Learning to Fly the AI Rocket Ship, Together
Being on parental leave these past several weeks has given me the opportunity to reflect on the nature of knowledge work in the age of AI - not only in our professional lives but also in how it integrates into our personal lives, and the other way around.
We’re all working to keep pace with the onslaught of new AI models, concepts, techniques, and tools. Podcasts, conference talks, open source repos, books, articles - all in a space moving so fast that we’re learning and sharing in public simultaneously. We’re all trying to figure out what “good” looks like.
At times, the velocity feels overwhelming. At other times, AI still feels like just a helpful tool - useful, but not yet delivering on what we’ve been promised, both professionally and personally.
I think both feelings are valid. And I think 2026 is when that changes.
My Personal AI Context
My journey in tech started with a Computer Engineering degree. After eight years in Management and Technology Consulting in the FinTech space, I pivoted to startups and Product Management.
What’s interesting is how all of these experiences are remixing now with AI. Business Operations and Intelligence, Systems Engineering and Architecture, User Experience, Product Management, Operating System Design, DevOps - these disciplines are being reshaped and rethought together to create the businesses and products of the very near future.
I don’t come at this as a pure AI researcher or ML engineer. I come at it as someone who has spent years thinking about how work gets done, how systems scale, and how to design for humans operating within complex systems.
A 2026 Thesis
I believe the promise of GenAI for business outcomes will be realized in 2026.
- 2023: We experienced the magic of ChatGPT
- 2024: We saw refined and new models emerge
- 2025: We saw accelerated business adoption - but the results aren’t fully there yet
We’ve found the boundaries of what raw LLM technology can provide. In 2026, I expect breakthroughs in engineered and applied AI-centric applications and platforms that address the friction points we’re all experiencing today.
Knowledge Work Will Get Restructured
Knowledge work is about to look very different. It’s going to be restructured by tools that allow human operators to manage and offload “mental load.”
Completely new AI-centric skills and habits will be required on top of the underlying domain knowledge we have today. We’ll still need that knowledge, but we won’t deploy it in the same way we have in the past.
And because of this, the real competitive pressure isn’t AI itself. It’s the person next to you who figures out how to multiply their output with these tools.
The 60-70% Promise Doesn’t Scale Linearly
There’s an explicit promise being bandied about: AI can do 60-70% of the work for you. So you should be able to do 3x more and achieve 3x business outcomes, right?
Yes and no.
Yes, AI can handle roughly 60% of the obvious, mechanical work. But that doesn’t scale to business outcomes linearly. The remaining 40% - judgment, context, synthesis, stakeholder navigation - is where the leverage actually lives.
No, you shouldn’t settle for 3x. You should be able to achieve much more. But you need to rethink your underlying assumptions about how work gets done with AI to realize that potential.
We need to keep going back to first principles.
What I’m Exploring
So that’s what I’ve been doing - going back to first principles. Specifically, I’ve been exploring how to structure and manage context for AI systems. Not just prompts, but the entire information architecture that feeds into these models. I believe that Context Maps - structured ways for knowledge workers to inject, manage, and refine conditional context - will matter a lot for how knowledge workers actually get things done. There are many of us here building, sharing, and learning in this space.
This isn’t about building another AI tool. It’s about understanding how humans and AI systems will work together, and what new disciplines emerge from that collaboration. I’m building this for myself first, using existing tools and concepts, and I plan to share what I learn along the way.
I’m fortunate to have the time to focus on this while I’m on leave. As I wrap up and return to my full-time Head of Product role, I’ll share more progress and learnings.
Looking forward to continuing to build and share with you all in 2026.
P.S. - Thought Leaders Worth Following
Shoutout to folks who have been valuable thought leaders in the AI and Product Management space. Thank you for your content and your authentic, personal style that adds taste and personality in a sea of generic playbooks and sales pitches. In a sea of grayscale content, you add much-needed hints of color: