How AI agents change org design

A specific question we get from leadership: "How does deploying AI agents change how we should structure our team?"

The honest answer: it should change it, but most companies haven't thought through how.

This post is for founders, CEOs, COOs, and org design leaders thinking about how the agent era affects how companies should be built.

The traditional org design logic

Companies are typically structured around:

→ Functions (engineering, sales, marketing, ops, finance) → Hierarchical reporting (ICs → managers → directors → VPs → C-suite) → Headcount scaling with revenue (more revenue = more people) → Specialization (people specialize as company grows)

This structure has worked for decades because it solved a specific problem: how to coordinate many humans doing knowledge work.

AI agents change one of the underlying assumptions: that knowledge work scales with humans.

What changes

Agent leverage scales differently from human leverage.

Hiring a person = step function increase in capacity, with overhead (salary, management, ramp time).

Deploying an agent = continuous increase in capacity, with low marginal cost. Once configured, runs at scale.

This affects optimal team size for given throughput.

Functions blur at the operational layer.

When agents absorb operational work across functions, the lines between marketing ops, sales ops, support ops, HR ops blur. The agents are doing similar work (drafting, classifying, summarizing) across functional contexts.

The "agent ops" role we covered in [post 225] sits across functions in a way that traditional ops roles didn't.

Specialization shifts.

People specialize less in operational tasks (because agents absorb those) and more in: → Judgment work → Relationship work → Strategic work → Creative work → Cross-functional integration

The skills that matter shift.

Managers' span of control can expand.

When agents absorb operational management work (status reports, scheduling, follow-ups, basic coaching prep), managers can effectively oversee more people. Direct report counts that previously felt unmanageable become manageable.

The "department of one" model becomes more viable.

For specific functions (HR, finance, legal at smaller scales, design at smaller scales, marketing in some configurations), one experienced person + agents can do work that previously required a team.

This is most visible in solo operators (see [post 153]) but scales upward.

What this means for hiring

Hire fewer "ops" roles. Hire higher-leverage strategic roles.

If agents absorb 60-80% of operational work, you don't need as many people whose primary job is operations.

You DO need more people who can: → Build relationships (sales, customer success, partnerships) → Make judgment calls (senior managers, strategic leadership) → Create (designers, writers, product strategists) → Drive cross-functional initiatives (program managers, ops leaders)

The hiring mix changes. Less middle-of-the-pyramid operations. More senior + creative + relational.

Hire generalists who can leverage agents.

The most valuable individual contributors become people who can: → Identify automation opportunities in their own work → Configure and maintain agents for their workflows → Train teammates on agent usage → Bring cross-functional perspective

These are generalists with AI fluency. They're harder to find than specialists in narrow functions.

Hire fewer junior-level operational roles.

The traditional "hire junior people to do operational work, train them up over years" model breaks when agents absorb the operational work.

This is uncomfortable. It also creates a separate career-pipeline problem: where do senior people come from if junior roles are diminished?

We don't have a clean answer to this. The companies that figure it out will have an advantage.

What this means for management

Managers manage more people, each at higher leverage.

The 1:8 manager-to-IC ratio that's been roughly standard could expand to 1:12 or 1:15 in agent-enabled orgs.

The ICs are doing more strategic work, less operational drudgery. Managers spend less time on routine status work, more on coaching + decision-making + cross-functional integration.

The middle of the org chart compresses.

Some companies are eliminating middle management levels. The reasoning: if managers can effectively oversee more people, you need fewer layers.

This is controversial. Some implementations are clumsy. Done well, it produces flatter, faster orgs.

Senior leadership gets more important.

In flatter orgs with higher individual leverage, senior leadership's strategic direction matters more. Cumulative impact of leadership decisions scales with each enabled individual.

What this means for functions

Engineering: Less change than other functions. Engineering already had high leverage per person. AI tools augment, but org structure doesn't shift as dramatically.

Sales: Agents absorb SDR-level work (lead qualification, research). Fewer SDRs needed. More AE-level work (relationships, deals). Possibly shift in compensation toward senior reps.

