To Buy or Not to Buy: FunnelAI Studies Consumers’ Intent to Purchase
San Antonio—Sridhar Kamma is the kind of car enthusiast who would meet up with other devotees a couple times a month to talk about the merits of a BMW M3 versus a Porsche Boxster, a Mustang versus a Camaro. He’s the type to post questions in online forums and blogs about where he could find a part for the classic BMW E24 he was restoring.
At least he was, before he started bootstrapping a startup. Since he co-founded FunnelAI, Kamma has traded in his M3 for a more sensible and cost-effective daily driver, a Mazda 3. He also spends more time selling FunnelAI’s artificial intelligence software, which can pour over online forums and social media posts in pursuit of consumers who might want to make a vehicle purchase, than he does reading about about his interest.
FunnelAI’s Python-based code aims to study the intent of an online comment, such as a social media post or a message in a Reddit forum about car repairs, to see if it can interpret whether the consumer might be interested in purchasing a car or possibly be in need of a service repair. The company’s AI software will flag posts that might be potential leads for customers based on the structure of a consumer’s sentence. Hits are given to FunnelAI’s customers, who can reach out to the customer directly to offer their service.
“We find strong purchase intent in these posts, and connect those people with businesses,” Kamma says. “As it learns, it gets better and better.”
FunnelAI charges a monthly fee of $2,400 for the leads it gives its customers. The company has started to find some traction, and has reached $1.2 million in annual recurring revenue, Kamma says. Anecdotally, Kamma says auto dealers that FunnelAI has sold its service to have reported selling an additional 20 to 30 cars per month.
FunnelAI is still early in its development, however, and is looking to raise a seed round of funding of about $3 million, Kamma says. The company previously received $125,000 by participating in the RealCo accelerator program, and added a $375,000 round of funding from Geekdom Fund, Active Capital, and a few angel investors.
Other businesses manage or analyze social media posts and other online content for businesses, from Spredfast to Brandwatch. The ability of FunnelAI’s AI software to discover purchase intent is what makes the business stand out, Kamma argues. FunnelAI is trying working to patent the specific use of artificial intelligence for purchase intent, which is still being processed. The company’s engineers train and have been training its software by labeling the posts it correctly flags, like “My car needs too many repairs, time to get a new one,” as a match and labeling incorrect flags, such as “Are you looking to buy a new car? Come by our dealership,” as a miss.
FunnelAI is initially eyeing the automotive industry, focusing first on selling to car dealerships and eventually expanding its focus to original equipment manufacturers and repair shops. The company also is starting to offer the service to the real estate market, and plans to expand into other verticals at some point, like the insurance sector.
Before moving into the startup world, Kamma studied artificial intelligence and data systems during a master’s degree program in Scotland and he then took a job in Dublin at Austin-based National Instruments. That job eventually brought him to Austin, where he bought his first BMW. Kamma moved to San Antonio to join the RealCo accelerator.
In a time where AI is increasingly becoming a buzzword used by executives in every sector from healthcare to automobiles, Kamma says FunnelAI’s software is based on statistical modeling and natural language processing that builds a dictionary based on the positioning of words in a sentence. Just like a human is taught the positions of words matter in the English language, so too is this software.
“It basically gets as good as it’s trained,” Kamma says. “We cannot train it all at once so that is where deep learning comes where the AI can continue to understand based on previous learnings.”