UX design
fromMedium
1 day agoThe deceptive nature of today's AI conversation design and how to fix it
Conversation design for non-human participants may be outdated and inefficient, raising questions about its effectiveness in user interactions.
Chronicle captures screenshots of your display, processes them into text summaries, and stores those summaries as local memory files, allowing the AI assistant to understand your recent activity without explanation.
For decades in SAAS, products reduced ambiguity. Users supplied constrained inputs, and the system handled the output. It's never been Minority Report cinematic, but it was predictable. By providing predictable environments for manipulating data, users learned by moving things, adjusting variables - and the outcome emerged through interaction.
LLMs have made AI assistants a standard feature across SaaS. AI assistants allow users to instantly retrieve information and interact with a system through text-based prompts. Mathias Biilmann, in his article " Introducing AX: Why Agent Experience Matters," discusses two distinct approaches to building AI assistants. The Closed Approach involves a conversational assistant embedded directly within a single SaaS product. Examples include Zoom's AI Companion, Salesforce CRM's Einstein, and Microsoft's Copilot. The Open Approach involves external conversational assistants, such as Claude, ChatGPT, and Gemini,
This is simply a mismatch between how UI is generated in AI tools and how products are actually built. So instead of avoiding constraints such as branding and data, they need to be integrated into the generation process. This way, the UI becomes more valuable and can be used to continue to iterate past a demo or proof of concept (POC) phase.