Introduction

If you are a CFO, finance leader, or someone trying to understand what AI in finance actually looks like in practice, this conversation is for you.

In this edition of our Fireside Chat series, "The Dream AI Agent for a CFO," we sat down with Lin Chan, CFO at rce.ai, a seasoned finance executive with decades of experience spanning startups, growth-stage companies, and large multinationals across cleantech, semiconductors, manufacturing, and SaaS. The session was moderated by Niyati Chhaya, Co-founder and VP of AI/ML at Hyperbots.

The conversation explored what a CFO's work really looks like day to day, where the biggest friction points are, and what an ideal AI-powered workspace would actually do. Lin brought a practitioner's candor to every question, covering everything from strategic decision-making at large enterprises to phishing scams that cost her company real money, and her vision of an AI assistant that wakes her up with the answers already prepared.

You can also watch the complete fireside chat with Lin Chan here: https://www.youtube.com/live/cvExNdiKWlA?si=I75bAeY2nPitn1PV

Key Takeaways

- A CFO’s priorities change with company stage, but cash visibility and sales forecasting remain the two numbers that determine survival.

- One spoofed email cost Lin’s company $60,000 in minutes, showing why fraud detection is no longer an IT problem but a finance priority.

- Most finance teams still manually move data between QuickBooks, Excel, and forecasting tools, creating hidden inefficiencies and avoidable errors.

- CFOs are expected to influence pricing, contracts, legal risk, hiring, and growth strategy, not just report financial numbers.

- AI becomes valuable when it delivers decisions already analyzed, anomalies already identified, and actions already prioritized before the CFO starts the day.

- Vendor analysis is still painfully manual, with finance leaders comparing pricing, logistics, payment terms, and specifications document by document. 

Summary of the Conversation

What Does a CFO Actually Do Every Day?

Niyati: What do the daily tasks of a CFO look like?

Lin: It depends on the size and phase of the company. For a pre-revenue startup, the number-one focus every day is cash. How much is in the bank, what payments are due, how long the runway lasts. Decisions have to be fast because there is very little structure. You are essentially making crisis decisions most of the time.

When you move to a Series B or C stage, around $20 million or more in revenue, the two big priorities are still cash and sales. How long does the runway last? You want at least two years of financing visibility. For larger companies, whether mid-sized up to a billion in revenue or beyond, cash is still critical if you are heavily leveraged. For everyone else at that scale, the lens shifts toward strategic performance and longer-horizon forecasting.

How Does Company Stage Shape the CFO's Weekly and Monthly Rhythm?

Niyati: How does that translate into weekly, monthly, or quarterly routines?

Lin: For a pre-revenue company, cash is reviewed almost weekly. For a well-run Series B or C company, it is usually a monthly look at overall forecasting. For large companies, at minimum monthly. But on a day-to-day basis there is always a mix of contract decisions, vendor negotiations, and operational calls that all feed into the broader financial picture.

The more successful you become at a startup, the less you end up doing yourself. Your role gets more defined and more strategic. Once the process has been set up for finance and accounting, it tends to run on its own, and you focus on the bigger decisions covering HR, treasury, insurance, legal, and financial strategy.

Which Metrics Matter for CFOs, and Why They Differ by Industry?

Niyati: What are the key metrics you track beyond cash and sales?

Lin: It depends entirely on the type of company. For SaaS businesses, I look at customer acquisition cost, churn rates, and revenue stickiness, specifically what the fall-off looks like. For manufacturing, I focus on labor and material variances, use rates across product lines, and working capital ratios. Inventory management is critical there. For retail and product companies, collections and accounts payable across potentially thousands of vendors become the central challenge.

When you are dealing with a department store situation, for example, you are juggling thousands of SKUs and thousands of vendor payment lines simultaneously. It is a constant operational juggle, and it is very different from managing a SaaS renewal schedule where most collections happen upfront.

How Are Approval Processes and Financial Controls Structured?

Niyati: How involved are you in payment approvals and financial controls? Is that a daily task or more of a policy-setting role?

Lin: For small companies, I am involved in almost every payment. For larger organizations, I set the policies covering approval thresholds and multi-level sign-off requirements, and then my role becomes periodic review and audit rather than daily execution. I always define approval limits based on company size and cash flow volume. For some organizations the CEO threshold is a million dollars or more. For others it might be just a few thousand, depending on the business.

