Last year, there was interest in agent-based AI approaches such as AutoGPT. This year, we're seeing startups implementing some of the “let AI do the work” use cases. Her Mindy, who advertises herself as an email-based executive assistant, is one example.
Using Mindy is very easy. You can email your questions to the assistant at m@mindy.com with questions like: “I want to know about all the events happening in my area.” [you have to specify the area] There is a link to book your tickets this weekend. ” or “Can you tabulate Apple’s sales for the past eight quarters?”
Mindy searches for the information on the web and responds via email after some time.
Image credits: mindy
Currently, Mindy is free to use, as the company is considering a suitable price for the service.
Why did the founders start Mindy?
Mindy was founded last year by Yu Pan, Cuong Do, and Benoit Berthoox. Mr. Pan was a founding engineer at both PayPal and YouTube. Do he join PayPal in 2002, but was a founding engineer at YouTube? Mr. Bertaud has worked for consulting firms such as Capgemini and Accenture.
The company raised $6 million in a seed round led by Sequoia Capital (with Roelof Botha as lead partner) and Founders Fund.
“The idea germinated in 2022, when I was done with the crypto side and Benoit was finishing up his own efforts. AI came along with the growing popularity of models like GPT-3.5. . Doing anything with these models required a lot of quick engineering. So we realized that since these models are parsing languages, we did a lot of work before using the tools. We thought there was no need to think about it,” Pang said in a phone call with TechCrunch about starting the company.
Berthoud said Mindy did market research and found that people use apps like ChatGPT but forget about it. Remarkably, despite this, ChatGPT still has over 100 million active users.
The company began testing Mindy in “semi-stealth” mode in November. However, he declined to say how many users interact with Mindy.
Choose email as a differentiator
Pan said the company considered building on email because it has a high “signal-to-noise” ratio of information and also focuses on specific topics.
“Email is focused on text, and our LLM is focused on text. So we thought it was a great marriage. Our view of these media is essentially: “It's like trying to marry the oldest technology in the book with the newest technology,” he said.
The company claims that many users, from financial analysts to students to defense contractors, use the bots to conduct research. Also, the AI assistant is not expected to respond instantly since he uses email rather than a web interface.
The company believes that because people are familiar with email and the conversational tone of this medium, there is no need to change workflows to adapt the product to their lives. Additionally, you can use it with a group of other people when coming up with plans and ideas.
The company says the asynchronous nature of email allows for greater agent autonomy and automation for more relevant results.
Although the company's primary focus is on email, it has also tested prototypes for a variety of use cases with other distribution challenges.
road ahead
The company currently has a team of 10 people and is looking to hire people to accelerate its growth through engineering courses. We're also looking at different monetization models, including subscriptions, query limits, and taking a cut from completed purchases. However, the specific approach has not yet been decided.
Agent-based approach
As my colleague Devin Coldewey wrote in a year-end article, we're looking at AI-focused startups that take an agent-based approach. Most recently, The Browser Company, the company behind Arc Browser, announced that it is developing an AI agent that will browse the web on your behalf.
An industry executive, who requested anonymity, said startups like Mindy looking to build agents need to build enough differentiation to avoid being overshadowed by the likes of ChatGPT. These tools offer very high value to customers, they said, and they need to find ways to acquire users cheaply.
Hemant Mohapatra, partner at Lightspeed, believes that AI assistant-based startups need to focus on user delight and reliability.
“Good consumer apps often experience rapid growth in their early stages due to high consumer satisfaction and word-of-mouth referrals,” he said. Mohapatra pointed out that the Assistant will need to focus more on compliance and security, as it will need more user data to handle end-to-end requests.