Think, decide and create.
DO
SEE
TRACK
A shared workspace for understanding situations, resolving questions, making decisions and creating work from the organisation’s maintained understanding.
Human interface to the model
FOR INVESTORS
Intelligence is becoming abundant. Organisational coherence is not.
Human-machine organisations need a state layer.
Models provide increasingly powerful intelligence. Agents provide increasingly powerful agency. But neither, by itself, maintains the organisation: what it currently believes, what it is trying to achieve, what constrains it, what has changed, what has been decided, and what those changes mean for what happens next.
Orient maintains that state continuously.
Every new observation is interpreted against the organisation’s existing reality and intent. Evidence can strengthen, qualify or contradict what is believed. Questions can resolve or reopen. Situations can change. Goals can become threatened. Constraints can become relevant. Decisions can require reconsideration. People and agents then reason and act from the same current state.
We call this Continuous Cognition.
Models reason. Agents act. Orient maintains the shared reality they operate from.
Reality
What is true, believed, uncertain, contested and decided?
Intent
What are we trying to achieve, under whose authority and for how long?
Judgement
Given what just happened, what should change, what should be preserved and what deserves attention?
Action
What are humans and machines now safe and authorised to do?
Orient exists because organisations do not have an information problem. They have a shared-reality problem.
AI can now generate, search, summarise and act across an organisation at extraordinary speed. But it does not inherently know what the organisation believes, what evidence supports those beliefs, what remains uncertain, what has been decided, or what has changed.
Every agent becomes another participant in the organisation. More intelligence and more activity make shared understanding more important—and harder to maintain.
Intelligence accelerated. Shared understanding did not.
We built Orient after watching this gap appear again and again. We saw increasingly capable models produce answers, recommendations and actions without knowing the reality of the organisation they were operating inside. They could sound certain without knowing what was contested. They could act without knowing what had already been decided. They could retrieve information without understanding which beliefs the organisation actually trusted.
We had encountered the same failure from different directions, across reasoning systems, language models, agents, optimisation and machine perception. The machines kept getting better. The world they needed to understand remained fragmented across people, documents, systems and time.
Orient is the missing layer.
It continuously forms and maintains a live model of organisational understanding and intent: what is known, believed, uncertain, contested and decided, alongside the goals and commitments the organisation is pursuing. That model is grounded in organisational work, updated as new evidence arrives, and shared by the people and machines that act from it.
Understanding becomes shared state. Shared state becomes coordinated action.
Orient maintains the model and determines what needs attention as reality changes. We call this Continuous Cognition.
The Orient Team
Stockholm, Sweden
PART I
The thesis
Why human-machine organisations need persistent shared state and why Orient is the layer that maintains it.
THESIS — 01 · WHY NOW
Every agent joins the work.
The first wave of AI sat beside people. It searched, summarised, generated and recommended. Humans still carried the context, judgement and memory of the organisation.
Now AI is entering the work itself.
Agents research customers, monitor operations, use tools, update systems, coordinate tasks and execute workflows. They are beginning to act on the same customers, plans, decisions and systems as people.
That creates a new failure mode.
A decision changes, but an agent keeps acting on yesterday’s version. Product updates an assumption, but sales and its agents never see it. One system treats something as settled while another still treats it as uncertain.
Every participant can be capable. The organisation can still become incoherent.
As agents multiply, intelligence does not automatically create alignment. Without shared understanding, it can multiply divergence.
The next constraint is not intelligence. It is coherence.
AI is adding participants to organisational work faster than organisations can give them a shared reality.
01 · HUMAN ORGANISATION
Humans carry coherence informally
Why we decided this
What we currently believe
What is uncertain
What changed
Who has authority
The organisation depends on human memory and judgement.
02 · MACHINE-AUGMENTED
Every actor reconstructs reality
Capable actors. No common centre.
AGENT A
Feature X is priority
AGENT B
Security is priority
AGENT C
Decision pending
WORKFLOW D
Old decision still active
NO COMMON STATE
≠
More capability. More divergence.
03 · COHERENT ORGANISATION
Independent actors. Shared reality.
SHARED ORGANISATIONAL STATE
One compatible shared state
People
Agents
Applications
Workflows
Coherence becomes infrastructure.
Models change. Agents come and go. The organisation’s understanding must endure.
Orient preserves what the organisation understands, intends and has authorised, so each new participant can continue from the state already established. Evidence, decisions, unresolved questions and history remain available as models, agents and providers change. The understanding accumulated by the organisation becomes an enduring asset.
The constraint is becoming visible.
