We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
SAINTCON 2023 - Jonathan Dutson - Build Your Own AI Assistant
Build your own AI assistant with expert tips from Jonathan Dutson. Learn how to select the right LLM, keep input specific, and overcome limitations for accurate outputs, and discover how to customize, integrate, and test for optimal results.
- Choose the right LLM: Select an LLM that fits your use case, considering factors like accuracy, customization, and cost.
- Keep it specific: Provide specific input to reduce ambiguity and improve accuracy.
- Give the model an out: Allow the model to admit when it doesn’t know the answer, reducing hallucinations.
- Customize with your data: Integrate your custom data into the LLM to create a tailored bot.
- Be aware of hallucinations: Recognize that LLMs can make mistakes and “hallucinate” due to lack of understanding.
- Make it interactive: Create a system prompt that takes user input and provides accurate outputs.
- Use an SDK: Utilize orchestration SDKs like Azure OpenAI or Semantic Kernel to integrate LLMs with other tools.
- Don’t rely solely on LLMs: Combine LLMs with human input and expertise to ensure accurate outputs.
- Network and learn: Join online communities, share knowledge, and stay up-to-date on the latest LLM developments.
- Test and trial: Experiment with different LLMs, frameworks, and applications to find the best fit for your needs.