The AI Labor Report
Enterprise AI is already working across your organization. The question is whether anyone is accountable for it. New research from 200 U.S. technology executives reveals the structural gap between what leaders believe about their AI investments and what is actually happening.
92% say they track it.
2% can prove it.
100%
Every organization in this study requires human review after AI generates work. Zero operate autonomously.
The AI your organization approved is probably not the AI doing most of the work.
67% say AI embedded in existing tools does the majority of the work. 53% say it runs through tools used without oversight. Managed AI is already a minority.
The state of enterprise AI in 2026.
Senior executives have quietly turned AI into a de facto workforce without giving themselves any reliable way to see where that work is happening, who owns it, or whether it actually pays off.
The enterprise has an AI workforce it cannot book.
78% view AI as both software and labor, but the P&L has no line for AI labor and the org chart has no box for an agent.
Executives say they track AI value. Their books say otherwise.
92% claim to track AI's financial impact. In practice, 51% record 30% or less of AI work as a business outcome.
Fully autonomous workflows remain fiction.
100% of organizations require human intervention after AI generates work. The predominant model is supervised machine labor, not autonomy.
The fastest-growing AI workforce may be the one no one governs.
53% say shadow apps perform most automated work. 67% say embedded AI does most of the work. Governed AI is already a minority.
The ROI debate is really an evidence crisis.
79% fear AI budgets will be cut. 76% lose ROI opportunities because they lack visibility into AI decisions.
Twelve percent have a clear methodology. Here is what separates them.
They treat AI execution cost as a labor line, allocate it to workflows, and report cost per unit of work.
The enterprise has an AI workforce it cannot book.
The functional organization was built to coordinate humans doing different work at scale. That assumption is now structurally false. 78% of enterprise executives now view AI as both software to license and govern and as a labor force that performs work. Only 17% see it as software only. The perception has shifted. The infrastructure has not.
When an agent handles ten thousand support tickets, it appears on the P&L as software expense. When a human did that work, it was labor. 63% of organizations do not record AI impact in formal financial systems at all. Performance reviews, bonus decisions, and promotion cases are being made on falsely attributed output.
View AI as both software and labor 78%
See AI as software only 17%
Other / unsure 5%
The majority already see AI as labor. Their accounting systems do not.
Always goes to employee
13%
Sometimes goes to employee
74%
Rarely goes to employee
14%
AI performs work
Output has no origin tag
Nothing enters financial records
CFO cannot see AI contribution
Budget faces scrutiny
Executives say they track AI value. Their books say otherwise.
92% of leaders agree their organization tracks the financial and efficiency impact of AI-generated work. In practice, only 2% record more than half of that work as a business outcome. 51% record 30% or less. The tracking is happening. The translation is not.
Less than 10%
1%
10-20%
10%
21-30%
40%
31-40%
39%
41-50%
9%
More than 50%
2%
51% record 30% or less. The most common answer is 21–30% – a fraction of the actual work AI performs.
If AI was involved, it contributed
43%
Educated guesses on correlation
38%
We don't attempt to separate AI's impact
8%
A clear methodology
12%
The 12% with a clear methodology are the benchmark. They are the only ones who can answer their CFO with confidence.
Quietly Cautious
Low confidence, but the records are clean.
The 12% Benchmark Group
Clear methodology, defensible numbers.
Flying Blind
Low confidence, no records.
Confident but Blind
Claims to track, but nothing in the books. Most organizations live here.
Actual recording
Hoow much AI work is actually captured in financial records
Confidence
How confident the organization is that they're tracking AI impact
Most organizations believe they track AI impact. Almost none can show where that impact lives in their financial records.
IT tracks AI spending, usage, and impact
56%
Centralized function estimates AI impacts across the business
52%
Best guess based on investments and estimated AI use
52%
Each department estimates AI impacts separately
44%
Recorded in formal financial systems
37%
We don't calculate this
2%
Only 37% record AI in formal financial systems. Everything else is estimation.
Fully autonomous workflows remain fiction inside large enterprises.
100% of organizations require human intervention after AI generates work. This is not a failure of AI capability. It is the current operating model. The cost of this human intervention — the verification, the editing, the rework — is absorbed invisibly into knowledge worker time with no attribution, no budget line, and no measurement.
100%
of organizations require human review after AI generates work. Zero operate autonomously.
Varies by task 36%
Substantially edited 34%
Quickly reviewed 24%
Reworked from scratch 7%
The predominant enterprise AI model is not autonomy. It is supervised machine labor.
Merging with legacy systems
61%
Accuracy of output
59%
Need for human oversight
56%
Tracking usage
54%
Employee confusion
37%
View AI as both software and labor 78%
See AI as software only 17%
Other / unsure 5%
The shift to viewing AI as labor is near-universal – but books, reviews, and budgets still treat it as software.
The fastest-growing AI workforce may be the one no one governs.
53% of leaders estimate the majority of automated work at their organization runs through unmonitored shadow applications. 67% say AI embedded in existing tools — Salesforce, Office, Adobe — accounts for more than half of all AI output. These are not the same tools. Together they suggest that formally governed AI is already a minority of enterprise AI work.
