Why Waldo
Tell Waldo what you want to get done.
Waldo understands your context, coordinates the right AI agents and tools, brings you in when your judgment matters, and carries the work until the outcome is actually resolved.
As AI becomes more capable, people should not have to become managers of AI.
Waldo turns intent into outcomes.
01
AI is splitting into two worlds.
Personal AI is getting better at understanding your messages, calendar, preferences, memory, and daily life.
At the same time, specialist agents are becoming much better at doing specific kinds of work like coding, research, browsing, communication, analysis, and operations.
We believe these two worlds will come together.
People will not want one AI that knows their life and a separate collection of systems for managing every agent they use at work.
They will want one agent that understands them and coordinates the rest around what they actually want to accomplish.
That is Waldo.
02
Today's AI agents can already perform increasingly complex tasks.
But completing a task is not the same as delivering the outcome you wanted.
An agent can finish writing the code while the release is still blocked.
A research agent can return a report while the decision is still unresolved.
An email can be drafted while the commitment behind it is still open.
The person still has to:
Agents execute tasks.The person still carries the outcome.
Waldo is built to change that.
03
You start with what you want to accomplish.
Ship this release by Friday.
Research these options and help me make the decision.
Plan my week around everything I need to finish.
Waldo keeps the larger outcome in view.
It can break the work down, coordinate the right specialist agents and tools, follow what happens, and bring you back when a meaningful decision requires you.
But Waldo does not treat an agent saying "done" as the end.
It keeps track of:
An outcome may be completed, reopened, deferred, transferred, or consciously dropped.
Either way, it does not silently disappear when an agent session ends.
Waldo stays with the outcome from intent to resolution.
04
A coding agent can write code.
A research agent can gather information.
A browser agent can complete an action.
A communication agent can draft or send a message.
Those are pieces of work.
Waldo keeps track of how those pieces connect to what you actually wanted to accomplish.
Specialist agents can change underneath it.
Your context, your outcome, and your relationship with Waldo remain.
05
The person using AI at work is the same person living the rest of their life.
Your meetings affect your focus.
Your health affects your capacity.
Your commitments affect your priorities.
Your work affects your day.
Most software treats these as separate worlds.
Waldo does not.
We believe a truly personal agent needs a continuing understanding of the person across these boundaries.
Being personal is not only about knowing who you are. It is also about understanding the state you are in.
Your priorities change. Your capacity changes. New commitments appear. Work becomes blocked. Plans move. Some days you have room to take on more; on others, protecting your attention matters more.
A personal agent should understand that changing state and adapt how it plans, prioritizes, delegates, interrupts, or leaves you alone.
Waldo is designed as one continuing relationship across work and life, instead of a different assistant for every part of you.
06
A truly personal agent should understand more than your inbox, calendar, and stored preferences.
It should have some understanding of the human behind them and how that person's state is changing over time.
With your permission, Waldo can use signals such as sleep, recovery, stress, activity, and changing capacity as part of that understanding.
Not as another health dashboard.
Not to make medical decisions for you.
But because the right plan can change depending on the person behind it.
A low-recovery day may mean protecting focus, moving lower-priority work, delegating more, or reducing unnecessary interruptions.
A high-capacity day may look different.
Health is one part of understanding your state.
It gives Waldo context that most software does not have: not just what needs to happen, but something about the person who has to live through it.
Health becomes context for agency, not another score for you to manage.
07
There is one Waldo, expressed through different surfaces.
Work
Kennel is Waldo's Mac work surface.
It is where you can see the outcomes Waldo is carrying, the specialist agents working underneath them, what has changed, what needs your judgment, and what remains unfinished.
What needs me, and what can Waldo keep handling?
We are starting here because founders, engineers, and investors working across multiple AI agents already feel this coordination problem.
Personal
Mobile becomes the personal side of Waldo.
It is where your day, communication, health context, priorities, decisions, and ongoing commitments come together.
It gives Waldo a better understanding of the person behind the work and gives you a simple place to make decisions, see what matters, and stay connected to the outcomes Waldo is carrying.
Everywhere else
Waldo should also be reachable where you already work and communicate.
Messaging and browser interfaces become lightweight ways to talk to the same Waldo rather than creating another separate assistant.
Underneath every surface is the same shared understanding of you, your permissions, your ongoing outcomes, and what remains.
08
You should not have to recreate yourself inside every new AI product.
Waldo should remain the persistent relationship while models, tools, and specialist agents change underneath it.
A coding agent may own a coding task. A research agent may own a research task. A browser agent may own an action.
Waldo keeps track of why that work matters to you, how it connects to the larger outcome, and what needs to happen after those agents stop.
More agent runs are not the goal.
More tasks completed are not necessarily the goal either.
The goal is for something meaningful in your world to become true.
Waldo begins with what you want to accomplish and works backward from there.
A successful tool call, finished agent run, commit, document, or message may be evidence of progress.
None of them automatically mean your outcome is complete.
Waldo keeps the difference visible.
Memory is part of being personal, but it is not enough.
A truly personal agent should understand both who you are and the state you are in.
Your priorities, commitments, relationships, preferences, corrections, capacity, unfinished work, and changing circumstances all shape what the right action is.
The same request may deserve a different response on two different days.
Waldo should understand enough of that changing context to plan, prioritize, delegate, interrupt, or leave you alone appropriately.
The purpose of Waldo is not to maximize the number of agents running around you.
It is to help you carry less.
Less re-briefing.
Less supervision.
Less context switching.
Less remembering everything yourself.
More time and attention for the decisions that actually need you.
09
We are starting with people already living in the agentic future: founders, engineers, and investors whose work already spans multiple AI agents, tools, and information streams.
Kennel gives Waldo a focused starting point where the problem is already visible.
Can one personal agent carry an outcome across multiple agents without making you the coordinator?
From there, Waldo grows into the broader personal agent: one agent that understands the person, coordinates the intelligence around them, and stays with what matters across work and life.
The long-term idea
Models will keep getting better.
Specialist agents will keep multiplying.
Tools will change.
Interfaces will change.
The part that should remain constant is the agent that works for you.
An agent that understands you, coordinates the intelligence around you, and carries your outcomes to completion.
That is Waldo.
One Waldo across work and life,
coordinating the intelligence around you and carrying what matters forward.
