What it is
An AI assistant is an interface that enables users to control complex software with natural language commands instead of memorizing menu trees, or keyboard inputs, or arbitrary machine language. This means that the most advanced functionality of your software becomes immediately accessible to even new users.
AI assistants work by using a large language model (LLM), or possibly a small language model (SLM) which is more cost effective to translate the endless variety of ways a user might phrase a command into something your software can understand. This means the LLM must be trained on both the available functions in your software and also the ways your users think and communicate.
Why you want it
Increase customer satisfaction
The more value your customers get from your product, the more the loyal they will be. They are also more likely to convert to higher tiers of service. AI assistance maximizes the value that users can derive by enabling them to use your product to its full potential.
Steal customers from your competitors
Steep learning curves are one of the top barriers to adoption. They are what keep users attached to a specific technology solution, as they dread the prospect of having to un-learn everything and start over. If you don't bring in users with an AI-assisted user experience, someone else will. This is a once-in-a-lifetime opportunity.
Prevent the competition from taking your customers
The most powerful use for AI is an assistant. What this means is an interface that enables users to control complex software with natural language commands instead of memorizing menu trees, or keyboard inputs, or arbitrary machine language. This means that the most advanced functionality of your software becomes immediately accessible to even new users.
Create brand new customers
But an AI assistant won't just win you the users from other competing products. They create an entirely new market of users who would otherwise not have used anything at all.
What I do
Planning and implementing an AI agent is a complex process that involves multiple steps.
Model the software: This means map out every function and how it relates to the other functions, as well as the product's information architecture. My preferred approach is my proprietary tool, XenBuild.
User research: For each function or combination of functions, there are a number of ways that your users may type them into a chat UI. I conduct user research to learn about how your users think and communicate.
Select of a language model: In most cases, a large language may not be necessary, and a small language model (SLM) may suffice. The best option is open source to give you full control.
Oversee technical implementation: The AI assistant needs to be built and connected to the code of your product. I can help oversee the process and ensure it remains to spec.
Training the AI: The AI must be trained on the product model and the user command language created during user research.
How to get started
If you are considering adding AI to your product, book a call with me, and we can discuss your unique needs. There are a lot of ways to do AI wrong, and they can cost you a lot of money without any return. Don't end up like the many companies currently throwing good money at bad projects. Let me help you stay on the cutting edge without getting shredded.