I think this really works. While at Datadog, at one point I would write deep dives on monitoring applications like HAProxy, OpenStack and others. I would start with what to look for, how to monitor those things with off-the-shelf tooling, and finally how to do it with Datadog. The response from the community was overall very positive, regardless of whether or not they ended up converting to users (though many did).
This is great. I once worked at a company called Paribus that would argue with customer service to get a refund on your behalf for retailers and credit cards with price drop protection. Wouldn't be surprised to see in the future a more maliciously-tuned bot for things like fraudulently claiming non-delivery of goods.
At GrubHub, I am often in need of a partial refund. Sometimes it is granted immediately, and sometimes it is flagged for a chat with CSR. It seems non-deterministic. Sometimes the CSR seems to deliberately misunderstand my request, warranting two or three chat sessions until it's all cleared up. A robot advocate could certainly conserve my blood pressure.
Until I stopped using DoorDash, they used AI agents. Typing "agent" transfers the chat to a real human who would often begin rapid negotiation. Complaining a little more results in a better inducement. That part could be gamified with an chatbot on the customer's side.
I just had a cyberpunk vision of illegal optical film manufacturing labs that exist so people can take pictures that can't be AI-censored by camera-chip firmware, web browser code, hard drive controllers, or USB chipsets.
A couple questions, if you don't mind. How did you go about finding clients? What is the nature of the work agreement—project-based, hourly, or something else?
In the author's defense, a primary usecase is listed in the first graphic under the top paragraph (and the sentence just below the image highlights it).
A bit off topic but I enjoyed reading the bar exam questions you linked and also found that Virginia publishes their answers. For those interested, you can view them by year (the answers also contain the question text): https://barexam.virginia.gov/bar/barsampleanswer.html