Congrats on launching! Don't stress about the low levels of response, I suspect that as you suggested, this will be more of a tool to help companies do tech support than a tool for consumers to get help.

Keep up the pace and iteration!

Hi Hacker News! I'm Griffin, co-founder of Circle. Here at Circle, we want to build tech support chatbots for large companies. We're building bots that are much more knowledgeable compared to current bots and give users a much better experience. We've made a chatbot that analyzes Apple's community discussion forums to find accurate solutions, which you can try out on our website.

I've often been frustrated by current support chatbots, and I've talked with many others who felt the same way. Most of the time they couldn't find answers to my questions, and I instead had to talk to live agents who could understand me and help troubleshoot.

I believe that the fundamental way these chatbots work prevents them from solving many problems and giving users a good experience. These chatbots use solutions that humans manually add to a database. What often happens is that the bots have answers for simple things like how to reset a forgotten password or cancel a subscription, but for many tech issues they won't have a pre-written answer at hand. This limited knowledge makes it hard for the bot to help. They can't problem-solve or try to further understand the problem like a support agent can. Talking with them is like trying to have a conversation with someone who can only respond by reading off a few pre-written sentences on a list — pretty frustrating.

We’ve noticed two major approaches that companies in this space take. One type makes bot-builders, like Ada (https://www.ada.cx), where employees at the company they're selling to manually enter every question and the corresponding answer. The other type, for example, Solvvy (https://solvvy.com), builds bots that hook into knowledge bases like Zendesk and search for a corresponding article. Both approaches lead to bots that read from a small list of manually-created answers and chatbots with restricted capability.

Today we're launching V1 of our product. We've built a tech support chatbot for Apple-related tech issues that analyzes Apple's community discussion forums to find the best solutions. The fundamental difference is that our chatbot has access to a lot more data: Apple's forums have millions of words of knowledge. It gives solutions that are very likely to work since people in community forums have firsthand experience with the problem.

We're also developing a future version that makes our chatbot even more helpful and easy to use.

We're building a chatbot that will bring in the helpful aspects of a human agent, a chatbot that can problem-solve and adapt to the situation. It will search the internet throughout the conversation to give clarification, answer further questions, and guide the user step-by-step through the solution. With that, it'll become an incredibly effective support agent, capable of solving tech problems to a high degree of accuracy. Without getting knowledge from the internet, that wouldn't be possible.

In the future, we want to monetize by partnering with companies, where people who contact tech support would interact with our bot. We're hoping to automate a significant amount of inquiries for companies, which can bring significant cost savings, while still giving users a good experience.

I'm so excited to share what we've been working on with you all. What are your thoughts? I'm all ears!