Monthly
Membership

CHF 10.00
Monthly Cancellable.
Register

yearly
Membership

CHF 100.00
Save CHF 20.00
Yearly Cancellable.
Register
sign up for your free day pass
Oops! Something went wrong while submitting the form.

Community Event | Implementing ChatGPT/LLM-based Tools - Issues and Solutions

17
.
11
.
2023
 – 
18:00
 – 
20:00

Details

We have a great line-up of speakers who will report on implementation issues and solutions they encountered building their AI-based tools.

Stefan Djokovic
ChatGPT Plugins and "GPTs" - Build your own powerful Agents inside of ChatGPT

In this talk Stefan will
- give an update on new features in OpenAIs offering, notably "GPTs"
- give a short description of the technical details of plugins and GPTs
- introduce his plugin FlashcardGenerator
- show how to "scrape" other plugins to learn more about their strategies
- implement in <5m a new plugin from online templates
- explain the good and bad he found from his experience

Engin Arslan
Implementation of visual ChatGPT interaction Heuristi.ca - lessons learned

Using a chat-based interface to work with LLMs is the prominent way of interaction for most people. However, a written or spoken dialogue can be too linear, constraining, and full of redundancies. Heuristi.ca is a visual and node-based way of interacting with ChatGPT.

In this talk, Engin will
- explain why using a different kind of UI might be beneficial in working with LLMs and
- report the lessons he learned as a developer when building this AI-powered tool.

Lucas Fievet
Overcoming challenges of integrating LLMs into LogicFlow's MAIA

LogicFlow’s MAIA is a tool to provide users with in-app help, allowing them to find content and workflows. By using LLMs to understand website content and merging that with LogicFlow’s proprietary site mapping tools it enables a new kind of user interface which drastically accelerates learning and avoids having to move outside the software to get answers.

In this talk, Lucas will share some of the challenges faced during the integration of LLMs into MAIA and the approaches taken to solve them:
(1) how to autonomously map out an application graph and label the different states and pages;
(2) how to efficiently retrieve workflows from the graph; and
(3) how to map user inputs onto the workflow actions