Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
IndustryInsight · On RadarHabr AI→ original

SuperMe raises $20M for AI clones of experts who answer in a single chat

Ask ChatGPT about careers or marketing and you will usually get an averaged answer. SuperMe offers something different: AI profiles of real specialists, built from their posts, interviews and working notes, which answer in one chat and even argue with each other. The company has raised $20M to attract both users and the experts themselves. For product people, it is a test of an idea: the value of an AI service may lie not in the model, but in people's unique data.

AI-processed from Habr AI; edited by Hamidun News
SuperMe raises $20M for AI clones of experts who answer in a single chat
Source: Habr AI. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceHabr AI ↗
Editor-in-Chief
Key metric #1The round was led by Kirsten Green of Forerunner Ventures, with Greylock participating
It works via MCPexperts can be queried directly in Claude and ChatGPT
Key metric #3It is free for now; a question limit will come later, and after it, payment

When a person needs advice on work or career, they go to someone who has already been through a similar situation. More and more often, a neural network answers instead of that person. The problem is that ChatGPT, Claude or Qwen produce an answer averaged across the entire internet. The startup SuperMe is trying to approach this from a different angle: to build a network of AI profiles based on the knowledge of specific living specialists. The company now has $20M to grow this network.

How the service works

A user asks a question, and several AI expert profiles join the chat. They do not simply take turns delivering lines. First, the profiles hold a short conversation among themselves and with the user's own profile, clarifying the context using data from LinkedIn. Then they answer the question itself and may disagree.

The author of the Habr piece tested this on the service's founders, Casey Winters and Ludo Antonov, or more precisely, on their AI copies. They explained the difference from ordinary LLMs this way: chatbots give an averaged answer, while SuperMe provides the experience of specific people who have solved the same problem.

The profiles draw on two kinds of knowledge sources. The first is open data: posts, interviews, podcasts. The second is materials that the experts connect themselves: Google Drive, Notion, Gmail, and meeting-notes services such as Granola. The idea is to extract experience that has never been published anywhere. The company acknowledges that such data can contain personal information and trade secrets, and assures that these do not end up in the answers. This cannot be verified from the outside, so the promise rests on the service's word.

One more detail for developers: SuperMe works via the MCP protocol. As a result, experts can be queried not only in the app but also inside Claude or ChatGPT.

Who stands behind the project and who provided the money

Winters and Antonov founded the startup in 2024. Winters worked at Grubhub and Pinterest and consulted for Airbnb and Canva. Antonov led development at Pinterest, Lyft and Whatnot. The $20M round was led by Kirsten Green, founder of Forerunner Ventures, with Greylock Partners participating.

Winters told TechCrunch that Silicon Valley has closed networks where it is known who is the best in a narrow field. He wants to make such a network more transparent and make the knowledge locked in specialists' heads accessible. Green, for her part, is betting on a network where every participant benefits from sharing their experience.

The service builds on a trend that has been developing for more than a year: from Eugenia Kuyda's chatbot for corresponding with a deceased friend (2016) to HereAfter AI and StoryFile. SuperMe's novelty is that the copy is made not for memory but for work tasks, and that it copies living people.

What remains in question

The first limitation is the composition of experts. Right now these are mainly Western specialists in product, growth and development. To broaden the range of topics, the company is hiring a person to search for new experts.

The second is motivation. An expert spends time and opens up working data, and what they get in return is not yet clear. The service is currently free. A limit on the number of questions will come later, after which payment will be required, but the material contains no scheme for paying experts.

The third is answer quality. The author compared four answers from different models to a PR task about selling tickets for a flight to observe an eclipse. The answers turned out similar in structure: in every case a major business publication was chosen for the exclusive, and a sales threshold on day seven was set. This is one example, not a benchmark, and it does not show that AI experts are consistently better than ordinary LLMs.

What this means in practice

For founders and product managers, the main lesson is this: when base models are available to everyone, the advantage shifts to data that is not on the open internet. SuperMe is building precisely such a layer, and MCP lets it be embedded into familiar chats without a separate app.

If you are building a similar product, think through two things in advance: how to keep experts' personal and commercial data out of the answers, and what to pay people for taking part. Whether the network grows to a useful size depends on it.

Story

Key Questions Answered

How the service works?

A user asks a question, and several AI expert profiles join the chat. They do not simply take turns delivering lines. First, the profiles hold a short conversation among themselves and with the user's own profile, clarifying the context using data from LinkedIn. Then they answer the question itself and may disagree. The author of the Habr piece tested this on the service's founders, Casey Winters and Ludo Antonov, or more precisely, on their AI copies.

Who stands behind the project and who provided the money?

Winters and Antonov founded the startup in 2024. Winters worked at Grubhub and Pinterest and consulted for Airbnb and Canva. Antonov led development at Pinterest, Lyft and Whatnot. The $20M round was led by Kirsten Green, founder of Forerunner Ventures, with Greylock Partners participating. Winters told TechCrunch that Silicon Valley has closed networks where it is known who is the best in a narrow field.

What remains in question?

The first limitation is the composition of experts. Right now these are mainly Western specialists in product, growth and development. To broaden the range of topics, the company is hiring a person to search for new experts. The second is motivation. An expert spends time and opens up working data, and what they get in return is not yet clear. The service is currently free.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Hands-On AI Engineering · Hamidun Academy

Build & Ship Production AI Agents Without Boilerplate Code

Start your free 72-hour trial at Hamidun Academy (academy.hamidun.com) — interactive lessons on Claude Code & Cursor, live Tech Radar, and production agent templates.

Start 72h Free Trial →Vibe-Coding Intensive
What do you think?
Loading comments…