Document 1 · v0.1
The Agentic Organization
Executive Summary
Every major technology wave has changed how organizations coordinate people, information, and capital. The agentic AI era is different because it introduces a new participant inside the organization: autonomous digital workers.
This is not simply another productivity improvement. It changes the composition of the workforce itself.
Today's enterprise software was designed around human work. Departments, approval chains, workflows, and SaaS applications all reflect the assumption that humans are the primary executors of business processes.
As AI agents become capable of planning, reasoning, collaborating, and executing work across systems, organizations will need to manage both human and AI workers. This shift creates new operational challenges that existing enterprise software was not designed to solve.
The companies that define the next decade of enterprise software will likely build the infrastructure required to operate AI-native organizations, not merely AI-powered applications.
Central Idea
The defining change of the next decade is not that AI becomes another software feature. It is that organizations gain a second workforce.
For over a century, organizations have been designed around one type of worker: humans.
The next generation of enterprises will operate with two:
- Human workers
- AI workers
Once organizations employ both, every operational system, from management to identity, governance, security, finance, and software architecture, must evolve.
Organizations Exist to Coordinate Work
At their core, organizations solve a coordination problem.
They exist to answer four questions:
- What needs to be done?
- Who should do it?
- How do we know it was done correctly?
- How do we improve over time?
Historically, the answer to the second question has almost always been a person.
This assumption shaped everything that followed.
Departments exist because humans specialize.
Managers exist because humans require coordination.
Enterprise software exists because humans need systems to organize work.
Agentic AI challenges this assumption.
AI Is Not Another Tool
Previous software made people more productive.
Spreadsheets made accountants faster.
CRM made sales teams more organized.
Email made communication easier.
Cloud software made collaboration simpler.
In every case, the human remained the primary executor.
Agentic AI changes the executor.
Instead of helping people perform work, AI increasingly performs work on behalf of people.
That distinction fundamentally changes how organizations operate.
The Enterprise Gains a Second Workforce
Future organizations will consist of three components:
- Human workers
- AI workers
- Enterprise systems
AI workers will not replace every employee, nor will they function as independent organizations. Instead, they will become part of the workforce, handling well-defined operational responsibilities while humans retain accountability for strategy, judgment, relationships, and governance.
This introduces a new management challenge:
Organizations must coordinate work across two fundamentally different types of workers.
Departments Become Capabilities
Traditional departments exist because expertise is organized around people.
Marketing hires marketers.
Finance hires accountants.
Legal hires lawyers.
In AI-native organizations, capabilities become more important than departments.
For example:
Customer Success may consist of:
- Human account managers
- AI onboarding agents
- AI support agents
- AI renewal analysts
- Human escalation specialists
The organizational boundary is no longer "who belongs to marketing" but "what capability delivers customer outcomes."
This makes organizations more modular and adaptive.
Managers Become System Designers
The role of management shifts.
Today, managers spend much of their time:
- assigning work,
- tracking progress,
- coordinating people,
- removing blockers,
- approving routine decisions.
As AI workers take on more operational execution, managers increasingly design and improve the systems in which both humans and AI operate.
Their responsibilities shift toward:
- defining goals,
- designing workflows,
- setting guardrails,
- allocating responsibility,
- monitoring performance,
- deciding when humans should intervene.
Management becomes less about supervision and more about organizational design.
Software Stops Being the Workplace
Today's employees spend much of their day inside software.
CRM.
ERP.
Project management tools.
Help desk platforms.
In the AI-native enterprise, software becomes operational infrastructure.
AI agents interact directly with enterprise systems through APIs.
Humans increasingly supervise outcomes rather than manually operating software.
Enterprise applications become systems that agents use, while humans move toward oversight, exception handling, and strategic decision-making.
Organizational Memory Becomes Strategic
Today's organizational knowledge is fragmented across documents, emails, chats, and individual employees.
AI workers require consistent context to perform reliably.
