# FY1 S2 W7 Idea Test on Community

# Study Plan

### Context

We are considering the AI Chatting channel community on Arcalive site in Korea as a target community. Community members have positive sentiment towards AI and actively discuss contents made with AI as well as tools for using AI.  

### Hypothesis

We hypothesize that members of Arcalive AI Chatting community desire a better interface and are willing to pay for such better interface. 

### Implication

If our hypothesis is true, then we can reliably start a business addressing desires of members in this community.

### Method

We will post a short post describing that we are planning to make a convenient, reliable mobile interface that grows over time on positive cash flow. We will study how they react. 

### Possible Outcomes

- Positive Comments validates our hypothesis.
- Negative Comments invalidates our hypothesis. Either community members do not desire such mobile interface, or community members have no willingness to pay for a better solution.

# Study Review

### Observations and Learnings

Community members responded with negative comments. 

Furthermore, community moderators banned us despite our acknowledgement of the community’s sentiment and announcement that we will not pursue the plan to develop and sell a mobile interface. 

- Community moderators are strongly negative against commercial activities. It’s reasonable for them to take a strong stance in order to protect their community from spam. For us, we will need to find alternative ways to form a relationship with the community.
- Community members seem to consider the interface as a low utility add, whereas they consider models as a high utility add. They are willing to pay for great models, but not for great interfaces.
- Community members are quite satisfied with features provided in the “Tavern” interface which is the dominant tool today. Benchmarking its features will probably prove sufficient.

### Action Plan

- Try providing parity with Tavern AI but in a mobile interface. From posts that they have posted unsolicited in the past, we can observe the desire for a mobile interface. But they are not interested in paying for the interface. So let them bring the API keys for OpenAI GPT models. Then we can sell computing capacity for local models later.
- If they are not open to the mobile interface with the above modification, then move on and find a different target community. Arcalive AI Chatting community remains an attractive target community for us, since 1) Text AI is easier to develop for as the first version than Image AI, and 2) Korean language speakers have more positive sentiments towards AI than English language speakers. Thus, starting with Korean Text AI community makes it easier for us to convince mainstream customers later.