Sam Altman-backed Slope raises $30M for AI-powered B2B payments platform

Tracking and ensuring the integrity of payments is no easy task for any business, but in the B2B space, there is the added difficulty of ensuring the companies you are selling to will actually make their payments on time, won’t stiff you on some baseless excuse, go out of business, or try to wriggle out of them through clever accounting, bankruptcy, litigation, or other means. These issues are far rarer in the B2C space.

How can B2B companies, such as wholesalers selling products in bulk, ensure that they will be paid for the products and services they are rendering to their customers? Slope, a two-year-old AI startup founded in San Francisco, is attempting to create the gold standard: a B2B payments tracking and receiving platform that is powered in part by its own “rules-based” tech and partly by OpenAI’s GPT-3.5 Turbo. Slope is also developing its own proprietary, in-house large language models (LLMs).

The company today announced a $30 million equity round led by Fred Wilson’s famed Union Square Ventures with participation from OpenAI CEO and co-founder Sam Altman, for a total funding to date of $187 million. That’s hefty cash for a lean team of just 18 full-time employees.

“We run very efficiently,” said Lawrence Lin Murata, Slope’s CEO and co-founder, in a video call with VentureBeat. Lin Murata said he experienced firsthand the challenges B2B vendors face from working at the wholesale goods business operated by his parents in their home country, Brazil.

“For B2B businesses, it’s their lifeblood to be able to get money fast and send money to their vendors,” he added.

Slope’s technology informs the entire B2B customer payments journey

The company will use the cash to continue building out its team and technology — which includes an online payments and invoicing tool that Slope’s customers can use to accept payments from their customers, including credit card, ACH (automated clearing house), and international payments.

“We start from customer onboarding, risk assessment, and go all the way to reconciling, including everything in between,” said Alice Deng, co-founder and chief product officer of Slope.

That includes Slope analyzing a buyer’s “credit risk, invoicing, billing, cash application claims, and reductions and all the way to syncing into [a B2B customers’] accounting system — it’s all handled in our platform,” Deng added.

Slope further offers financing for its customers’ customers, allowing those who can’t pay upfront to be granted credit directly through Slope’s payments system.

It also delivers a newfound level of “visibility,” into B2B payment workflows that can be “old school” and more obscure, according to Deng.

One example of this newfound visibility is Slope Timeline, a feature that uses Slope’s understanding of a business’s transactions to keep the B2B vendor and their customers informed of payment and product shipping statuses in near realtime.

“Both the buyer and the seller understand exactly what stage there are in to the millisecond,” said Deng. That’s a marked improvement from businesses traditionally wondering, “oh, is my order open yet, did it ship? Did my wire transfer get reconciled?”

A foundation of ‘clean data’

Key to Slope’s approach to understanding, providing updated information on, and helping de-risk B2B payments for its enterprise customers is its focus on obtaining “clean data” from them.

“The foundation of that is clean data, it powers everything in the system,” Deng said. “We’re an AI company, but we’re actually a clean data company.”

To achieve “clean data,” Slope works with its enterprise customers to gather all of the data about the orders those customers are receiving, processing, and shipping out.

“That data is formatted and surfaced in ways that are useful to the customer within our platform,” said Lin Murata.

SlopeGPT leverages GPT in a novel way, turning enterprise transaction data into a fraud risk analysis tool

Here’s where some of Slope’s AI approach comes into play: Slope is able to assess a B2B buyer’s creditworthiness and their fraud risk, and extend them the optimal credit to incur the minimal risk on the seller/Slope’s customer, using SlopeGPT, a tool unveiled in April.

SlopeGPT takes a Slope enterprise customer’s transaction and purchase order data, runs it through a dedicated instance of OpenAI’s GPT (not on the public internet or accessible to third-parties), and clusters the data into embeddings that can determine which types of payments are regular and which are anomalous. Slope then uses these embeddings, and other rules-based data management techniques, to surface relevant data and suggestions to both its customers and their customers.

When analyzing a buyer’s risk for granting them financing/credit, SlopeGPT can also look for clues.

“If there’s anomalous activities, or somebody’s pretending to be a different business, if they’ve stolen information from another business…if they’re intentionally trying to show strong cash flow, and then do sudden transfers out, we can detect a lot of these anomalies to prevent payments,” Lin Murata said.

Slope discovered the power of GPT for this purpose by feeding it is own 2.5 million bank transactions over 18 months.

The company has also developed its own proprietary LLM — trained on public data — that performs even better at accurately identifying risk and will be released soon, according to the founders.