Outmaneuvering Friction, Stages of Agents, & Gamification of Everything
In this edition we explore how AI is ending intentional friction, the five stages of agent functionality, gamification, and a round of insights for modern product leaders.
Edition #28 of Implications.
This edition explores forecasts and implications around: (1) how friction was once, but will no longer be, a strategy, (2) the stages of functionality of AI agents, (3) the growing proliferation of gamification, (4) another round of insights for the modern product leader, and (5) some surprises at the end, as always.
If you’re new, here’s the rundown on what to expect. This ~monthly analysis is written for founders + investors I work with, colleagues, and a select group of subscribers. I aim for quality, density, and provocation vs. frequency and trendiness. We don’t cover news; we explore the implications of what’s happening. My goal is to ignite discussion, socialize edges that may someday become the center, and help all of us connect dots.
If you missed the big annual analysis or more recent editions of Implications, check out recent analysis and archives here. A few recommendations based on reader engagement:
Conformative software is a new wave of apps that become more tailor-made the more you use them. These apps are designed and built with a new form of accommodative user interfaces that adjust to your skill level, your needs, and your preferences.
A few things I expect to see in 2025, from new use cases of betting markets and DIY software to the rise of Cognico companies and AI talent flocking from the companies that build AI to the industries that will be most transformed by AI.
Moving past the prompt-based era of creativity, unleashing creative risk, and creative meritocracy.
But first…Behold the world of storytelling.
A couple weeks ago, I shared a shift in my own career as I transition from leading strategy, design, and emerging products at Adobe to joining A24, an independent studio I have long admired, as a partner and leader of new efforts across innovation and technology. I will miss my teams and products at Adobe, but also couldn’t be more proud of the leadership team we’ve built and the roadmap ahead. I am also excited to spend more time as a customer using some of the products and services we’ve made over the years (and especially some of what’s coming soon!).
A24 is a special team and place, and has always had a model that not only empowers talent but ultimately works in partnership with the world’s greatest storytellers (a friend sent me a good write-up on A24’s origin story here). Suffice to say, I’m excited for the steep learning curve ahead and applying all I’ve learned serving the creative industry through Behance, Adobe, and my side projects/writing towards A24’s business and future. Bigger picture, as the space of media and technology evolve and many “skills” are offloaded to compute, I believe that taste and curation will be key differentiators. In an era of abundance in which every brand floods the zone with content, I think humans will crave better stories, with deeper meaning, craft, and new forms and mediums of shared experiences. And amidst so much volatility around the world, I have long believed that stories, in the form of films, shows, books, and music, are the most effective methods to transport payloads of hard truths.
I hope for this monthly “implications” exercise to continue, because it - and you - have helped me synthesize so many changes in technology and culture over the years — and their implications. It turns out that a wide spectrum of inputs, in the form of ideas and experiences and conversations and feedback, impacts your outputs. For this I am grateful.
Assertions & Implications
Lets jump into three assertions with implications worth noting, as well as another section of insights for the modern product leader.
Friction is no longer a strategy.
Deconstructing “agents” into the stages of functionality.
Gamification of everything.
Insights from The Modern Product Leader Playbook, Part 4
Friction is no longer a strategy.
One early use case of AI’s new “Operator” capability that I found particularly intriguing was people tasking the AI to submit claims - whether to health insurance, for refunds, entering online sweepstakes, claiming redemptions of credits, getting payment from a class action lawsuit, or other means of holding companies accountable for their obligations - in repeated and increasingly creative ways to ultimately get payment. Whereas most humans who value their time will tire with these activities, AI never tires and has no fear of failure. As a result, as these “automated browser type” capabilities become more widely available and fine-tuned for particular use cases, companies can no longer rely on the additional margin accrued from the “friction” of these activities. This is wonderful for consumers, at least initially, but may ultimately result in new policies or higher prices to account for this new world.
One company that has been ahead of the pack on this is DoNotPay, a service that bills itself as “the AI consumer companion” and has tried to productize some of these capabilities - like contesting a parking ticket or getting refunded when your internet goes out. But I suspect many other tools and AI-powered hacks will emerge wherever human friction remains a strategy of sorts for companies and governments. The use cases of AI removing friction to foster equal opportunity and righteousness are seldom discussed but very interesting. The people who suffer the most from intentional friction are often the most disadvantaged (who also have the least time to spare). Let’s hope AI is a source of productive change.
