AE Studio · Executive Webinar
Agents in the Org Chart.
What leaders need to change as AI reshapes how work gets done.
Melanie PlazaChief Technology Officer
·
Husam MachloviHead of Product
AE Studio · an AI Strategy & Implementation Studio
The shift
Your org chart was built for a world without AI.
Every week, smart people gather updates, rewrite them, reformat them, and route them upward. AI changes the cost of that work. Which means it changes what your org chart is optimized for.
The gap
AI is in the building.
Value isn't.
78%
of organizations report using AI in 2024.
McKinsey Global Survey on AI
5%
are capturing AI value at scale.
BCG
1%
of leaders describe their organization as mature in AI deployment.
McKinsey
The limiting factor isn't access to tools. It's the operating model.
Where the leading edge is investing
The innovators are reshaping the org around AI.
Block · Jack Dorsey · Mar 2026
"From hierarchy to intelligence."
Three roles: DRIs who own direction, player-coaches who develop craft and people, and builders. No permanent middle-management layer whose main purpose is information routing.
Published essay · block.xyz/inside
Microsoft · 2025 Work Trend Index
"Intelligence on tap will rewire business. Every leader needs a new blueprint."
The rise of the "Frontier Firm." Lean teams assembled around goals, agents as digital colleagues.
Annual report · 31K surveyed + M365 telemetry
Microsoft · 2025 Work Trend Index
"Every employee becomes an agent boss."
The new baseline role: every person manages a team of agents for the work they own, not just uses a tool.
Annual report · 31K surveyed + M365 telemetry
Shopify · Tobi Lütke · Apr 2025
"Show why AI can't do it, before asking for headcount."
AI usage is now a baseline expectation. Every team proposal has to show the AI version first.
Public internal memo
Meta · Reality Labs · Mar 2026
AI Builder. AI Pod Lead. AI Org Lead.
New role taxonomy. Small, AI-native pods. Fewer layers between builders and leadership.
Business Insider reporting
About AE Studio.
We're AE Studio, an AI strategy and implementation studio. For ten years we've worked with hundreds of companies, helping leaders put new technology to work. For the last few, that's meant two things: helping companies adopt AI, and running frontier AI alignment research.
A few of the teams we've built with
Why it's happening everywhere at once
AI lowers the cost of coordination and intelligence.
Coordination
Work that used to consume the middle of your company.
gathersummarizecompareroutedecide
These five verbs are what most managers and program owners do all day. They're exactly what language models are best at.
Intelligence
Work that used to be cognitively expensive gets cheap.
analyzesynthesizeinvestigatereasonevaluate
A decent first-pass analysis, research synthesis, or reasoned tradeoff used to require an expensive senior. Now it's a prompt away.
Management doesn't disappear. It splits. The routing layer evaporates; judgment and taste get more important.
The question most leaders dodge
When output goes up, what do you buy with it?
Say AI makes your team 30% faster. That's a strategic choice, not a spreadsheet exercise. Most companies default to option 01 because it's the only one easy to count.
01
Same output, fewer people.
Pure efficiency. Same value, cheaper. Sometimes the right call, and the one that makes your best people update their LinkedIn.
02
Get more from the same team.
Higher quality. Ship the backlog of nice-to-haves. Reshape roles around where people add the most value. Engineers pick up product work, the product engineer pattern.
03
Do things you couldn't before.
Assess every user individually. Rebuild a content library. Respond to every customer personally. New categories of product, not same products cheaper.
Efficiency-only is the boring answer. The interesting answer is new value.
Where you are
Seven levels of AI leverage.
Most companies are here ↓
Stuck in Phase 1. The leverage starts at Level 3.
Stuck in Phase 1. The leverage starts at Level 3.
Phase 2, where execution changes ↑
Not you with AI. Your workflow with AI.
Not you with AI. Your workflow with AI.
Phase 3, structural leverage ↑
The whole org gets redesigned. Dorsey is here.
The whole org gets redesigned. Dorsey is here.
ArchitectPhase 3 · Your org + AI · Structural leverage
7
Org redesign around AI-native workflows
Roles and reporting lines reflect where humans actually add value. AI is the default path.
strategistAI operates
6
Multi-agent teams
Humans manage teams of agents. The work is coordination, QA, and exception handling.
managerAI teams
Redesign the process, not just the tooling
OperatorPhase 2 · Your workflow + AI · Process leverage
5
No human in the loop (for qualified workflows)
The agent runs end to end. You monitor dashboards, not drafts.
monitorAI autonomous
4
Connected agents with tool access
Data pipes in, output routes out. Review at checkpoints, not every step.
reviewerAI connected
3
Single-agent task execution
One agent, one task. You provide input, review output, iterate. You're still driving.