This creates demand for a shared organizational memory that captures:
- policies,
- processes,
- decisions,
- historical context,
- customer knowledge,
- operational knowledge.
Organizational memory becomes foundational infrastructure rather than documentation.
Decision-Making Becomes Distributed
Today's organizations centralize many operational decisions because humans are scarce and expensive.
AI workers allow thousands of low-risk operational decisions to occur continuously.
Examples include:
- prioritizing support tickets,
- reallocating marketing budgets within predefined limits,
- identifying procurement anomalies,
- adjusting inventory forecasts,
- escalating compliance risks.
Humans increasingly focus on strategic, ambiguous, and high-consequence decisions.
Operational decisions become distributed throughout the organization.
New Constraints Replace Old Ones
The limiting factor shifts.
Historically, organizations optimized for human productivity.
Future organizations optimize for coordinated execution across humans and AI workers.
New constraints emerge:
- Trust
- Governance
- Identity
- Delegation
- Accountability
- Organizational memory
- Evaluation
- Cost management
These constraints are likely to define the next generation of enterprise infrastructure.
What Doesn't Change
Despite rapid advances in AI, several organizational principles remain constant.
Organizations will still require:
- trust,
- accountability,
- governance,
- compliance,
- human judgment,
- leadership,
- incentives,
- culture.
AI changes how work is executed, not why organizations exist.
Companies that ignore these enduring principles are unlikely to succeed.
Implications
If this idea holds up, several conclusions follow.
Organizations will manage AI workers as operational resources.
Not because AI is human, but because autonomous execution introduces lifecycle management, permissions, monitoring, and performance considerations that resemble workforce management.
Enterprise software shifts from recording work to executing work.
Systems of record remain important, but competitive advantage increasingly moves toward systems that coordinate and execute work autonomously.
New enterprise infrastructure becomes inevitable.
Organizations will require capabilities that traditional SaaS platforms were never designed to provide, including AI identity, governance, orchestration, memory, evaluation, and operational visibility.
Competitive advantage moves up the stack.
As foundation models become more capable and more widely available, value shifts toward the infrastructure that enables organizations to deploy them safely and effectively.
Open Questions
This document intentionally leaves several questions unresolved.
They should guide what we explore next rather than be answered prematurely.
- What is the minimum infrastructure every AI-native enterprise requires?
- What becomes the new system of record?
- Which organizational constraints become the largest markets?
- Which existing enterprise platforms adapt successfully?
- Which new buying centers emerge?
- How quickly will large enterprises adopt autonomous execution?
- What regulatory or governance barriers slow adoption?
- Which industries transition first?
Where We're Starting
The next generation of enterprise software will not be defined by better AI applications. It will be defined by the infrastructure required to operate organizations composed of both human and AI workers.
The most valuable companies of the agentic era may not build the smartest agents.
They may build the operating infrastructure that makes employing millions of AI workers practical, secure, and trustworthy.
Predictions
Good frameworks make claims that can be proven wrong. The following are not certainties. They are predictions we track over time, updating as we learn from the market and from conversations with people inside organizations.
1. AI Workforce Management Becomes a New Enterprise Category
Organizations will eventually manage hundreds or thousands of AI workers alongside human employees. This creates demand for platforms that provision, monitor, govern, and optimize AI workers across the enterprise.
What would validate this?
- Enterprises begin tracking the number of AI agents deployed.
- New budget lines emerge for AI workforce management.
- Vendors position themselves around operating AI workers rather than AI assistants.
2. Every Enterprise Builds an Internal AI Platform Team
Just as cloud engineering and platform engineering became standard functions, organizations will establish teams responsible for AI infrastructure, governance, and operations.
These teams will own the internal AI platform rather than individual AI applications.
What would validate this?
- Job postings for AI Platform Engineering, AI Operations, or similar functions become common.
