Memory
It remembers, so you don't have to.
Watch the agent recall context from a conversation weeks ago — and use it to act, not just summarize.
Camelot
Privacy-First · Agentic Operating SystemThe Personal Assistant that learns to think the way you want.
CamelotOS is a privacy-first personal assistant — a native macOS app with an iOS companion — built so it grows smarter over time, keeps your data yours, and works seamlessly across the tools you already use.
The Premise
The operating system for your personal assistant.
Most AI assistants live in the cloud, forget who you are between sessions, and ask you to trust them with everything. CamelotOS takes the opposite path: an environment built around your privacy, your data, and an agent that genuinely improves the more you use it — no matter which AI models power it underneath.
What Sets Us Apart
Different by design — not by feature list.
These are the principles competitors can’t easily copy, because they’re built into the foundation rather than added on top.
Memory That Compounds
Designed to remember the way a person remembers — not like a database. It gets sharper the more you use it, and that depth carries forward no matter which model is underneath.
Privacy by Architecture
Privacy isn't a setting we offer — it's the foundation we build on. Your information stays yours by default, not by exception.
Local First, Best Cloud When It Counts
It can seem like a personalized local model means sacrificing performance. It's the opposite: our hybrid architecture runs everyday work on-device, then calls on the best cloud models the moment a task outgrows what local can handle quickly and accurately — so quality is never the price of privacy or cost.
Connected to Your World
Ecosystem-wide connectors integrate the third-party services you already rely on — with the catalogue of integrations expanding continually.
Growing dailyBiometric Authentication
The moments that matter are gated behind biometric approval — keeping a human in the loop precisely when trust is on the line.
The sum is the moat.
Any one of these is hard. Together — model-agnostic compounding memory, privacy-first design, hybrid compute, and an open ecosystem — they form something no cloud-only assistant can retrofit.
How it feels · scroll to assemble
Always Learning
Memory that gets better the more you use it.
It picks up your voice, your judgment, your priorities — and gets sharper the longer it serves you. Less machine, more memory.
your voiceyour judgmentyour prioritiesyour way
Always Capable
Across every tool you already use.
Connectors let it work inside the apps you already live in — picking the right one for the moment, without you having to ask.
MailCalendarFilesResearch
Always Yours
Yours by default. Not by exception.
Privacy isn’t a setting. It’s the foundation. On-device first, encrypted in transit and at rest, and never sold to train someone else’s model.
on-deviceencryptedportableyours
See It
Moments from the product.
Short clips of the agent doing real work — not slideware. We’ll add captures as features land.
Local-first hybrid
On-device by default. Cloud when it counts.
See routine work happen instantly on-device, with the cloud only entering for the moments that truly need it.
Connectors
Plugged into the tools you already use.
A walkthrough of authorizing a connector and watching the agent pick up the task end-to-end.
Biometric trust
Human-in-the-loop for the moments that matter.
Face/Touch ID gates the actions that move money, send messages, or change anything you care about.
The Economics
Every task you hand to the cloud keeps a meter running.
As individuals and teams delegate more of their daily work to AI, cloud API costs don’t just add up — they compound. More tasks, more steps, more usage, every single day. And the more capable the model, the more every call costs. For one person it’s a creeping bill; across an enterprise, it’s multiplied by every seat.
designed reduction in recurring inference cost over time — and the savings deepen the more the system learns your work.
Zero‑cost compute first
Orchestration prioritizes free, on‑device local compute whenever it’s capable of the task — escalating to paid cloud models only when the work genuinely demands it.
Savings that compound
Adaptive skill learning means CamelotOS handles more on‑device the longer you use it. Your cost curve bends downward over time instead of climbing.
Model‑agnostic by design
It makes the most of whatever compute is available — local or cloud, today’s models or tomorrow’s — always running the most cost‑effective option for the job, with no lock‑in.
On The Horizon
Giving you back your time — for the human connections that matter.
Automation that only makes you faster misses the point. CamelotOS is built so the hours your agent reclaims go where they matter most — to the human connections that actually move work, and life, forward.
Next · Enterprise
Connection & Collaboration Layer
Within a team, agents quietly absorb the coordination tax — scheduling, status, meeting busywork — and surface the connections that count: routing the person with a question to the one who has the answer, linking the people circling the same problem, turning lost coordination time into real collaboration on ideas and hard technical challenges. Efficiency between people, not just speed for one.
Built on the same privacy foundation — agents exchange suggestions, never your private data, and you always decide what’s shared.
Later · Beyond Work
A Social Connection Layer
The same framework extends outward — agent-brokered, intent-driven introductions that create genuine connections instead of endless feeds. It’s social networking with the busywork taken out and the meaning left in — you connect when there’s a real reason to, with people worth knowing.
A planned expansion of the platform, built on the foundation shipping today.
Our North Star
The first personal assistant that learns to
Not a smarter chatbot. A companion that adapts to your judgment, your priorities, and your way of working — and keeps that understanding yours alone.
Why Back CamelotOS
Built for the agentic-first era.
The market is racing toward autonomous agents. The winners will be the ones people actually trust to act on their behalf — and trust is exactly what we engineer for.
01 — DIFFERENTIATED
A path no one else is taking
While the field optimizes for bigger cloud models, we optimize for privacy, personalization, and a hybrid approach that compounds in value over time.
02 — DEFENSIBLE
An advantage that deepens
Our edge isn't a single feature that can be cloned. It's an architecture and a compounding memory layer that gets stronger with every user and every day.
03 — TIMELY
The right product, right now
As agents move from novelty to daily utility, the demand for one that's private, persistent, and genuinely yours is arriving fast. We're built for that moment.
Join the Waitlist
Be early to the agent that
We’re opening access in waves — the macOS app first, with the iOS companion close behind. Drop your email and we’ll send a confirmation now, then a note when it’s your turn. No spam, no list-sharing, ever.
Join the Team
Help us build the operating system that respects the user.
CamelotOS is being built by a small founding team. If building the privacy-first personal assistant is the kind of thing you’d work nights on, leave a comfortable job for, or stake real time against — we’d like to hear from you.
Who fits
Founding-team operators across the stack.
We’re open to engineers, designers, and operators who can ship without supervision, who care more about the product than the title, and who’d rather build something that matters than work on something that doesn’t.
- Builders who finish things.
- People positioned to take a real risk right now.
- Privacy-first thinking treated as a value, not a feature.
In Summary
We’re not building a smarter agent.
We’re building the operating system that makes one truly yours.
True privacy‑first, human‑level personal and business assistance — achieved not by chasing ever‑smarter models, but through smarter architecture that simply knows you better.