Week #5 #

Feedbacks #

Feedback collection plan

This week our team is focused on collecting and analyzing user feedback. To collect user feedback, our team decided to organize feedback sessions with other students from the University. During these sessions, we will give the users an opportunity to test our project. These sessions include the usability tests of both the website and the telegram bot.

After each session, we asked the students to answer several questions regarding their user experience to obtain insights, new ideas and overall user feedback.

The questionnaire for the users will include the following questions:

How relevant do you find our project in terms of its potential uses? Can you see yourself or others benefiting from this AI model?
How would you rate the usability of our website? Was it easy to navigate and use?
How easy was it for you to receive and use the token to access the AI model via the Telegram bot? Were there any challenges you faced?
How accurately does the AI model imitate the communication style of the person from the uploaded chat history? Do you find the imitation convincing and precise?

We believe that these questionnaire will provide us with valuable insights that will help us improve our project.

Conducted user surveys or feedback sessions

Our team has conducted two feedback sessions with the users. Here are the results of the questionnaire for both respondents.

First user: #

Q: How relevant do you find our project in terms of its potential uses? Can you see yourself or others benefiting from this AI model?
A: I think the project has potential, but I’m unsure how often I would use it personally. It might be more appealing for people looking to create chatbots for customer service or personal use.

Q: How would you rate the usability of our website? Was it easy to navigate and use?
A: The website is user-friendly and easy to use. However, not everything is implemented yet. For example, it is not possible to upload more messages to the mode yet.

Q: How easy was it for you to receive and use the token to access the AI model via the Telegram bot? Were there any challenges you faced?
A: The process of receiving the token was smooth and everything worked well. However, the bot lacks the ability to stop the conversation and start a new one with another token.

Q: How accurately does the AI model imitate the communication style of the person from the uploaded chat history? Do you find the imitation convincing and precise?
A: The AI model is impressive in some respects, but it sometimes it misses the nuances of the person’s communication style. It’s good, but there’s room for improvement in capturing unique expressions.

Second user: #

Q: How relevant do you find our project in terms of its potential uses? Can you see yourself or others benefiting from this AI model?
A: I think the project does not add something new to the market. However, it can be useful and interesting fot Russian market, since it function in Russian language.

Q: How would you rate the usability of our website? Was it easy to navigate and use?
A: The website is fine and fulfills the basic needs for a user.

Q: How easy was it for you to receive and use the token to access the AI model via the Telegram bot? Were there any challenges you faced?
A: Generally, the process of receiving a token is good. However, I think the users would require more information on receiving the token. For example, it would be a nice addition if the user could read an instruction on how to export the chat history from Telegram.

Q: How accurately does the AI model imitate the communication style of the person from the uploaded chat history? Do you find the imitation convincing and precise?
A: The AI model did a decent job imitating the communication style, but there were noticeable inconsistencies. Sometimes, the responses felt generic and not specific to the person whose chat history was uploaded.

Analyzing feedback, identifying and prioritizing issues

The feedback provided a few insights for our team, and we have made several important conclusion regarding the future development of our project. Here is the list of prioritized tasks that we have to complete, based on the received feedback:

Website usability. Our team has to work on the website usability and functions. More exactly, we may add instructions for the user on how to export his telegram chat history to upload it to the website.
Token accessibility and Telegram bot usability. Based on the user feedback, our team has decided to allow users to cancel current conversations with the model. Such that the user can access another model via a different token.
AI model accuracy. Our Machine Learning developers have to tackle the model’s inconsistencies and improve the model’s overall performance.

Roadmap: #

Our team has created a roadmap of our project. This roadmap proposes the future prospects of our application, including its future features and functionalities:

Creating the MVP. Finalize main functionality. <- We are here.
Finish the website. Complete the website secondary functions. For example, premium accounts.
Customize the telegram bot. Make the telegram bot more user-friendly, add more instructions for the user about model’s usage.
Dockerization. Containerize the application using Docker.
Production. Put the application to production. Post the project to a service like AWS.

Weekly Progress Report: #

Frontend development:
This week our frontend developers have finished developing the design of the website. Now the website is ready to be finalized.

Backend development:
Frontend developers have connected the machine learning unit to the database and the website.

Machine learning:
Our machine learning developers are actively developing the machine learning unit that is almost ready to be used on any telegram chat histories provided by the user.

Challenges & Solutions #

The main challenge of this week was to develop a questionnaire tor receive valuable user feedback. Moreover, our team members had to find people to test our application.

We have organized several discussion sessions as well as feedback sessions with the users. During our discussion session we have collective created the questionnaire and found people to test our application.

Conclusions & Next Steps #

This week we have achieved valuable results. We are actively working on our project to present a completely functional and presentable MVP.

The next steps:

Finalization of the website to be fully functional and usable.
Finalization of the telegram bot.
Finalization of the Machine Learning Unit development.
Start implementing Docker for future production.