Weekend Reads: Five Links
From automation anxiety to intentional integration, AI has a constantly shifting set of expectations around productivity, mental space, and human agency.
Designing useful experiences with AI at the core, all while trying to match expectation with capability is a delicate balance. Frameworks help.
How “wasting” time teaches the importance of mental autonomy
Our relationship with time and deciding on what to spend is always top of mind. Assuming you have the luxury of time to do some ponderin', this post from the team at Medium offers great POV on the value of what we might call "wasted" time - those moments when we're not actively producing or achieving something measurable.
The round up post from the Newsletter brings an interesting perspective that seemingly unproductive moments are actually essential for our mental autonomy. We are obsessed with productivity metrics and optimization (I’ve written a bit about this here), deliberately finding periods of unstructured time creates space for peace and creativity that simply can't exist when we're constantly "on."
Many philosophers and thinkers / gurus, whatever, have advocated for reclaiming your attention through activities that don't have an obvious ROI. Bird watching, deep listening, or just being present without an agenda - these aren't "productive" in the conventional sense, but they're incredibly valuable for our wellbeing and creative thinking.
I'm guilty of filling every moment with something "useful" - podcasts during walks, emails during breakfast, reading before bed. I'm a self-described "content junky", and at times, I find I’m losing my talent of being bored. But this article was a good reminder that mental breathing room is a necessity for maintaining our sense of self in an increasingly demanding and stressful world. It also makes me appreciate that the luxury of time, is in fact a luxury.
I'd be interested to know if you have found ways to protect some "wasted" time in your schedule? I'm trying to build more of these moments into my days, even if it means setting aside the productivity mindset that's so deeply ingrained in our professional culture.
The Medium Newsletter, prompt by Marian Bull
The Magic Minimum
This article offers a great perspective on designing for the "Magic Minimum" and how it might mark a shift in how to measure product success.
Traditional product design has adhered to the "toothbrush test". But, with agents and agentic workflows, value can be delivered even when users only engage once or twice a month.
What I find interesting is how this changes the math for product designers. Are we still constrained to measuring success based on daily-use tools? Clicks? Can we develop highly specialized and personalized solutions that work proactively in the background, that only surface only when they're done?
It's like using a professional service (like a therapist or accountant) that delivers high value despite infrequent direct interaction.
I'm particularly interested in how we'll need to rethink our success metrics (again, read here), but when value comes from quality of interventions rather than time spent in product, how do we effectively measure and communicate that value? How do we create intuitive explanations of complex background processes so that users can understand and appreciate?
I think designing for long term goal facilitation and management is the Magic here.
Stanford HAI: What workers really want from AI
TL;DR: People want AI to take out the work, not the fun.
This is an amazing research on what people really want from AI. According to the study, workers want to use AI for repetitive tasks but prefer to maintain oversight. However, this highlights a significant gap between expectations and current capabilities.
Here's what I found to be an interesting truth: workers are totally cool with AI handling the boring stuff, but they want to stay in the driver's seat. This highlights a pretty significant disconnect between what people want from AI and what it can actually do right now.
Trust is a major hurdle - folks are worried about accuracy, losing their jobs, and machines making decisions without human oversight. This creates an opportunity for us to design products that emphasize strong collaboration rather than full automation.
Looking at the research in this article, there are some clear product design implications: The future seems to favor things that will enhance uniquely human skills like prioritization, teaching, and communication. I think building products that compliment these areas while addressing the trust gap is the sweet spot.
Agents, Tools, and Simulators
More Glossary stuff:
This article has a straight-forward, conceptual framework for understanding AI that breaks down thinking into three distinct lenses. I've found it's one of those articles that helps organize the mental models for building AI systems novel and complex interactions.
Here are the three ways they view AI:
Agent Lens: this perspective sees AI as an entity with persistent goals that takes deliberate steps to achieve them. When we worry about AI having its own agenda or taking unexpected actions to fulfill its objectives, we're looking through this lens.
Tool Lens: Views AI as essentially useless without human direction. Through this perspective, AI is just an extension of human intention, providing affordances that amplify what people want to accomplish.
Simulator Lens: A little harder to internalize, but it conceptualizes AI as modeling generative processes and patterns rather than pursuing specific goals. We've been calling these "parameters" in the MCP work.
What's interesting is the overlaps between these frameworks. For instance, when Agent and Tool perspectives intersect, we can end up with collaborative relationships, prompter / worker dynamics, or alignment scenarios depending on how goals align.
These systems operate as simulators with a layer of agency on top, which is a helpful nuance, that can help explain some of their behaviors and limitations and help us create mental maps for users.
Cool tool: Plumb
If you like n8n and Zapier, then Plumb could be considered their lovechild. The interaction feels much more intuitive than the first two and gives me a much better sense of what is happening. AND, it's also an aspiring store front for using, or selling, your agent workflow complications.
✌🏼