driverAI executes
DoerPhase 1 · You + AI · Individual leverage
2
Do what you'd otherwise skip
Deeper research, call prep, pressure-testing. AI changes your outcomes, not just your speed.
doerAI augments
1
Offload the busywork
Summarize, reformat, synthesize. Fastest win, lowest bar.
doerAI assists
What changes for your people
Every role moves up the stack.
| Role | Before AI | With AI | More valuable now |
|---|---|---|---|
| Product | PRDs, requirements, handoffs | AI drafts specs, synthesizes feedback | prioritization, judgment, customer insight |
| Engineering | manual code, tests, docs | AI-assisted build, test, docs | architecture, review, verification |
| Ops / Program | status, reporting, chasing updates | AI automates reporting and follow-up | exception handling, governance |
| Sales / Marketing | research, proposals, assets | AI scales drafts and analysis | positioning, narrative, relationships |
| Support | routine tickets, FAQs | AI resolves repetitive issues | trust repair, complex cases, insight loops |
| Managers | route information, enforce process | AI improves visibility and routing | coaching, accountability, system design |
So what do you actually do?
Start with a workflow.
Then redesign a team.
Then rethink the org.
Most companies reverse this. They try to "set an AI strategy" before they've redesigned a single workflow. Don't. Start small. Ship something visible. Then scale the change.
Step 1
Find one workflow worth redesigning.
FrequentHappens every week.
PainfulPeople complain about it.
MeasurableYou can count hours or dollars.
Low-riskMistakes won't blow up customers.
Clear ownerOne person has the pen.
Repetition-heavyLots of handoffs and rework.
Look at your own calendar
What do you do every week that's repetitive? Where do you read, summarize, rewrite, chase, or report?
Ask your team what they already use AI for
Shadow adoption is already in your org. Some of the best redesign ideas are happening quietly. Learn from them.
Step 2
Do the workflow math.
Before you talk about "AI strategy," count one workflow. Weekly operating-review prep, the updates, decks, and status docs produced every Monday.
| Step | Who | Time / run | Frequency | AI help? |
|---|---|---|---|---|
| Gather updates from teams | 8 managers | 30 min | weekly | yes |
| Reformat & summarize | 8 managers | 45 min | weekly | yes |
| Exec review & decide | 1 exec | 20 min | weekly | partly |
620 min / week · About 540 hours / year · before counting slower decisions, delayed meetings, or opportunity cost.
Step 3
Hand off the work. Keep the judgment.
Agents don't just draft anymore. They research, execute, run entire workflows, and manage other agents. The real question isn't what AI can do. It's which parts of the work you still want to own, and being honest about which mode you're in.
Hand it off when…
- you know what good output looks like
- the work repeats or is well-scoped
- you can check the result without redoing it
- mistakes are cheap to catch and fix
Stay in when…
- the first result changes what you ask next
- the problem is a category you haven't seen
- judgment is the product, not the artifact
- trust or a real decision is on the line
Delegate the doing. Own the decisions.
What's actually in the way
The real blockers aren't technical.
Every room says the same four things. You can't build adoption without naming them, out loud, from the top.
Fear 01 · Identity
"It's going to take my job."
Reframe. AI gives you superpowers. The repetitive work shrinks, the judgment work grows, and you deliver way more value than you could before. The question isn't whether the job changes. It's who does the changing.
Fear 02 · Security
"What about data and privacy?"
Reframe. The same governance that already covers your SaaS stack covers AI. Scope, permissions, audit logs, DLP. The boring answer is the real answer, and it's your job to make it visible.
Fear 03 · Trust
"It hallucinates. I can't rely on it."
Reframe. Of course it does. Structure the work so humans judge the output, not whether the output exists. That's Level 3 and up. The model's failure mode becomes a design constraint, not a blocker.
Fear 04 · Relevance
"We're too complex for this to work."
Reframe. Every company that said this about websites, cloud, and mobile is not the one that won the next decade. Complexity is the reason to start, not the reason to wait.
"In our culture, the cost of being wrong is higher than the cost of being slow."
That math used to make sense. It doesn't anymore. AI makes execution cheap, so the cost of standing still compounds fast. If your culture punishes being wrong more than being slow, none of the four reframes above will land. This is the one you work on first.
What has to be true before agents produce gain
Agents multiply what's there. Including the problems.
If the underlying org can't do the thing, agents can't either. They just make the dysfunction faster and more visible. Three prerequisites that have to be in place before multiplication produces gain instead of mess.
01
Legibility.
Does the data exist, and can agents reach it? Availability and access. Documented requirements, a real source of truth. Without this, agents guess. Confidently.
02
Authority.
Can humans close what agents surface? Named owners with real decision rights, and decision windows measured in days, not quarters.
Watch: agents surface decisions faster than most orgs can close them.
03
Culture.