- CIOs and CTOs describe AI as enterprise infrastructure rather than isolated projects.
3. Systems of Record Are No Longer Enough
Traditional enterprise software will continue to store business data, but competitive advantage shifts toward systems that coordinate and execute work.
Organizations will increasingly evaluate software based on how much work it autonomously completes, not simply how well it records information.
What would validate this?
- Enterprise buying criteria shift from reporting capabilities to execution capabilities.
- Vendors compete on autonomous outcomes rather than workflow automation.
4. Organizational Memory Becomes Strategic Infrastructure
AI workers cannot operate effectively with fragmented knowledge.
Enterprises will invest in persistent organizational memory that provides consistent context across people, agents, and applications.
This becomes as foundational as identity or cloud infrastructure.
What would validate this?
- Companies create dedicated initiatives around enterprise knowledge architecture.
- Organizational memory becomes a standalone software category rather than a feature.
5. Identity Expands Beyond Humans
Identity systems will increasingly manage AI workers alongside employees, contractors, applications, and service accounts.
Permissions, accountability, auditability, and trust will extend to autonomous actors.
What would validate this?
- Identity providers introduce first-class support for AI agents.
- Security teams begin measuring and governing AI identities.
6. AI Governance Moves from Compliance to Operations
Governance will no longer focus solely on approving AI usage.
It will become an operational capability that continuously monitors behavior, risk, cost, and performance across thousands of AI workers.
What would validate this?
- Enterprises deploy continuous AI monitoring platforms.
- Governance budgets shift from policy creation to operational tooling.
7. Human Work Shifts Up the Value Chain
Humans will spend less time executing repeatable operational tasks and more time defining goals, making strategic decisions, exercising judgment, building relationships, and handling exceptional situations.
The manager's role increasingly becomes designing systems rather than assigning tasks.
What would validate this?
- Job descriptions emphasize orchestration, oversight, and system design.
- Operational metrics increasingly measure leverage rather than individual output.
8. A New Enterprise Buying Center Emerges
Within the next decade, large organizations will establish a dedicated function responsible for deploying and operating AI across the enterprise.
This may evolve from IT, engineering, or security, but it will eventually become a distinct organizational capability with its own budget and executive ownership.
What would validate this?
- New executive titles become common (e.g., Head of AI Platform, VP of AI Operations).
- Procurement processes distinguish AI infrastructure from traditional SaaS.
9. Enterprise Value Shifts Toward AI Infrastructure
As foundation models continue to improve and become more accessible, sustainable enterprise value will increasingly accrue to companies that provide the infrastructure enabling organizations to safely, reliably, and efficiently employ AI at scale.
The long-term winners are more likely to own operational infrastructure than individual AI applications.
What would validate this?
- Infrastructure companies consistently achieve larger enterprise contracts than single-purpose AI applications.
- Customers consolidate around a small number of foundational AI platforms.
10. Every Enterprise Will Operate a Hybrid Workforce
The future organization will not be fully autonomous.
It will consist of humans and AI workers operating together, each contributing where they create the greatest value.
The defining challenge of the next decade will not be replacing people with AI, but designing organizations where both can work together effectively.
What would validate this?
- Enterprises report AI worker counts alongside human workforce metrics.
- Organizational design frameworks explicitly incorporate both human and AI roles.
What Would Change Our Mind?
We should actively seek evidence that disproves our assumptions.
We would need to reconsider this view if one or more of the following occur:
- AI agents remain narrow tools rather than trusted autonomous workers.
- Enterprises overwhelmingly prefer department-specific AI applications over shared infrastructure.
- Existing enterprise platforms absorb these capabilities so effectively that no new infrastructure layer emerges.
- Regulation, security, or liability concerns prevent organizations from deploying AI agents at meaningful scale.
- The economics of autonomous AI remain too expensive relative to human labor for broad adoption.
- Organizational change proves to be the primary bottleneck, limiting adoption despite technical progress.