Each square represents 1% of respondents. More than half work at organizations where shadow applications perform most of the AI work.
Less than 50%
34%
50–75%
55%
More than 75%
12%
Mix of leaders and departments
57%
IT department
35%
Operating executives
8%
Not being done
1%
Accountability is distributed, which means it belongs to no one.
The ROI debate is really an evidence crisis.
74% of leaders say managers at their organizations at least sometimes struggle to show how AI investments affect key business metrics. 90% have no single dedicated function responsible for demonstrating AI's return on investment. AI performance is everyone's problem and therefore often no one's clear mandate.
74%
of managers regularly cannot demonstrate AI's impact on key business metrics
79%
are concerned AI budgets will be cut because they cannot be tied to revenue or profit
Measuring impact of every task
54%
Tracking against core business metrics
49%
Measuring efficiency gains
43%
Measuring indirect benefits
40%
Isolating project-specific impact
40%
Knowing every tool that contains AI
38%
Knowing how often AI is used
34%
Organizations know what they want to measure. They cannot connect AI activity to any of it.
Profitability
61%
OpEx efficiency
59%
Revenue per employee
53%
Cost of goods sold
49%
Net profit margin
45%
Organizations are watching the right numbers. The problem is connecting AI activity to them.
Multiple departments share responsibility
55%
Varies by task or tool
27%
Different departments use different approaches
9%
Dedicated person or function
10%
Organizations are watching the right numbers. The problem is connecting AI activity to them.
When responsibility is shared equally, it belongs to no one.
Twelve percent of organizations have a clear methodology. Here is what separates them.
What separates the 12% is not the sophistication of their models or the size of their AI budget. It is three management decisions.
01
Treat AI cost as a labor line, not an IT expense.
They allocate AI execution cost the way they allocate human labor — to the workflows it serves, not to a central IT budget. This makes AI visible to every P&L owner.
02
Build a defensible attribution methodology.
They establish how causation between AI activity and business results is proven, not assumed. If AI was involved, they can show what it did and what it produced.
03
Record AI contributions in systems of record.
They move AI impact out of decks and informal estimates and into the financial systems executives trust — the same place human labor appears.
Until these three issues are addressed, AI budgets will remain politically vulnerable, performance attribution will remain distorted, and the true unit economics of AI-enabled work will stay partially invisible.
Most organizations can see AI where work is conversational. Few can see it where value compounds.
58% of organizations can identify AI-assisted content in support tickets and customer emails. 57% can track it in internal documents. But only 39% can track it in CRM or financial records. Only 35% can identify AI-generated code in their own repositories. The visibility gap is not random. It follows a pattern: organizations can see AI where output is visible and conversational. They lose sight of it precisely where enterprise value accumulates.
CRM systems
39%
CRM systems
39%
Financial records
39%
Code repositories
35%
Support tickets
58%
Customer emails
58%
Documents
57%
Support tickets
58%
Customer emails
58%
Internal documents
57%
CRM records
39%
Financial records
39%
Code repositories
35%
Blue: where leaders can see AI. Coral: where the gap lives — and where the business value sits.
The measurement failures in this report are not operational problems. They are funding threats.
79% fear AI budgets will be cut. 76% lose ROI opportunities because they lack visibility into AI decisions. The CFO conversation is already happening. Most organizations are not ready for it.
Very concerned 11%
Slightly concerned 68%
Not very concerned 16%
Not at all 5%
$500K–$1M
21%
$1M–$5M
39%
$5M–$10M
32%
$10M–$50M
7%
$50M+
2%
71% of organizations surveyed control between $1M and $10M in annual AI spend. The stakes are large enough to get the CFO's attention. The measurement systems are not there yet.
Often
4%
Sometimes
72%
Rarely
21%
Never
4%
Blue: where leaders can see AI. Coral: where the gap lives — and where the business value sits.
Lack of visibility is not a future risk. It is an active, recurring cost.
About this research.
The AI Labor Report is based on original research conducted among senior technology leaders and executives at large U.S. organizations. All respondents were screened for decision-making authority and company size before participating. The findings represent the operational realities of the people directly responsible for AI outcomes at scale — not general opinions about technology.
Sample size
200
Geography
United States
Company size
1,000+ employees
Field dates
March 20 – April 8, 2026
Confidence level
95%
Margin of error
±6.9 percentage points
Fieldwork conducted by Wakefield Research.
68%
C-level executives (CIO, CAIO, CTO, COO, CFO)
100%
hold final or significant AI decision-making authority
68%
control annual AI budgets of $1 million or more
CIO
27%
CAIO
24%
CTO
21%
COO
15%
CFO
12%
These are not observers. They are the people writing the checks and owning the outcomes.
IT / Software
18%
Healthcare
17%
Banking / Finance
14%
Retail
14%
Manufacturing
12%
Transportation
6%
Automotive
5%
Cross-industry. No single sector distortion.
Sign up for a demo and start managing your AI workforce today.
See how Lanai turns shadow AI into a measurable, accountable labor line — with attribution, oversight, and a system of record your CFO will accept.
The operating system for AI value creation. See it. Steer it. Scale it.