In a typical group, finance is without doubt one of the most necessary capabilities. But groups are sometimes slowed down by guide workflows. In line with a survey by Paylocity, an HR software program supplier, 38% of finance groups spend greater than a fourth of their time on guide jobs, like reviewing invoices.
Matthieu Hafemeister, an ex-fintech investor at Andreessen Horowitz, says he’s seen many finance orgs wrestle to scale up on account of all of the work they’re doing by hand.
“The established order for finance is numerous level options which might be cobbled collectively inside the finance division,” Hafemeister informed TechCrunch. “Excel continues to be the bottom widespread denominator, limiting the promise of automation.”
To Hafemeister’s level, most finance departments are certainly closely reliant on spreadsheets. One survey discovered that 82% nonetheless use Excel recordsdata for budgeting, forecasting, and different core monetary planning actions.
After experiencing these frustrations firsthand whereas main progress at fintech agency Jeeves, Hafemeister determined to workforce up with Ted Michaels, Jeeves’ earlier head of finance and an previous buddy, to launch a platform to automate monetary duties.
Referred to as Concourse, the platform connects to a companies’ monetary programs to let finance groups retrieve and analyze knowledge, generate charts, and ask ad-hoc questions equivalent to “What’s our non-GAAP income?”
“Concourse can proactively floor insights that permit finance groups to be higher ready by enabling them to remain forward of developments,” Hafemeister mentioned. “As an alternative of a instrument that tries to enhance the velocity or effectivity of finishing a process, Concourse will be given discrete duties to do fully by itself.”
Now, finance automation isn’t precisely new know-how. Linq lately emerged from stealth with AI to automate facets of analysis for monetary analysts. Ledge and Doopla are additionally constructing a spread of finance-specific generative modeling instruments.
However what makes Concourse totally different, in keeping with Hafemeister, is its skill to execute monetary workflows with “advanced, multi-step operations.” For instance, the platform can retrieve knowledge from an organization’s NetSuite dashboard to obtain CSV recordsdata, then copy that knowledge to an Excel spreadsheet.
“We leverage giant language fashions to do what they’re finest suited to and pair them with extra conventional strategies of knowledge evaluation,” Hafemeister defined.
There’s nice curiosity in AI for finance. One ballot discovered that 58% of finance groups are actually utilizing some type of AI know-how, up 21% from 2023. Grand View Analysis estimates that the “AI in fintech” section, value $9.45 billion three years in the past, is rising 16.5% yearly.
However to face an opportunity of constructing a dent available in the market for finance automation tech, Concourse must display its product’s ROI — a difficult feat. Per Gartner, exhibiting or estimating the worth of AI is a high barrier to adopting it for near half of corporations.
Concourse can even need to assuage potential clients’ fears of AI-introduced errors and hallucinations. In a ballot of U.Ok.-based executives by HR specialist Peninsula, 40% mentioned inaccuracies from AI instruments had been a key concern, adopted by issues round knowledge confidentiality.
Hafemeister mentioned that Concourse employs “a wide range of instruments and strategies” for fact-checking and validation to strive to make sure its AI performs duties as supposed. He added that Concourse doesn’t use corporations’ knowledge to coach its AI fashions — at the least not with out specific permission — and that the platform solely collects knowledge clients share with it.
“Information accuracy is paramount in finance, the place solutions are usually both fully appropriate or fully incorrect,” Hafemeister mentioned. “As such, at Concourse we’ve spent a number of effort and time on delivering AI that may precisely carry out the duty it’s been assigned. We additionally take knowledge privateness and safety very critically, and have constructed Concourse utilizing trade finest practices.”
Of us appear keen to be take Hafemeister at his phrase.
Concourse, which remains to be in beta forward of a broader launch deliberate for subsequent yr, has a number of clients, together with Instabase and Shef, and $4.7 million in capital. Hafemeister’s ex-employer, a16z, has invested within the startup, together with Y Combinator, CRV, and Field Group.
Hafemeister says the main target in the intervening time is product growth and rising New York-based Concourse’s six-person employees.
“We raised cash to rent extra engineers, construct out extra workflows that our AI can tackle, improve protection on knowledge integrations, and begin to scale our go-to-market operate,” he mentioned. “The robust concentrate on engineering recruiting is to rent backend, machine studying, and AI engineers.”
