Key Finding 8:
Users Want Relevance, Transparency, and Control in Chat-Based Discovery
In responses to Q9 (“Imagine you had a chat-based assistant that helps you explore, try, and compare AI tools — what would you want it to do?”), participants across both high-stress single-tool users and multi-tool users revealed a core expectation: chat-based discovery must be efficient, explainable, and user-aligned — not vague or generic.
Shared Expectations Across Both Groups:
Participants expressed several clear, consistent needs:
Help me get the best tool for my task, not a long list
Tailor recommendations based on task, prompt, or need
Summarise options clearly — via high-level comparisons or sandbox-style overviews
Explain the reasoning behind recommendations
Provide supporting evidence, such as reviews or user data
Be concise and purposeful — avoid rambling, vague answers
Enable an informative conversation, not just keyword matching
Avoid generic or unfounded suggestions — several users, especially among multi-tool respondents, explicitly stated their distrust of chatbots for AI tool discovery unless transparency and validation are present
Specific to Multi-Tool Users:
This group — experienced and efficiency-focused — raised more structured demands:
Speed: deliver results quickly, without unnecessary friction
Goal alignment: evaluate tools against specific objectives like budget, time, or scope
Extended exploration: include links or prompts to follow-up learning or documentation
Specific to High-Stress Single-Tool Users:
These participants leaned more toward guidance and support:
Prompt scaffolding: help them phrase or refine their search intent
Friendly UX: clarity and tone matter in chat
Openness to AI: this group expressed fewer reservations about chatbot-led discovery and showed more curiosity toward engaging with it
The overarching insight is that trust, relevance, and efficiency define the success of chat-based recommendations. Multi-tool users will disengage quickly if the experience lacks rigour or evidence. High-stress single-tool users will engage more readily — but require confidence-building through clarity and conversational structure.