The bigger the company, the more my job becomes stewardship and strategy rather than transactions. Once a solid process is in place, accounting tends to run on its own and I focus on higher-level decisions. The CFO's job actually becomes more complex the larger the organization, even as it becomes less transactional.

How Deeply Is the CFO Involved in Business Strategy and Pricing?

Niyati: Are you involved in business model discussions, things like pricing and target market?

Lin: Absolutely. The CFO has to be. If the CEO makes a major decision without consulting finance, the outcome can be disastrous. For example, if a CEO intends to file a patent infringement lawsuit, the CFO needs to confirm the company has the cash flow to sustain that, because litigation can cost millions of dollars and drag on for years.

Pricing is another area where finance must be central, not just for revenue impact but for revenue recognition implications. I have worked with companies where analyzing the financial impact of a large vendor contract against prior projections took significant time and resources. The CFO is often the one who catches what the contract actually says versus what everyone assumed it said.

I see the CFO almost as a second person to the CEO. Sales, marketing, and R&D all have to move in the same direction, and the CFO is the one connecting financial reality to that shared direction. Even in a public company context, you are working very closely with sales to make sure you hit your numbers and can present a credible story to analysts.

Why Financial Fraud Is a Day-to-Day Threat, Not a Rare Event

Niyati: You raised concerns about phishing and cybersecurity. Can you share what you have personally experienced?

Lin: Two incidents hit me in the past year or two, and both were significant.

In the first case, my controller received what appeared to be an email from our CEO. Before we had any formal approval policies in place, he wired $60,000 and only realized five minutes later that the email was a spoof. Insurance covered about 90% of the loss, but the exposure and embarrassment were very real. If money is involved, you are a target. It does not matter how small the company is.

The second case was more sophisticated. My attorney was processing a settlement payment to me. A scammer had intercepted our email exchange at the attorney's network level. My emails to her were silently deleted, and her emails to me were rerouted to the attacker. The scammer then instructed her to wire approximately $50,000 to $60,000 to a fraudulent account. By the time we understood what had happened, the funds were gone. The attorney filed a police report and recovered about 90% of it, but not before serious disruption.

What made that second case especially difficult to catch was that my name was spelled correctly in the email body, but the reply-to address had "CHEN" instead of "CHAN." Nobody clicks into every link in a long reply chain when they are processing dozens of messages a day. That is exactly what scammers count on. I would even receive invoices with the CEO's email attached claiming he had approved the payment, and when I asked him directly, he had no idea what it was.

What I would love is an AI agent built for fraud detection that flags any inbound payment request not matching a known, approved vendor, and catches the spoofed email address before it ever reaches the CFO's inbox.

What AI Tools Are Being Used Today by CFOs, and Where They Fall Short?

Niyati: What AI tools are you currently using, and what for?

Lin: We use ChatGPT for research, things like market sizing, opportunity analysis, and preliminary ROI framing. It is useful for getting started, but it cannot access our internal data, does not know our vendor agreements, and requires us to manually feed it everything. For presentations and some graphs, AI tools help with formatting and initial drafts.

But the core financial work, forecasting, variance analysis, and document review, is still largely manual. One of the biggest time sinks for a small company is the data handoff: downloading from QuickBooks, reformatting in Excel, then uploading into a forecast model. Every manual step is a potential error. At larger companies I have set up Oracle with Hyperion or similar planning tools where data flows automatically via API, eliminating that entire chain. Most small and mid-sized finance teams are still stuck in that manual loop.

Rolling forecasts are an important part of any finance department. You update with actuals, look at what has changed, and track the trend. The FP&A group handles the analysis side, looking at where the data is heading. The problem is that when you are deep in the data transactions, you sometimes lose the high-level view of where things are going.

How CFOs Manage Document Storage, Vendor Analysis, and Financial Insights

Niyati: How do you store and access key documents today, and how do you generate insights from them?

Lin: For a company our size, everything lives on Google Drive, covering legal agreements, vendor contracts, employment records, purchase orders, and financial reports. I like to maintain a proper data room with clear organization: one section for HR, one for vendors, one for finance and sales. The CEO and CFO typically have access across all of it.

