Remark raises $16M to build out human-powered expert models for e-commerce

Startups working on AI-powered e-commerce tools often rely on external data or user signals to build and improve their models. However, a company called Remark is taking a different approach by letting thousands of human experts chat with users while they are purchasing items, and then using that knowledge to train models.

Remark also makes those experts “available” all the time by having their AI-powered persona answer questions. Remark said that this resulted in a 10% net revenue gain for partners, which was a great sign for investors.

The startup on Tuesday announced $16 million in Series A funding led by Inspired Capital with participation from Stripe, Neo, Spero Ventures, Shine Capital, and Visible Ventures. This brings the startup’s total funding to date to $27 million. Remark plans to use the funds to grow its team of 25 people and spend on model training.

Remark joins a number of startups that are plugging AI into various areas of e-commerce, such as search, try-ons, chatbots, and dynamic page generation that is suited for AI-powered experiences. While these companies operate in different areas, merchants can only spend so much to enhance their online experience. In this market, Remark could have challenges in scaling and competing with other e-commerce startups using AI.

While Remark previously depended on a cut of the sale to generate revenue, the company has moved to a SaaS-based (software-as-a-service) model for better cash flow and now takes a fee based on site traffic.

The company’s CEO, Theo Satloff, said that physical stores tend to see high conversion rates of 30-35% compared to online stores that convert at roughly 1.5%. With Remark, the startup wanted to change that.

“The reason for low conversion rates on e-commerce sites is that people have a lot of questions when they shop. They have to go on Reddit or ask their friends to get advice about the purchase. In a store, a person with product knowledge helps them, so the conversion rate is high,” he explained.

“At Remark, we want to bring that kind of expertise to online shopping for our partners,” Satloff said in a call with TechCrunch.

When a user visits a site that uses Remark, the startup’s tool generates a set of dynamic questions that can drive the user’s purchasing decision. If a user is looking for pots and pans, the site could ask what kind of stovetop they have, or if they use diesel conduction or gas. When the user clicks on one of these questions, a chat interface opens where they’re matched with an expert who can offer advice about buying an item.

Remark matches an expert with a user based on skill set and locale, it says. If an expert is not available, the user is matched with an AI bot based on those experts who can help them with their shopping.

The company is slowly expanding its expert network. Satloff said Remark works with a variety of accreditation agencies to verify the expertise of the people who work with them. (This is akin to Airbnb adding various experience providers for its platform.)

Experts are paid for their time on the platform on a per-chat basis. They are also paid when their AI persona converts a sale. While the startup didn’t provide an exact figure for the average earnings that an expert receives, it said its top 20% earn $60,000-$70,000 per year by spending more than 15-20 hours a week offering advice.

The catch is that the human expert needs to be alert to requests from users. When a user activates a chat, experts get a notification on their phone on Remark’s app, and they have five to eight seconds to claim the chat and respond to the buyer.

Remark also dynamically rewrites the content on the website based on the user’s conversation with the expert and what they might be looking for.

The company has been growing quickly. Following its $10 million fundraise in 2023, Remark has seen 4x revenue growth. It also increased its network of experts from 50,000 to 60,000.

Kamran Ali, a Principal at Inspired Capital, said that Remark’s utilization of human insights was an important factor in the VC firm investing in the startup.

“As AI continues to take over the Internet, we’re going to continue to be inundated with just a ton of AI-generated content everywhere we go. The amount of AI-generated content we see today is dramatically more than a year ago, and that’s there’s no end in sight. So, having human insights and preferences will actually be at a premium, and that’s what attracted us to invest in Remark,” he told TechCrunch over a call.

Remark is now working on the ability to generate blog posts that recommend products based on expert conversations. Later, it plans to send personalized follow-up emails to users about a conversation they had with an expert.