Agents make the org transparent. Decisions, latency, and reasoning all become visible for the first time. That visibility is the real prize.
The leverage: leaders who act on what they see first compound. The ones who don't, plateau.
Invest here first. Everything agents do compounds from this foundation.
How leaders actually move the org through this
Two ways this goes wrong.
Every real transformation holds high standards and high support at the same time. Pick one side, and you get one of these two failures.
Failure 01 · Standards without support
Mandate without tooling.
Duolingo, April 2025 → April 2026. Announced AI usage would be evaluated in performance reviews. Twelve months later, they quietly walked it back. Employees got squeezed before the tools, training, and management support could meet them.
Raise the bar before the org can deliver, and you burn credibility. Standards without support is extraction.
Failure 02 · Support without standards
AI theater.
Usage leaderboards, LinkedIn posts, no outcomes. One company we watched mandated AI use, ran usage leaderboards, and had its CEO posting about prompt counts weekly. The output was slop. No metric moved. A lot of motion, no expectation of outcome, and the best people stopped trying because it was clearly performance.
Measure outcomes, not prompts. Cycle time, cost to serve, quality, decision speed, hours saved. Support without standards is a hobby.
What to do about the fears
Your advocates move the org faster than your policies.
The fastest way to dissolve AI fear isn't a training program. It's Bob from accounting standing up and showing the room the agent he built, and taking live questions about it.
01
Find the builders.
They're already in your Slack or Teams, shipping quiet wins. Self-selected. Curious. Past the fear.
Move: Search messaging for "chatgpt" or "claude." Tap the top five names by volume.
02
Give them air cover.
10% of their time, a named exec sponsor, a dedicated channel, and permission to break small things in public.
Move: Put them on the same calendar as your AI steering group. They're not a side project.
03
Make the work public.
A 60-minute office hour each week. Anyone drops in with a real problem. Advocate drives the screen. Everyone leaves with a workable next step.
Move: Record sessions. Post one short clip a week in #ai. Adoption bends on seeing peers, not policy.
"Fear dissolves when the answer comes from the next cubicle, not from a keynote."
What to actually do
Suggested moves.
Four moves that separate companies who unlock AI's second derivative from companies who just license it.
01
Treat AI as a learning curve, not a light switch.
Meet people where they are. Raise the bar as the tools get better.
02
Remove every constraint between your people and the AI.
No token caps. No access tiers. Pre-connected tools. Kill the IT queue.
03
Give people a stage, not just a mandate.
All-hands demos, shared channels, leaderboards. Make building visible.
04
Expect the tools to churn.
Your best internal tool from three months ago should feel obsolete today.
Proof · Ramp · one year in
Fintech expense-management company, ~1,000 employees. Ran this playbook at full tilt and published the receipts.
99.5%
of employees active on AI tools
1,500+
internal apps in six weeks, from 800+ non-engineer builders
12%
of production PRs from non-engineers
And one we're in right now
From AI audit to agents in production.
We're working with a marketing organization that's running 150+ campaigns a year with more than 40 people on staff, and they want to scale their communications to more segments. We're using AI to help them do this by discovering many hours spent on administrative and routine tasks.
184
Hours / week recoverable
The opportunity concentrates in these agents.
Phase 1 ships the first three into production on their Claude enterprise environment.
JIRA / PM agent
43 hrs/wk recovered
Performance / analytics agent
25 hrs/wk recovered
Campaign brief / strategy agent
12 hrs/wk recovered
Copy drafting agent
14 hrs/wk recovered
QC validation agent
10 hrs/wk recovered
Localization agent
33 hrs/wk recovered
By Monday
Lead the change. Don't chase it.
01
Pick one workflow from your own week.
The one you'd be embarrassed to tell a board it still looks this way.
02
If you're unsure, do the math on it.
People × hours × frequency × loaded cost. Now you have a budget conversation, not a decision-fatigue conversation.
03
Bring it to your team on Monday.
One workflow. One team. And tell them out loud this is a test, it won't be perfect, and that's the point.
The question isn't whether AI will affect your organization. It already has. You don't have to do this alone.
Keep the conversation going
Thank you.
If one workflow on your desk fits the pattern we described today, we'd love to see it. And if you want help strategizing, adjusting your org chart, or implementing any of this, reach out. Send a note, we read everything.
Let's get into it
Four questions for the room.
- Where in your organization are talented people acting like routers, gathering, reformatting, and passing information up and across?
- Is there anyone on your leadership team already leading this change by doing the work themselves? If not, what's one quick win you could take on yourself to show the org you're in it too?
- What human judgment are you unwilling to automate, and why? (e.g., hiring calls, strategic pivots, compensation decisions, customer escalations.)
- If AI made your team 30% faster this quarter, which of the three options, same output fewer people, more from the same team, or things you couldn't do before, would your company actually choose? Which should it choose?