Anthropic
Jakub Pachocki · Chief Scientist, OpenAI
OpenAI’s Chief Scientist does not name Orient. He describes increasingly powerful AI systems that act alongside people and other AIs, become harder to inspect, and challenge our ability to keep humans in control.
Pachocki describes the problem.
Orient is our architectural response.
Our thesis: if humans must remain in control of intelligence they cannot fully inspect, shared reality, authority, evidence and human judgment cannot live only inside the model. They need an institutional layer outside it. Orient is building that layer.
Make organisational cognition legible.
As more organisational reasoning is performed by machines, people need to inspect what was believed, why it changed, which evidence mattered, who authorised the decision and what followed. Orient’s institutional role is to make that history visible and open to challenge.
Independent research; not endorsements of Orient.
THESIS — 02 · WHAT ORIENT IS TODAY
One live model.
Two ways to operate from it.
People and agents work from the same organisational understanding.
Research, monitor and assist from shared context.
UNDERSTAND
OPERATE
Authority · Constraints · Human review
ACT
Reason · Propose · Update
Agents operate from maintained organisational context before they reason or act, rather than reconstructing the organisation independently.
Machine participants in the model
The live organisational model.
A continuously updated model of what the organisation knows, believes, decides, does and learns.
The model everything operates from
What authority does this action require?
Intelligence can recommend an action. The organisation’s maintained state establishes the evidence, intent and authority against which that action must be judged.
People can authorise bounded classes of actions, with explicit scope, conditions and limits. The proposed Orient Control layer would use this state to check whether those conditions still hold and route exceptions or changed circumstances to human judgement.
One change. A shared understanding of what it means.
Continuous Cognition in Action
Your Organisation and the World
People and Conversations
Meetings · messages · expertise
Documents and Knowledge
Files · plans · shared knowledge
Research and Discovery
Studies · search · investigation
News and World Events
Markets · events · regulation
Apps and Systems
Data · events · internal outputs
Work and Agent Outputs
Analysis · actions · results
OrientOne Live
Model.
Organisational understanding.
Continuously maintained.
Continuously maintained.
Working Assumption
Supplier expected to qualify.
Current Understanding
March launch
is on track.
is on track.
Existing Commitment · Preserved
March promised to the client.
Dependency
Launch depends on supplier qualification.
Evidence AttachedTraceable
Monitoring AssumptionsState 01
From the Same State
People
Confirm launch
readiness.
AI Agents
Monitor supplier
readiness.
Findings, decisions and outcomes return to the model.
Understanding changes. Commitments change through authorised decisions.Illustrative launch scenario
Continuous Cognition
Organisational understanding that keeps working—even when nobody asks.
Orient continuously maintains a live model of what the organisation knows, believes, has decided and is trying to achieve. As evidence, decisions and circumstances change, it determines what deserves attention: what to reconsider, investigate, correct, preserve or escalate to human judgement.
The ambition is an organisation whose understanding keeps pace with its activity—across every person, agent and system.
The live model is what Orient maintains. Continuous Cognition is how that understanding stays alive.
What changed—and why does it matter to what we are trying to achieve?
The architectural ambition brings reality and intent into the same maintained model: goals, objectives, principles, commitments and measures, with owners, authority and time attached. People establish and ratify that intent. Orient can then identify which changes affect it and propose what needs attention.
Sensemaker: Continuous Cognition applied to an objective.
Consider a research agent reading a new paper. It interprets the findings, relates them to existing beliefs and an authorised research goal, traces the consequences for product and research decisions, and determines whether anything deserves attention now. The next step may be to update or preserve a belief, challenge an assumption, open a question, propose research or abstain. This illustrates how the architecture turns new information into judgement in service of an objective.
Form → Transition → Reconcile
Form what evidence justifies. Transition what new evidence changes—and preserve what remains valid. Reconcile accumulated understanding against the underlying observations. Discrepancies return as proposed changes, subject to the same judgement and authority.
Shared understanding can include unresolved disagreement. The model preserves what is uncertain and who has authority to decide.
THESIS — 03 · WHY IT COMPOUNDS
Orient gets harder to copy every time it is used.
The advantage is not the information Orient ingests. It is the organisational understanding created as that information is interpreted, challenged, resolved and acted on.
01
Meaning must persist
Claims, evidence, assumptions and decisions must survive individual conversations and applications.
02
New evidence must reconcile
Each observation must be understood in relation to what the organisation already believes.
03
Change must propagate
When understanding changes, dependent decisions, plans and expectations must be surfaced.
04
Change and action must be governed
Evidence, uncertainty, authority and ratification must remain explicit.
