Pilot on the AI-driven ‘financial back office’

Many startups fail due to cash flow challenges. Keeping an eye on financials is important, but bookkeeping and back office tasks might not be every startup’s strength. Especially when operating under tight seed money and capital, employing an entire back office might not make sound business sense. This is where Pilot, winner of VentureBeat’s 2021 AI Business Application Innovation award, comes in.

Founder and CEO Waseem Daher said Pilot runs “the financial back office for startups.” Daher said the company handles bookkeeping, tax preparation, budgeting, forecasting, and other business needs. Pilot hires full-time, U.S. based employees who specialize in working with startups to form the back office.

The startup’s financial challenges

While all companies have to keep an eye on the balance sheet, a startup has special financial challenges. The early stages for example, demand tracking compliance with employment laws and tax returns and the cash balance and burn rate. As the company grows, founders might want to track and plan for strategic hiring and growth. “They might want to really get in the weeds of a forecast or a budget or work with a fractional CFO to make sure they have a plan and are tracking against that plan,” Daher says.

“The biggest challenge is making sure you’re putting into place the appropriate groundwork for the business stage you’re in,” Daher said. Pilot, which was founded in 2017, has offices in San Francisco and Nashville and a deep bench of experts that keep an eye on such groundwork.

A little help from AI

The team at Pilot uses AI strategically. AI helps with automated visibility of errors and their management, predictive insights and context-specific reporting. The company anonymizes and aggregates financial data from the more than 1,000 startups on its rolls to create supervised machine learning models that can detect, for example, anomalies and unexpected variance in balance sheets. Vendor overpayments and tracking overdue invoices are just two of the many routine bookkeeping tasks AI models keep track of.

AI provides automated visibility, which means it flags non-traditional behavior and spending, while keeping aggregated charts of what spending (the burn rate) should look like. Context-specific insights are key, which means Pilot’s experts sift through the recommendations to separate the signal from the noise. The models learn as they go along.

Daher sees the AI models becoming increasingly sophisticated as analysis can be sliced and diced according to a number of variables such as geography — how do you fare with respect to other startups in your area — or growth stage or industry.

The challenges, Daher said, is in making sure to use AI where it shines best: on big data analytics and the routine gruntwork of making bookkeeping follow rules. AI is a workhorse and should be treated as such. The stars are still the human employees and the customers, Daher said.

Avoiding overfitting of models is another trap that Pilot makes sure it doesn’t fall into. For example, Daher pointed out, some startups might consider LinkedIn a recruiting expense while others might categorize it as a sales expense. Models need to be aware that one data point does not make a rule. “You may end up with an overly simplistic view of the world that may not always be accurate,” Daher said. “You have to design AI models with that sort of generality in mind.”

“Our engineers have built a ton of software to help our team do the work more accurately, more reliably, and more consistently. The AI is really the Iron Man suit for our internal finance teams,” Daher said. He expects Pilot to move to more unsupervised learning models in the future, which can yield aggregate insights in a scalable and automated fashion.

But, Daher added, “it’s not about using AI just for the sake of using AI. It’s about figuring out what’s going to yield the best possible customer experience and working backward from there.”