Deconstructing “agents” into the stages of functionality.
This is a topic I am brewing for a future issue of Implications, but figured I’d share some early thoughts on the agent development efforts I am seeing across startups I meet with or within big enterprise projects, and apply some organization to these functionalities. Here are the five stages of agent capabilities as I see them at the moment:
Glorified Personalized Help: The most basic (and least impressive) functionality of agents is glorified personalized help akin to an FAQ being more personalized for you and the problem you’re trying to solve.
Reactive Recommendations: You request things and get responses with some degree of work done on your behalf, whether you are summoning code, assets, or some other prepared personalized useful artifact based on your request. Most startups pitching agent-based functionality, and the agents I have played with, fall into this camp.
Proactive Recommendations: The next stage is when agents within a product experience or workflow proactively suggest something you may want to do but didn’t think of. Such agents often require multi-modal LLM models under the hood that can see what you’re doing, ascertain the context in which you are doing it and what you hope to accomplish, and can proactively and reliably make recommendations. Apart from structured systems like code generation workflows, I have not seen many great examples of this yet…but they are coming.
Proactive Action: Going further up the pyramid of possibility, the fourth stage is when agents not only proactively recommend things we might want, but actually do these things for us. This is the stage where an agent becomes a collaborator or colleague — where we start working side-by-side. We’re seeing this form of agent emerge in some coding applications, and I am excited to see this vision really come to life.
Autonomous Workflows: Finally, the upper most stage that I can imagine at the moment is fully autonomous agents that run end-to-end workflows on their own. Such workflows may involve agents making purchases on our behalf, agents working (or negotiating with) other agents on our behalf, and agents staffed with specific jobs and OKRs (objective key results) that they must meet. There are so many exciting implications of this stage, including new tools for observability and new human roles in organizations to set the goals and “rules” for agents to abide by. I explored the vast implications of this level of capability in my “Cognicos” edition.
Finally, a major caveat or rebuttal to my own outlook and analysis: The challenge to this way of thinking is whether or not we are “personifying” AI too much and trying to figure out how it fits with how humans work. When we talk about the rise of “virtual AI colleagues” are we falling into the “faster horse” trap that has always delayed the value realization of any new technology? I remember the early days of the iPhone, when Apple used skeumorphic design principles like adding fake physical thread stitching to the contacts app to make it look more leather bound and help us all adopt these new technologies through familiar patterns. But longer term, these efforts to make technology familiar ended up localizing the potential outcomes. How do we think about the use of “agents” without the biases and assumptions we have from the context of human work over the last 300,000 years? Perhaps this is a topic for another edition… ;-)
Gamification of everything: widespread and unexpected proliferation of game mechanics.
There aren’t many apps that would NOT benefit from gamification, whether it is in the form of leader boards, native currencies and rewards, daily challenges, etc. While gamification tactics has been used in popular consumer apps (Duolingo, “achievements” in diet and fitness applications, medals on our Apple Watches, etc) have been used for years, I have long wondered why such mechanisms wouldn’t work in the enterprise? Don’t we enjoy a meeting or work more when it’s fun? What if daily practices and accomplishments in the workplace had more gamified elements? Would such mechanisms increase our engagement and performance in school or at work, much like as they do in the games that we play?
As an angel investor, I am seeing more pitches in spaces beyond traditional consumer tech that incorporate game mechanics. Some education technology companies are developing entirely new game mechanics that make learning more exciting and competitive amongst peers. Good Inside, a company devoted to equipping parents with scripts, workshops, and guidance to navigate all sorts of parenthood challenges (where my wife is co-founder and COO) is also exploring the role of game mechanics to help parents navigate the many modules and milestones of parenting. Another team in the dev ops space I advise is exploring gamification of code commits and general efficiency and effectiveness for developer teams in the form of leaderboards and other tactics that will be cool to talk about once launched. (Wand when I saw recent prototypes, this felt super- powerful to me.).