HISTORY CREATED THROUGH USE
01
What was believed
02
What challenged it
03
What evidence mattered
04
What was accepted or rejected
05
What was ratified
06
What was superseded
07
What was decided
08
What action followed
09
What was expected
10
What actually happened
PART II — THE BUSINESS
Orient is live. The next questions are commercial: who buys first, how adoption expands, why the model compounds, and how this becomes infrastructure.
COMPANY — 01 · PROOF
The core system is built and running.
These are operating product views—not concept diagrams. Source activity enters the workspace while Orient maintains a current view and status through time.
INSIDE THE OPERATING PRODUCT
Current state
A maintained situation view: current understanding, change status and the assumptions that still need attention.
Evidence entering the model
New source material enters the workspace and joins the context the organisation already holds.
BUILT AND RUNNING
The core product is live on real organisational material. It maintains sources, questions, situations, current views and status as new information arrives.
THE NEXT PROOF
Repeatable customer deployment.
COMPANY — 02 · WHY US
Built by people who have lived both sides of the problem.
Orient combines deep experience in reasoning systems with the company-building discipline required to make new infrastructure adoptable.
Mikael Larsson
Co-founder · Company + Commercial
Mikael turns technical ambition into an operating company. He has spent 20+ years building and scaling SaaS, digital media and technology businesses from 0→1 through commercial growth.
SELECTED EXPERIENCE
Co-founder, Learnster · Former MD, DigitasLBi Nordics · fundraising, organisation building and commercial scale
Jay Solomon
Co-founder · Product + Intelligence
Jay conceived Orient’s thesis and built the system. He has spent 20+ years building AI and reasoning systems across intelligence, industry, retail and media.
SELECTED EXPERIENCE
Cambridge Mathematical Tripos, Signal Processing · UK Government · Mitsubishi Electric Research Laboratories · Northvolt
The technical depth to build a new organisational layer, paired with the company-building experience to make it adoptable.
CUSTOMER + WEDGE
Start where shared reality matters most.
Orient enters through a high-context organisational problem: a situation where decisions, evidence, assumptions, priorities and outputs are changing continuously and several participants need to remain aligned.
ONE WEDGE
High-context organisational problem
That problem may begin inside a single team or span several functions. The underlying wedge is the same: an important part of the organisation needs a reliable, persistent shared reality.
ENTRY SHAPE 01 · CONTAINED ENTRY
A team with a recurring high-context problem
A strategy team maintaining a changing body of decisions, evidence, assumptions, priorities and outputs.
Orient can begin with a bounded organisational domain where the cost of lost context and repeated reconstruction is already visible.
ENTRY SHAPE 02 · CROSS-FUNCTIONAL ENTRY
A situation that crosses organisational boundaries
A major programme where product, operations, finance and leadership need to maintain the same evolving understanding.
The initial domain can also span several teams when the problem itself is inherently cross-functional.
The initial domain is deliberately bounded. Once Orient becomes useful there, the same shared reality can extend into adjacent teams, workflows, systems and machine participants.
THE COST ORIENT REDUCES
Reconstruction cost
Teams rediscover context already known elsewhere.
Coordination cost
Actors decide from inconsistent assumptions.
Decision latency
Understanding is assembled manually before action.
Error cost
Stale assumptions propagate into plans and execution.
AI coordination cost
Every agent reconstructs incomplete context.
COMPANY — 04 · GO TO MARKET
Start bounded.
Expand from the state created there.
We begin with decisions that cross organisational boundaries—not with a mandate to model the whole organisation on day one.
ENTERPRISE SALES MOTION
01
Design partner
Choose one recurring decision and agree a measurable baseline.
02
Bounded deployment
Maintain the relevant state and agree a bounded intervention.
03
Prove decision value
Measure the outcome and revise understanding from what happens.
04
Expand the model
Pull in adjacent teams and decisions.
05
Connect agents + systems
Turn shared state into infrastructure.
THE GTM INSIGHT
We do not need to model the whole organisation on day one. We begin where fragmented understanding is already expensive and expand from the state created there.
Each deployment should follow a complete decision loop: establish a baseline, make a bounded intervention, observe the outcome and revise the shared understanding. We will measure relevant outcomes such as reconstruction time, earlier detection of consequential changes, decision latency or avoidable error. Commercial proof comes from improvement against the agreed baseline.
Continuous attention. Selective computation.
How much intelligence is this decision worth?
Most organisational state needs no new computation most of the time. The research goal is to choose the next useful step: use a model, retrieve evidence, ask a person, wait or stop. Routine judgements should use inexpensive specialised models; deeper reasoning is warranted when uncertainty or consequence makes it valuable.
