appliedai_unsupervised-learners

Homepage
Example of the questionaire
Use Cases for AI projects
Implementation details for one of the use cases

Inspiration

Many companies, especially in the German Mittelstand, are lacking the resources for AI consultancies and do not have access to relevant domain experts. However large language models nowadays have the power to act with precise prompting as exactly the consultants, people need. Our vision is to democratize AI by helping people use their power.

What it does

Our Web App utilizes the broad knowledge of large language models to help guiding the decision makers of companies towards the best use cases for AI projects in their companies.

How we built it

We developed a questionnaire, which is very specific, so we can generate a profile of the user´s company and the problems he is facing, but also broad enough, so companies of different areas can use our consulting app. With these information and highly-engineered prompts we are able to get very good suggestions for use cases, which are tailored to the user and moreover hands-on implementation plans for setting them into practice.

Challenges we ran into

We learned to speak the same language as the large language models to give very specific prompts to generate the same outputs in the same structure for every repetition of the query. This was very important as it allowed us so to visualize the information generated by the AI for a better understanding of the user.

Accomplishments that we're proud of

We are proud of our working structure and time management. At the beginning we focused on brainstorming and developing the concept of the app before starting to code. Thanks to that we did not have to do big changes in our code or concept. We were able to develop a clean and professional looking app and a working backend which we tested with multiple examples.

What we learned

We learned to write very specific prompts to generate precise and helpful information. Additionally we learned to split tasks in the group to develop a working MVP in 41 hours.

What's next for AI for Everyone

Our next milestones are the implementation of further cool functionalities we have thought of: for example a visualization of the profile of the user, a nicer interface which we have already specified in our presentation and a way to playfully let the user explore different use cases of AI within a educational playground. Another feature will be the possibility to upload process data which out model analyzes to generate use cases which are even more tailored to the user´s current state.