This wave of adoption is not just about the readiness of technology or the tactics to drive gamification, it is also about culture. As one founder explained to me, “over 50% of kids under 12 in the U.S. play Minecraft, and 50% of kids under 16 play Roblox. These kids spend almost two hours playing each day and most of their lives are virtual and in-game, from attending concerts to hosting parties.” The logic is, if the next generation of talent is so immersed in game mechanics throughout their childhood, why not capitalize on these patterns and tendencies across other aspects of their digital lives when they become adults? Especially for those born after the emergence of the smartphone, gamification leverages a set of reflexes for engagement and self-optimization that run strong.
The challenge will be incorporating meaningful incentives while avoiding situations in which people “game” work. We already know that some people manipulate OKRs and other measures to make their work seem more impactful than it is. We’ll need to make sure that gamification is designed and implemented in such a way that gets us doing more of what we should be doing, and less of what we shouldn’t.
Insights from The Modern Product Leader Playbook, Part 4
Here is the fourth installment of a set of insights I share with or learn from modern product leaders I admire. Pretty sure I’m done writing books, so figured this was an optimal place and audience to share these insights.
As a startup, you should only do half of what you want to do (only half the options, half the tabs, half the offerings, and half the target audience) to compound your chances of true PMF (product-market fit). Most founders I advise and work with are incredibly ambitious and optimistic — two traits that help and hurt when building a product that really nails its core value. Early on in my product-building days at Behance, I often wanted to hedge our team’s likelihood of success by including more features given I lacked the certainty of which features would ultimately work. But over time I learned that, when you kill secondary features, people engage with the primary purpose of the product more. The truly experienced product builders are like Bonsai masters, they prune and cut the most beautiful branches to strengthen the trunk.
Natural > Rational. Great products capitalize on natural tendencies of humanity. Consider this thought experiment: You’re on a subway and overhear two people you don’t know talking about a downtown restaurant as the greatest Italian place they’ve ever been to. If you’re in the mood for Italian food, which opinion do you trust more, the two strangers talking on the train or the average rating of lots of strangers on Yelp? Rationally, you should trust Yelp’s average more. But most people would go with the overheard opinion on the subway. For some reason, people tend to trust people over averages and networks. Even though this isn’t completely rational, it feels natural. There are lots of these instances in which we choose what feels natural over what is strictly rational. Valuing natural tendencies over rational thinking can help you make your products more engaging and better able to hook customers faster than the time it takes for them to realize long-term value. The human brain hasn’t evolved so much in hundreds of thousands of years, so it is worth asking yourself, across each part of your user experience and especially at choice points, “what would the natural and familiar instinct of a user be?” When designing a product (and the copy, options, and UX), seek to understand your user’s natural tendencies rather than what they should rationally do.
Copy what drives familiarity, not what sets your product apart. In early Behance days, we wanted to launch a create spin on hosting group conversations for creative pros online (we called them “Circles”) and having people identify their creative fields of interest (we called them “Realms”). And in the effort to be creative and different with these features, we made them harder to understand. Ultimately, we renamed these features Groups and Fields, and then eventually integrated them into other parts of the product (basically removed them all together). The classic mistake we made was failing to use familiar patterns every chance we got. Over the years working with dozens of other products and founders, it became very clear that you should make ~90% of your product using familiar patterns, and only retrain your users ~10% of the time for the few parts of your product that are truly differentiating. Ask yourself: What is your product do that is unique? It is these few specific things - and the functions that enable them - that you should neither outsource nor emulate. As for everything else, copy what works and don’t reinvent the wheel.
Ideas, Missives & Mentions
**Finally, here’s a set of ideas and worthwhile mentions (and stuff I want to keep out of web-scraper reach) intended for those I work with (free for founders in my portfolio, and colleagues…ping me!) and a smaller group of subscribers. We’ll cover a few things that caught my eye and remained on my mind, including how algorithms are programming us, whether AI advice will become self-fulfilling and what’s happening in in-car advertising, among other areas of interest. Subscriptions go toward organizations I support including the Museum of Modern Art. Thanks again for following along, and to those who have reached out with ideas and feedback.
Keep reading with a 7-day free trial
Subscribe to Implications, by Scott Belsky to keep reading this post and get 7 days of free access to the full post archives.