The target is a lower total cost of correctly maintained organisational understanding. That includes inference, retrieval, verification, human attention and the cost of delay. The test is whether Orient can preserve reliability while spending less effort on work that does not change the decision.
We will test specialised models and selective escalation against both capable low-cost APIs and frontier-model baselines. Savings and reliability improvements are research outcomes to demonstrate, not assumptions behind the business.
Owned inference offers capacity and deployment control, but does not automatically beat API pricing. These experiments will establish how continuous maintenance can support the commercial margin targets.
THE BUSINESS
The economics
Land in a bounded domain. Expand as the model becomes more valuable.
Initial deployment size varies with the scope of the organisational domain — from contained team or use-case deployments to larger cross-functional environments.
01
LAND — Start with a bounded organisational domain where shared reality has immediate value.
The initial deployment may cover a contained team, recurring use case or larger cross-functional situation. Scope determines the initial commercial size and deployment model.
Larger enterprise deployments are expected to begin around €50k–€100k ACV.
€50k–€100k ACV
< 14 days
Illustrative larger-deployment footprint
1 organisational domain
Initial model scope
The objective is not an enterprise-wide transformation project. It is to establish useful shared state around one problem where fragmented understanding already has a measurable cost.
02
MODEL — Orient builds a persistent model of the domain.
People, decisions, evidence, priorities, dependencies, history and active situations accumulate into organisational state. The initial cost of forming the model is concentrated early; its value increases as continuity and context accumulate.
The highest implementation cost is concentrated in establishing the initial domain. Subsequent participants and adjacent use cases build on state that already exists.
The marginal cost of expansion falls while the value of the accumulated model increases.
The existing organisational model expands through three connected surfaces:
More organisational domains, systems and history represented in the live model.
More people understanding, deciding and creating from shared organisational state.
More machine work operating from the same maintained context.
Seat-based recurring revenue
Capacity and usage-based recurring revenue
This is important because Orient’s long-term economics are therefore not bounded by employee count.
3–7×
Account expansion
A €70k land growing into €350k is 5× account expansion.
You don’t need massive logo acquisition if successful deployments systematically grow in that way.
Land → Model → Expand → Machines
€€
Account expansion
COMPANY — 07 · MARKET + CATEGORY
The market expands with machine agency.
Orient begins in organisational intelligence and decision coordination. As persistent shared state becomes necessary for people, agents and applications, the opportunity expands beneath multiple software categories.
THE STATE LAYER
Shared organisational reality
SITS BENEATH
Enterprise AI
Agents
Knowledge work
Strategy
Operations
Organisational memory
Workflow automation
Machine governance
THE CATEGORY EXPANSION
The opportunity is not to replace one category. It is to become a common state layer beneath several of them.
PRODUCT EXPANSION
Shared reality creates new expansion paths.
Orient first establishes a persistent shared reality for the organisation. Once that foundation exists, new products can be built on top of it — using the same organisational state, history, evidence, decisions and context.
CURRENT FOUNDATION
Creates a persistent shared organisational reality.
Orient Core maintains the organisation’s beliefs, decisions, goals, commitments, evidence, authority and history through time.
This shared organisational reality is the asset that unlocks expansion.
Shared organisational reality
Two potential expansion paths become especially powerful once Core is established.
POTENTIAL EXPANSION
Build from what the organisation actually needs.
Once Orient understands the organisation’s live state, recurring needs can become purpose-built tools, workflows and agents — grounded in the same shared organisational reality.
Build is a downstream opportunity created by Core, not a separate starting product.
Potential applications include:
Tools · Workflows · Agents
Research agents · Briefing agents
Approval workflows · Monitoring agents
Internal decision tools · Reporting and reconciliation tools
Expansion logic
Core supplies the organisational state; Build turns recurring needs within that state into purpose-built capability.
POTENTIAL EXPANSION
Control machines through the same shared reality.
As more agents operate inside the organisation, Orient can provide a live control layer for their objectives, authority, capabilities, actions, contradictions and divergence from organisational intent.
Orient can eventually govern machine participation because it already understands the organisational state those machines are acting within.
A future control layer could maintain a view of:
Objectives · Authority · Capabilities · Actions · Contradictions · Divergence
This is not a generic agent-monitoring dashboard. Its advantage comes from connecting machine behaviour to the organisation’s live state, decisions and intent.
The governing question becomes:
Not simply:
Can this agent perform this action?
But:
Why is it doing this, what organisational objective does it serve, who authorised it, and should it still be doing it?
