Reducing the number of clicks is a misguided metric. A process with eight trivially easy clicks is better than one with two fraught, confusing decisions. Each decision burns cognitive energy and risks making the user feel stupid. The ultimate design goal should be to prevent users from having to think.

Counterintuitively, the path to full automation isn't just analyzing conversation transcripts. Cresta's CEO found that you must first observe and instrument what human agents are doing on their desktops—navigating legacy systems and UIs—to truly understand and automate the complete workflow.

A critical error in AI integration is automating existing, often clunky, processes. Instead, companies should use AI as an opportunity to fundamentally rethink and redesign workflows from the ground up to achieve the desired outcome in a more efficient and customer-centric way.

A delightful user experience should be as intuitive as answering a phone call. If users need to learn a multi-step process for a core feature, the product's design has failed to solve the problem simply.

Traditional software GUIs are valuable because they embed expert knowledge into a structured workflow, limiting user choices to what's relevant. A blank chatbot prompt forces the user to design the entire process from first principles, a significant and often overlooked barrier to adoption.

The best agentic UX isn't a generic chat overlay. Instead, identify where users struggle with complex inputs like formulas or code. Replace these friction points with a native, natural language interface that directly integrates the AI into the core product workflow, making it feel seamless and powerful.

Instead of a broad onboarding, focus the entire initial user experience on achieving one specific, "brag-worthy" value event as quickly as possible. Structure this as a sprint: define the event, remove all friction, design a "click, click, value" path, and use alerts to nudge users along to that singular 'win'.

Open-ended prompts overwhelm new users who don't know what's possible. A better approach is to productize AI into specific features. Use familiar UI like sliders and dropdowns to gather user intent, which then constructs a complex prompt behind the scenes, making powerful AI accessible without requiring prompt engineering skills.

The most effective application of AI isn't a visible chatbot feature. It's an invisible layer that intelligently removes friction from existing user workflows. Instead of creating new work for users (like prompt engineering), AI should simplify experiences, like automatically surfacing a 'pay bill' link without the user ever consciously 'using AI.'