Persistent memory
What should remain known.
Durable facts, history and knowledge that should survive the current task.
Gunner.
The operating system that lets AI decide and act.
GUNNER IN ONE PARAGRAPH
What it is, before any picture.
Technically, Gunner is a stateful runtime built on Linux. It coordinates specialized agents, three forms of memory, probabilistic and declarative reasoning, and software tools within one persistent operating environment, working toward a human-defined objective. Okren is its first commercial use.
Just an exampleThis page uses a restaurant to make each part easy to picture. Lines marked “Analogy (restaurant)” are examples only. Gunner is not a restaurant product, and the restaurant is not a customer.
In real termsMost sections also give the technical description. The pictures are there to help, not to replace it.
The runtime
What Runs Underneath the Workday
On the home page, Gunner does a person’s workday at a computer. This is what is running underneath. Gunner operates as a stateful software runtime above the Linux kernel. Linux provides the underlying environment for processes, networking, storage, and computing resources. Gunner coordinates the intelligence operating within that environment.
The runtime maintains the active objective, current state, relevant memory, agent responsibilities, and available tools. It receives information from connected systems, coordinates specialized work, applies probabilistic and declarative reasoning, executes actions through software interfaces, and observes what happens next.
Analogy (restaurant): Linux is the building and the utilities. Gunner is everything that happens inside once the doors open.
The runtime’s main responsibilities
Linux hostProvides the processes, networking, filesystem, and computing resources on which Gunner operates. Analogy (restaurant): the building, the gas and the electricity.
Runtime stateMaintains the current objective, environmental context, active work, and changing system state. Analogy (restaurant): the ticket rail showing what is being cooked right now.
Listeners and connectorsReceive information through APIs, events, connected applications, and external sources. Analogy (restaurant): the host at the door and the phone for deliveries.
Agent coordinationOrganizes specialized agents and their work across multiple tasks and successive operations. Analogy (restaurant): the expediter who makes sure the plates leave as one meal.
Memory servicesMaintain persistent knowledge, active-session context, and immediate working state. Analogy (restaurant): the regulars’ book, tonight’s reservations and the tickets on the rail.
Hybrid reasoningCombines probabilistic AI for interpretation and exploration with declarative logic for structured reasoning and constraints. Analogy (restaurant): the chef’s taste for the special and the oven timer for when it is done.
Tools and executionConnect to software and external systems, perform actions, observe results, and feed those results back into the ongoing process. Analogy (restaurant): the stoves, knives and ovens that do the actual cooking.
The human question
A Workday Is a System.
A person listens.
Remembers.
Investigates.
Uses different kinds of reasoning.
Asks specialists for help.
Acts.
Watches what happened.
Then carries that experience into what happens next.
The model is one component. The system is the intelligence.
Natural language is the interface. The architecture carries the work.
Why this architecture began
Gunner began as an attempt to do what one person does at a computer, start to finish. That goal led us to build an operating system around the model.
We built Gunner to explore what persistent intelligence needs in order to operate — to understand what is happening, remember what matters, reason, act, observe, and continue.
A familiar point of reference
If you understand the role Palantir’s Ontology plays inside an enterprise, you have a useful starting point for understanding Gunner’s role in persistent intelligence.
For enterprise operations
Palantir’s Ontology provides a shared operational model connecting an organization’s data, relationships, logic and actions.
For persistent intelligence
Gunner keeps the objective, context, memory, reasoning, models, agents and tools coordinated as work continues and the state changes.
The comparison is about architectural roles. These are different systems solving different problems.
The operating loop
It Keeps Going Because It Is a Loop.
One objective can produce many actions. Each result changes the state of the work.
Analogy (restaurant): an order comes in, the kitchen cooks, the plate goes out, and what the diner says shapes the next order.
The outcome of one action becomes the context for the next.
Understand + remember
Context Has Different Lifetimes.
The world changes between prompts. Gunner keeps the right kind of memory for the right span of work.
Analogy (restaurant): the regulars’ book is what stays known. Tonight’s reservations are what this evening needs. The tickets on the rail are what is happening right now.
Session memory
What this body of work needs.
The details that matter while an active objective is under way.
Working memory
What is happening now.
Live agents, tools, environment and the current operation.
Persistent
Maria is a regular. She never eats shellfish.what should remain knownSession
Tonight is her birthday dinner for six.what this body of work needsWorking
Her order has just gone to the kitchen.what is happening nowExample (restaurant): what a system with continuity should get right. It illustrates the idea and is not a deployment.
When the night is over, Maria is still a regular. The temporary state changes.
The proof is not that it remembers the past. The proof is that the past changes what it does next.
3 memories. 3 lifetimes. 1 operating system.
A practical consequence of the design
When the Model Gets Better, Gunner Keeps Going.
An old chat transcript tells a new model what was said. An operating system needs to carry more: what the system is trying to do, what it knows, what changed and what should happen next.