Marketing: Agents absorb content drafting, social, campaign analysis. Headcount needed shifts toward strategic + creative + brand.

Support: Agents handle tier-1 work. Fewer tier-1 needed. More tier-2/tier-3 with deeper product knowledge.

Operations: This is the most affected function. Agent ops becomes critical. Traditional ops work absorbed.

HR: Less time on operational HR. More time on talent strategy + org design + leadership development.

Finance: Agents absorb invoice processing, reconciliation, reporting. Finance focuses on strategic analysis + capital allocation.

Different functions reshape differently. The pattern: less middle-of-the-pyramid operational work, more senior strategic work.

What this means for company size

Smaller companies can do more.

A 50-person company in 2026 with strong agent leverage operates like an 80-person company from 2022. Lower coordination overhead. Higher individual leverage. Same output potential.

This is the "small but mighty" company pattern that gets more achievable.

Mid-size companies feel the squeeze.

Companies at 200-500 employees often have built infrastructure for operational scale that's now over-built. Layoff cycles in 2024-2025 partially reflected this realization (though many were misframed as pure cost cuts).

The mid-size companies that figure out the right shape will benefit. The ones that don't will continue right-sizing.

Large companies have governance challenges.

500+ person companies face the agent governance complexity we covered in [post 228]. Coordinating agent deployment across many teams + many functions becomes a real organizational challenge.

Larger companies that build agent ops infrastructure well will compete. Ones that try to scale agents without infrastructure will struggle.

What this means for individuals

If you're an individual thinking about career:

Develop skills that complement agents.

→ Judgment (better with experience) → Relationship-building (harder to automate) → Creative work (genuinely novel thinking) → Cross-functional integration → Strategic synthesis

These compound in value as agents absorb adjacent operational work.

Develop fluency with agent tooling.

The people who can effectively deploy agents for their own workflow become disproportionately valuable. Configure, debug, maintain, improve.

Be wary of roles that are primarily operational.

If your role is largely "process work that's well-defined and repetitive," agents will eventually absorb significant portions. Plan for this. Either lean into judgment/relationship/strategic work, or develop agent ops skills.

Reskill toward agent-resilient work.

Hard problems. Novel situations. Customer relationships. Strategic decisions. Creative output. These don't go away.

What we'd tell leaders

If you're a CEO/COO thinking about how agents change your org:

→ Start with an audit of operational work. What % of each function's time is operational vs. strategic?

→ Identify automation opportunities by function. Which work fits agent absorption? Which doesn't?

→ Pilot small. Don't restructure entire functions. Deploy agents, measure capacity gain, then think about structure.

→ Think about career pipelines. If junior operational roles diminish, where do senior people come from?

→ Invest in agent ops capability. Either as part-time role initially or full-time as you grow.

→ Update hiring criteria. Less specialist operational, more generalist strategic.

→ Communicate transparently with the team. Don't hide the fact that you're reorganizing around agents. Honest communication builds trust.

What we wouldn't recommend

Don't lay off operational staff and announce it's because of AI. Bad for culture, bad for brand, often premature.

Don't restructure before validating that agents actually deliver in your context. Pilot first.

Don't assume agent leverage is uniform across functions. Some functions benefit more than others. Engineering changes less than ops, for example.

Don't ignore the human side of transition. People whose roles are changing need support, training, communication.

The big picture

Org design has been evolving slowly for 50 years. The agent era accelerates the evolution.

Specifically: → Smaller, higher-leverage teams → Flatter org structures → More senior + strategic + creative → Less middle-of-pyramid operational → New "agent ops" function → Cross-functional fluency more important → Generalist fluency over narrow specialization

Companies that adapt org design thoughtfully to the agent era will operate at structural advantage over companies that bolt agents on top of existing structures.

The thoughtful adaptation isn't dramatic. It's a series of small, deliberate choices that compound over years.

→ avery.software — Free Desktop tier. The agent platform that enables the org design transformations companies are figuring out.