Introducing Kinelo and the New Hybrid Team

Managing and working with “AI” will eventually feel like managing and working with human coworkers.

As 2026 begins, there is substantial skepticism around such claims because while products bearing the “AI” label have surged in usage and revenue, the impact of AI products has remained concentrated in seeds of specific workflows and tasks, without fundamentally transforming how work gets done.

With this post I am introducing Kinelo. Kinelo is a product built on philosophies and beliefs that are slower moving, more invariant than the state of an industry as it moves through a cycle.

Kinelo aims to transform how people work with each other, and how humans and “AI coworkers”, work together in the organizations of the future.

Finding invariants is hard

But it allows us freedom to worry less about time and uncertainty.

I was at a dinner in San Francisco with CEOs of some other early stage “AI” companies, sponsored by a prominent VC firm. The host sparked a debate with the prompting question: When will AGI arrive?

I am generally an optimistic skeptic. I have been in the machine learning and AI industry for over a decade–founding an ML/NLP startup in 2015, exiting it to a big tech company a few years later, and managing AI and ML teams–and I’m somewhat old school. I even still latch onto the term “machine learning” or “ML” instead of “AI”, though that is a losing battle.

Around the table I heard speculation on the timing of AGI, but I retorted with a different question: do we even have a definition of AGI? If not, how can we be talking about dates?

To me AGI means something close to a single system that can learn to drive in a snowstorm, recognize sarcasm, write an essay about Aristotle’s Poetics, and speculate on whether a particular song will go viral on TikTok or not. Truly general. And we’re not there yet.

But what if we ignore the timing and think about the end state? What would we as entrepreneurs choose to build towards?

Discovering Kinelo

With Kinelo we believe that AI capabilities will continue to improve. At some point—whether 2027 or 2035—AI systems will be capable enough to work alongside humans as genuine collaborators. And that in that state, certain requirements will need to be met to make them effective.

Kinelo is a product whose end destination is to provide the foundational pieces needed to make the human/AI integration a reality at work.

For the past few years, my team and I have been quietly working on Avy, a software product designed to help humans deal with the chaos of information scattered around them at work. Our mission has been to make computers feel truly intelligent, proactively bringing information from across systems, communications, coworkers, and entire organizations. We built technology that ran at the OS level and observed the human user's behavior, predicting their needs in real time using on-device AI and machine learning.

The long-term vision with Avy was that if we started helping individuals with their personal productivity at work, we could then start to observe interactions between people, model processes and workflows, and eventually orchestrate and coordinate those workflows across people and teams.

"Assembly line in the knowledge work factory" has been our north star since the beginning.

At a recent offsite in Utah our small team was sitting around a conference table. We were discussing our own internal use of AI products. While we do use them for many tasks–from programming to market analysis to reviewing legal contracts–AI has not fundamentally changed the way our team works together. Even though we are ostensibly at the forefront of the industry.

The team made a few observations:

1. Current AI tools produce inconsistent output. Raw hallucinations yes, but also just not meeting expectations in how they approach problems. They feel like excited-to-please interns: smart, but naive and without the confidence to ask questions to clarify intent and deliver desired results.

2. Current AI tools are out-of-sight, out-of-mind. They exist at specific points in workflows for specific tasks. They are reactive and invoked by people. And they are isolated from each other and from the human workflow in a way that makes them oblivious to the information and context required to perform well.

We see it all the time. If you're on an engineering team, you’re probably at the forefront of AI adoption compared to the rest of the world, but it’s still likely you've felt this. You ask an AI coding assistant for help. You give it a ticket description, maybe some context about your codebase. But it doesn't know that the API you're building depends on a schema change another engineer mentioned in standup yesterday. It doesn't know the PM clarified scope in a Slack thread this morning. It doesn't know the deadline moved because of a customer escalation. So it produces code that’s just not right. It feels like slop.

Fundamentally, the model is smart. It's just not in the loop, and it acts dumb as a result.

Then we noticed something else. As we humans sat around the table looking at each other in 3D, we noticed an absence: this hypothetical “AI” was not in the room with us.

And that led to the insight: How would AI have any hope of realizing its promised potential–that of being a truly collaborative coworker on par with us–if it wasn't even in the room with us?

The very things humans need to succeed at work—clear goals, background knowledge, ways to ask for help, the ability to participate in conversations—are the same things AI would need to participate meaningfully. Without those affordances, AI remains a disconnected tool you invoke. With them, it becomes a teammate.

So, what is Kinelo?

Kinelo is building the foundational technology to bring AI a "seat at the table." Not AI as a tool you invoke, but AI as a participant in teams at parity with humans.

This isn’t just a technology play. We’re looking at how humans work together for inspiration, because we believe that in the limit, managing, coordinating, and facilitating a team of human+AI coworkers will look a fair amount like managing a team of only humans today. And therefore we need to think about psychology, communications, and focus. As AI increases the pace of productivity, these will become increasingly important considerations for the humans working alongside them.

The first version of Kinelo helps human software engineering teams ship more effectively. We're starting there because that's where we have the deepest experience and where the pain is most acute.

Underneath the surface, Kinelo is a persistent intelligence layer that lives within your company, continuously building understanding of goals, commitments, and progress across all the places work happens. It forms a shared context that both humans and AI can draw from.

Our goal is that Kinelo will eventually manage entire companies. In the near term, those companies will consist of humans with AI tooling assisting them. Over time, they will increasingly consist of hybrid teams—humans and AI coworkers working together.

We've been working on Avy and Kinelo for some time, and we're excited to share more soon.

When successful, Kinelo will help AI products feel smarter, more effective, more capable than they are today. Yet simultaneously they will feel more integrated, more transparently part of workflows, more natural. And humans will experience a less chaotic, calmer, more productive work environment.