Analogy (restaurant): swap the chef and the restaurant is still there — same kitchen, same orders, same regulars.
The objective belongs to the continuing system, rather than to one conversation.
Memory, decisions and observed results remain part of the work as the world changes.
A different model can be coordinated around the existing objective and state, instead of starting from zero.
This is the architectural direction: continuity belongs to the system, not to a particular model. Full portability across arbitrary models remains a capability to prove.
Reason
Ask AI the same question three times. You may get different answers.
That is how probabilistic AI works. It weighs possibilities and gives you the most likely answer, not a fixed one. For research, language and unclear situations, that is exactly what you want.
A declarative system works the other way. It follows rules.
The same question, asked three times“Is this chicken still safe to serve?”Illustration. Not real output.
Probabilistic AI
Explores. Adapts. Proposes.
It reasons from likelihood. It is strong wherever the answer has to be discovered, and a different answer each time is part of how it explores.
- Asked 1st“Looks fine. Serve it.”
- Asked 2nd“It has been out a while. Hold it.”
- Asked 3rd“Serve it, if it was cooked through.”
Three answers. Useful for exploring. Not what you want from a rule.
Deterministic, rule-based
Follows the rules. Every time.
It starts from facts that are accepted and rules that are written down, then works out what follows. It does not guess, so the same facts give the same result.
The ruleFood left out past the safety limit gets thrown away.
- Asked 1stThrow it away.
- Asked 2ndThrow it away.
- Asked 3rdThrow it away.
Same facts. Same rule. Same answer.
How Gunner uses both
Gunner gives each kind of intelligence the job it is good at. Probabilistic AI explores and proposes. Rules and logic handle what has to come out the same every time.
Analogy (restaurant): the chef’s taste decides tonight’s special. The oven timer decides when it is done. Nobody argues with the timer.
Curious where the answer is unknown.
Exact where it has to be.
Exact where it has to be.
Curiosity finds the options. Rules keep the answer the same every time.
Metacognition
Intent Has to Survive the Journey to Action.
A
What the person asked for.
B
What the model understood.
C
What the model told the tools to do.
Each handoff can change what the system understands or asks a tool to do. The objective must still guide the next action.
Analogy (restaurant): the ticket says “no nuts.” The server writes “light on nuts.” The kitchen adds almonds. The intent has to survive every handoff.
What am I doing? Why am I doing it? Does the next action still serve the objective?
Coordinate + act
Agents Do Tasks. Gunner Runs the Whole Job.
The agent
Has a job.
Research. Calculate. Investigate. Write software. Watch a source. Test an approach.
Gunner
Holds the objective.
The operating system carries the state, memory and coordination around the workers.
Analogy (restaurant): the line cook, the pastry chef and the sommelier each do their own job. The expediter makes sure it arrives as one meal.
An agent performs a job.
Gunner coordinates the system that keeps the work moving.
Gunner coordinates the system that keeps the work moving.
Observe + continue
When Something Breaks, the Result Is New Information.
The tool returned an error.
The software worked.
The research contradicted the assumption.
The environment moved.
That changed state becomes part of the next decision.
Analogy (restaurant): the dishwasher breaks. The expediter reroutes the plates and the night goes on.
ONE TECHNOLOGY SHOULDN’T DO EVERY JOB EITHER
Different Technologies. Different Jobs. One Operating System.
Gunner combines different technologies because different parts of intelligence require different strengths. The innovation is how those pieces work together around the same objective, memory and changing state.
Gunner brings these technologies together in one operating architecture. Linux provides the foundation. The runtime coordinates the other components around the same objective, memory, and changing state.
Analogy (restaurant): a gas range, a walk-in fridge, an oven timer and a ticket printer. Different equipment, one kitchen.
LinuxThe operating foundation.
Gunner is Linux-based. Linux provides the underlying environment in which the larger intelligence system operates.
C++Close to the machine.
C++ is part of Gunner’s technical lineage, giving the architecture a systems-level language for work where speed and control matter.
PythonConnected to the ecosystem.
Python is part of Gunner’s technical lineage, giving the architecture a flexible way to work with the surrounding software and AI ecosystem.
Probabilistic AIHandles the uncertain.
Used for language, research, ambiguity and situations where the answer cannot be completely known in advance.
Declarative LogicHandles what can be known precisely.
Accepted facts, relationships and explicit rules let the system reason consistently where it should not have to guess.
Memory SystemsKeeps context alive across time.
Persistent, session and working memory preserve what should stay known, what matters during the current period, and what is in play right now.
Different technologies do different jobs. Gunner keeps them working as one persistent system.
The proving ground
We Built It to Recreate a Human.
Crypto trading is where we proved its capability because the environment changes constantly and the consequences are real.
Top floorTrusted, higher-consequence responsibility.
