most of the disappointment comes from.
Four things I’d tell every CFO right now: First, do a data audit before anything else. Second, start with one workflow. Third, define human review checkpoints before you go live. Fourth, measure the baseline before you start…These four steps sound straightforward, and they are. But most organizations skip two or three of them because they’re eager to get to the technology. That’s where most of the disappointment comes from.
Follow Along
You can follow Andrew Lokenauth's work on LinkedIn, YouTube, and his personal website. Or check out The Finance Newsletter and his BeFluentInFinance.com.
More expert interviews to come on The CFO Club!