When I need to verify something from a document today, I open it and read it. All document-level insights are manual. What I would love is an AI that can compare vendor proposals automatically. Right now, if I receive several quotes for a manufacturing source, I have to read every single one manually, align the specs, compare logistics and delivery terms, and evaluate pricing side by side. A vendor management co-pilot that does that comparison instantly, across all dimensions, would save enormous time.

What the Dream AI-Powered CFO Workspace Actually Looks Like

Niyati: If you had an AI-powered workspace that you opened first thing every morning, what would you want it to show you?

Lin: Every morning, the data has already been pulled, processed, and analyzed. I wake up and the AI has already done the thinking for me. That is the core feature, not more data but the right insights derived from the data.

For a company with sales, I want to see what the numbers look like against forecast, what contracts have been signed, and where the cash stands with forward projections. For a SaaS company, I want to see the waterfall. For manufacturing, I want the variance view. Then I want to know about AR: who owes what, what is overdue, and who needs a follow-up call. I also want visibility into salesperson performance, especially in early-stage companies where tracking ramp-up against targets matters a lot. And I want anomaly flags, anything that looks off, explained in plain language with a suggested next action.

On the communications side, I would be comfortable with AI scheduling meetings or sending a nudge on my behalf, something like "Lin would like to connect about the Q3 variance, can you find time this week?" But the actual conversation has to happen with a human. When something is off, you need to read the room, ask the right questions, and understand the real reason behind a number. The analysis is what I want AI to own. The judgment call is mine.

Why Human Judgment Still Matters for CFOs in AI-Driven Finance Operations 

Niyati: Would you want AI to take over communications and follow-ups entirely?

Lin: For scheduling and reminders, yes. But a variance conversation or a collections follow-up cannot be fully handled by AI. You do not always know what is behind a number until you have the conversation. The best way is to get on a call and ask directly. How come the use for this product line dropped? What is the situation and how is it being resolved?

The proactive people on your team will already flag the reason when they see a variance, and sometimes you already know the answer. But you still want to pick up the phone to ask how the situation is being resolved and whether the team has what they need to fix it. That is the relational and interpretive layer that cannot be delegated to a machine. Everything upstream of that judgment call, the data gathering, the analysis, the anomaly detection, the scheduling, absolutely can be.

I sometimes think of the CFO role as being like the mother in the family, always making sure the emotional and financial health of the company is on track, that each part of the organization is growing in the right direction, and that the warning signs get caught before they become crises.

How Hyperbots Helps CFOs Automate Finance Operations and Improve Decision-Making 

Finance leaders like Lin spend too much of their day on work that should not require their attention at all. Manually pulling data from QuickBooks, re-entering it into Excel, reading through vendor proposals one by one, chasing down suspicious invoices that should have been flagged before they landed in the inbox. That is the real cost hiding inside most finance operations, and it compounds every month.

Hyperbots is built to remove that cost at the source. The Invoice Processing Co-Pilot delivers 80% straight-through processing and 99.8% extraction accuracy, so the manual review cycle that consumes finance team hours gets handled automatically and with full auditability. Fraud detection runs continuously in the background, flagging payment anomalies and vendor mismatches before they reach the CFO's desk, exactly the kind of protection Lin described needing after her own fraud incidents.

The Vendor Management Co-Pilot replaces the manual proposal comparison process with AI-driven extraction across specs, pricing, logistics, and payment terms, turning a multi-day exercise into minutes. Vendor onboarding, compliance checks, and ongoing communication are all handled within the same workflow, giving the CFO a clean, organized view of every supplier relationship without the administrative overhead.

Across the broader finance operation, Hyperbots reduces accruals processing cost by 80% with less than 5% variance between accrued and actual costs. On collections, finance teams see a 40% improvement in DSO and a 70% reduction in collections cost, freeing the team to focus on the analysis and strategy work that actually moves the business forward.

The CFO that Lin describes, someone who wakes up to insights already prepared, who spends the morning on decisions rather than data gathering, who catches fraud before it happens and vendor issues before they escalate, that is what Hyperbots makes possible today.

See it in action with a demo or start your free trial today.