Middle floorsSmaller live responsibilities. Approaches earn or lose trust.
Lower floorsResearch. Training. Testing. Competing strategies.
It has run in a live environment: information in, specialized intelligence in parallel, actions proposed, outcomes observed, state carried forward.
The application was trading. The architecture was the point.
2025 · Founder-reported: Gunner ran a live trading account that grew more than 3,000% in under four months.
The turn
Then Capability Became Consequence.
Change software.
Move a machine.
Commit resources.
Place an order.
The question changes when intelligence can make something real.
Capability does not create authority.
The more it can do, the more it matters that people stay in charge. Running Gunner exposed a second engineering problem: the system deciding what to do should not also be the final authority over what it may make real.
Gunner
Operates.
Persistent Intelligence carries the work forward.
Okren
Supervises.
Independent supervision: software that lets a person see every money move before it happens. It runs as separate software inside the customer’s environment and does not require Gunner.
Gunner operates. Okren supervises.
How to check us
Don’t Take Our Word for It.
Here is what is documented, what OA describes, and what it keeps private.
Documented
Public, dated sources show the industry adding persistent memory, agents, tools and runtime responsibilities around models.
Described by OA
Gunner’s architecture: three memory scopes, probabilistic and declarative reasoning, agents as workers, action, observation and continuation. This is OA’s own description.
Founder-reported
Work on Gunner began in 2021. The architecture has run in live environments, including one where its decisions had real consequences.
Not disclosed
Private implementation: schemas, retrieval methods, orchestration internals, prompts, security and source code.
The picture changes if another system integrates the full loop and publishes capability-level evidence, or if models themselves absorb more of these responsibilities.
Questions people usually ask
The Short Answers.
Isn’t this just an agent?
No. An agent is a cook at one station: a worker with a job, tools, memory and some ability to keep working. In the analogy, Gunner is the restaurant. Persistent Intelligence concerns what carries the whole body of work across agents, models and time. In OA’s frame, an agent is a worker inside the system.
Isn’t this just memory?
As an analogy: a restaurant that remembers your name is still not a restaurant that runs a full night. Memory is one responsibility. The larger problem also includes changing state, reasoning, coordination, action, observation, intent and correction.
How is this different from an agent framework?
Frameworks give you ingredients: memory, workflows, tools, controls. in the analogy, Gunner is the restaurant built from them, one system where everything works together on the same job. Two restaurants can use the same ingredients and still be completely different restaurants.
Is Gunner a runtime or an operating system?
Both terms describe different aspects of Gunner. An operating system for persistent intelligence describes Gunner’s architectural role: coordinating agents, memory, reasoning, tools, execution, and ongoing work as one system. A runtime describes the software that actually carries out those responsibilities. Technically, Gunner is a stateful runtime built on Linux. Linux provides the underlying operating-system foundation. Gunner provides the intelligence operating environment above it. Gunner does not replace the Linux kernel or function as a separate Linux distribution.
Does “operating system” mean replacing Linux?
Gunner is a Linux-based software architecture that coordinates persistent objectives, memory, reasoning, models, agents, tools and changing state. “Operating system” describes that coordinating responsibility. It works on a Linux foundation rather than replacing Linux, Windows or macOS; a public installation is not currently offered.
What persists if the model changes?
Analogy: swap the chef and the restaurant is still there. In real terms, Gunner is designed so the work, its state and its knowledge live in the system, not in one model.
How does Gunner relate to Okren?
Gunner operates. Okren supervises. In real terms, Gunner is OA’s architecture for running AI agents, and Okren is separate software, inside a customer’s environment, that lets a person check every money move before it happens. It does not need a Gunner install. Okren came second, as Gunner’s commercial proof, when intelligent action became consequential. See the product.
Is Okren independent of Gunner?
Yes. Okren runs as separate software and does not require Gunner. It supervises any AI, including Gunner’s own money moves.
WHAT WE CHOSE NOT TO DO
What we chose not to do with this technology.
The same architecture could become many things. Here are three we looked at, and why each comes second.
Swipe or tap a use to change the example
A use we passed on · A phone you just talk toAn old handset, a new operating system, and no apps.
Why it comes second
Who it is for
What it would do
Why we passed
Each of these asks someone to trust AI with real responsibility, so each needs the same answer first: what is the AI allowed to do, and who decides? Okren is that answer, starting with money. That is why it is the product.
WHERE TO GO NEXT
Gunner operates. Okren supervises.
The instant check before the money moves, like every card payment.
Okren is where this research goes to work: independent supervision for AI that moves money, live with its first paying customer.
Why is Okren called the commercial proof?
Gunner is OA’s architecture. Okren is the first product built from it, put to a real job with a real customer need. It is separate software and runs on its own.
Go deeper
Want to Talk Through the Technology?
Technical conversations go deeper into runtime, system diagrams and the evidence behind the architecture.