Core maintains what the organisation understands and what it is trying to achieve, with evidence and authority attached. Control can then assess whether machines are acting consistently with both—a downstream capability built on the same state.
Machine governance becomes a downstream capability of shared reality.
These are not separate starting products. They are expansion opportunities created by the same underlying asset: a persistent model of organisational reality.
COMPANY — 08 · THE MOAT
The moat compounds at four levels.
The defensibility is architectural, proprietary, embedded and cumulative.
Architecture
Orient maintains temporal, epistemic organisational state rather than reconstructing isolated context.
Proprietary state
Use creates information that did not previously exist. That history cannot simply be re-imported from source systems.
Embedded dependency
As humans, agents and applications increasingly operate from Orient, the model becomes operational infrastructure.
Organisational learning
Every resolved uncertainty, decision and observed outcome improves the model from which future work begins.
THE DEEPEST MOAT
The deepest moat is not the software alone. It is the organisational history that only exists because Orient was there while the organisation learned.
COMPANY — 09 · AMBITION
We do not need to own general intelligence.
We intend to own the intelligence that maintains organisational reality.
Frontier models will continue to become more capable, more numerous and increasingly interchangeable.
Orient is building a different kind of proprietary intelligence.
Every time an organisation receives new evidence, something has to determine what that evidence means in relation to everything already known:
01
What changed?
02
What remains true?
03
What is now uncertain?
04
What contradicts what?
05
Which decisions are affected?
06
And when is it safe to act?
This is the intelligence required to maintain a live organisational model through time.
STATE T₀
Current organisational reality
+
NEW EVIDENCE
JUSTIFIED CHANGE
Evidence · Contradiction · Time · Uncertainty · Authority · Ratification
STATE T₁
Updated organisational reality
Today, general-purpose models can help perform parts of this process.
Over time, Orient can increasingly own the specialised systems that perform it: the models, organisational state, judgement data, benchmarks, evaluation systems and learning infrastructure required to maintain organisational reality.
USE
Organisations use Orient
JUDGEMENT
Real state-transition problems appear
DATA
Difficult judgements become proprietary data
INTELLIGENCE
Specialised intelligence improves
Orient maintains reality more accurately
DEPENDENCY
More people, applications and agents can depend on it
↺ USE
DEPENDENCY
General intelligence becomes an input.
Orient does not need to build the model that knows everything.
Explore Orient Research →
COMPANY — 10 · COMPETITION
Incumbents are built around a different starting point.
Other companies can reproduce individual capabilities. That is not the advantage.
Orient begins with a different core model: maintained organisational understanding over time.
For an incumbent to become Orient, it would have to rebuild around that primitive, changing not only features, but its data model, architecture and product logic. That is a much larger move than adding another AI capability.
WHAT EACH CATEGORY OPTIMISES
Search companies
Retrieval—finding relevant information.
Work platforms
Workflows—coordinating tasks, projects and communication.
Action—giving machines tools and agency.
Databases + graphs
Representation—storing entities, relationships and state.
Foundation model labs
Intelligence—improving general reasoning.
THE STRATEGIC QUESTION
Would an incumbent have to change its core model, architecture and incentives to become Orient?
A DIFFERENT CORE PRIMITIVE
Maintained organisational understanding over time.
Evidence, beliefs, uncertainty, decisions, dependencies, authority and outcomes remain connected as the organisation changes.
Others give intelligence better context.
Orient gives the organisation a shared reality.
COMPANY — 11 · RISKS
What we still need to prove.
A working system establishes technical feasibility. The company thesis now depends on answering four commercial questions.
Adoption
Will organisations adopt a persistent model across existing boundaries?
Wedge
Which initial problem produces the fastest path to repeatable value?
Expansion
Does a successful deployment naturally pull in adjacent teams, agents and systems?
Economics
Can Orient be sold and deployed with enterprise software margins?
COMPANY — 12 · THE ROUND
From a working system to repeatable deployment.
This round funds product hardening, core integrations, implementation capacity and a focused enterprise sales motion.
WHAT THIS ROUND PROVES
To prove that maintained understanding reduces reconstruction work, catches consequential changes earlier and improves a recurring organisational decision—then expands as more teams and agents depend on the same model.
THE VENTURE OUTCOME
If we are right.
Organisations will not allow every human, model, agent and application to maintain its own incompatible version of reality. They will need a persistent shared model beneath them.
01
More workflows begin there
02
More applications read from it
03
More agents depend on it
04
More organisational history accumulates inside it
05
Switching costs increase
06
The model becomes strategic infrastructure
IF ORIENT OWNS THAT MODEL