Why 57% of Enterprises Waste AI Budgets—And How to Fix It

57% of enterprises still can't outpace AI spend with ROI—unchanged for 2 years. New data reveals the last-mile gap separating losers from winners.

By Rajesh Beri·July 25, 2026·8 min read
Share:
THE DAILY BRIEF
Enterprise AIAI ROIAgentic AIAI GovernanceAI Strategy
Why 57% of Enterprises Waste AI Budgets—And How to Fix It

57% of enterprises still can't outpace AI spend with ROI—unchanged for 2 years. New data reveals the last-mile gap separating losers from winners.

By Rajesh Beri·July 25, 2026·8 min read

Here's the number that should be on every CIO's dashboard right now: 57%. That's the share of enterprises whose AI return on investment still isn't outpacing their spending—identical to last year, unchanged after two full years of record AI investment. This isn't a temporary trough. It's a plateau that 639 senior enterprise AI leaders just confirmed in Domino Data Lab's Fifth Annual Enterprise AI Report, released July 21, 2026.

The uncomfortable part? These same organizations report that 93% have improved their ability to move AI from experimentation to production—up from 88% in 2025. They're building better. They're deploying more. And they're still not making money on it.

If you're a CTO wondering why your AI stack keeps expanding while the CFO keeps asking for ROI proof, or a CFO wondering why AI line items keep growing without a matching revenue impact, this report explains why—and more importantly, what the enterprises actually succeeding are doing differently.

The Production-to-Profit Disconnect

The enterprise AI playbook for the last three years looked something like this: get models into production, show the board impressive deployment metrics, and let the ROI follow. That playbook is broken.

The Domino study identifies what researchers are calling the last-mile gap: the space between a model running in production and a business user actually deriving value from it. Getting a model deployed and getting a business user to act on what that model found are two entirely different problems—and most enterprises have only solved the first one.

Look at how business users actually access AI-generated insights today:

  • 34% of organizations report a mix of AI access methods that varies by business unit—no consistent delivery layer, no standard experience
  • 40% still rely on at least one fully mediated access method: a scheduled report from a data science team, or a request submitted to an analyst who runs the model and returns results manually

Think about what that second number means in practice. A VP of Sales wants to know which accounts are most likely to churn. The AI model that could answer that question in seconds is running in production. But to get the answer, she has to submit a ticket to the data science team, wait two to five business days, and receive a static PDF. By the time she acts on it, three of those accounts have already churned.

The model worked. The ROI didn't happen.

Why agentic AI Makes This Worse Before It Gets Better

Here's where the stakes escalate. The top organizational priority for enterprise AI leaders in 2026 is expanding agentic AI—tied at 38.5% with upskilling business users, and ahead of every other investment category.

But agentic AI is accelerating the last-mile gap problem, not solving it. When an AI agent takes autonomous actions—updating records, sending communications, triggering workflows—the distance between the model decision and the business outcome collapses to near-zero. There's no human mediating the loop. That's the upside.

The downside is governance. And the governance picture is alarming:

  • 43% of organizations have agentic AI running in governed production
  • 41% are piloting (12%) or scaling (29%) agentic AI without governance in place

Let that second number sink in. Nearly half of enterprises running agentic AI at scale today have no governance framework managing it. Those actively scaling without governance outnumber those merely piloting by more than two to one.

In conversations with operations leaders across regulated industries, I keep hearing the same thing: "We know we need governance, we just haven't had time to build it yet." That answer is acceptable when you're running a proof of concept. It's not acceptable when your agents are autonomously making pricing decisions, approving credit applications, or routing customer escalations.

The Governance Advantage Is Real and Measurable

Here's what makes this data more than a cautionary tale: the study identifies exactly what separates the enterprises generating ROI from the 57% that aren't. It's governance—not a bigger model, not a better vendor, not more data.

The numbers are stark:

  • Among organizations whose governance is fully keeping pace with AI activity: 67.5% have agentic AI running in governed production
  • Among organizations where governance is only partially keeping pace: that figure drops to 17.2%
  • Organizations with fully integrated AI governance are 3.9 times more likely to have reached governed agentic deployment

And the velocity impact is equally clear:

  • 75% of organizations with fully integrated governance report significantly improved AI delivery velocity
  • Compare that to 23% among organizations where governance is falling behind

The enterprises with the strictest governance policies aren't moving slower—they're moving three times faster and generating better results. Financial services, banking, and insurance organizations lead every vertical in both governance maturity and production velocity, despite being among the most heavily regulated. They built governance infrastructure first. Then they scaled.

This inverts the conventional wisdom that governance slows AI delivery. The data says the opposite: governance is the accelerant.

The Regional Split Worth Watching

The ROI plateau is global, but the severity varies significantly by geography—and the patterns reveal something important about where the implementation gaps are deepest.

On ROI failure:

  • 51.1% of North American organizations report ROI growing at the same pace as investment or slower
  • 66.9% in the UK
  • 67.0% in Europe

North America is faring better on ROI, but worse on something else: 12.8% of North American organizations report that business users have no direct access to AI-generated insights at all, compared to 1.4% in the UK and 6.4% in Europe. North American enterprises are deploying more but delivering less to the people who actually need it.

Europe has the most acute governance gap for agentic AI. European organizations report the lowest rate of fully integrated governance at 42.6%, compared to roughly 51% in both North America and the UK. Nearly half of European organizations are piloting or scaling agentic AI without governance—a compliance exposure that will become a regulatory liability as the EU AI Act enforcement matures.

What the CFO Needs to Hear

If you're presenting AI budget justification in the second half of 2026, the Domino data gives you a clear narrative:

The issue isn't your AI models. In 93% of cases, the production infrastructure is working. The issue is the delivery layer—the last mile between the model output and the business decision it should be informing.

The ROI math changes completely when you close this gap. Consider the difference between:

  • A customer success team that receives a weekly static report on churn risk (current state for 40% of enterprises)
  • A customer success team with a purpose-built application that shows real-time churn probability scores, recommended interventions, and tracks whether interventions worked

The underlying model is identical. The business impact is not.

Build for the business user, not the data scientist. The enterprises generating ROI have recognized that AI infrastructure is table stakes. The competitive advantage is in the application layer—purpose-built interfaces that let non-technical users act on model outputs without submitting tickets or waiting for analyst handoffs.

What the CTO/CIO Needs to Do Next

The technical implications of this data are straightforward, even if the execution isn't:

1. Audit your last-mile gap immediately. Map every production AI model to its business delivery mechanism. How does a VP of Marketing access campaign performance predictions? How does a logistics manager get supply chain disruption alerts? If the answer is "they submit a request to our data team," you've found your ROI leak.

2. Treat governance as a velocity tool, not a speed bump. The 3.9x deployment advantage for fully governed organizations isn't theoretical—it's measured across 639 organizations in regulated industries. Every week you delay governance is a week you're building technical debt that will slow you down later.

3. Prioritize agentic governance before agentic scaling. The 41% of enterprises scaling agentic AI without governance are not ahead of the curve—they're accumulating risk. The top risks cited by respondents: 44% cite regulatory sanctions from ungoverned AI outputs, 42% cite revenue loss from competitors moving faster with governed AI products.

Note that second risk. Your competitors who govern first are going to move faster, not slower. The organizations running ungoverned agentic AI today are setting up regulatory exposure that will force a governance retrofit later—which costs far more in time and money than building governance correctly from the start.

4. Measure business user access as a primary KPI. Right now, most AI programs measure model performance, deployment velocity, and cost per token. Add a fourth metric: what percentage of your business users can access AI-generated insights directly, without intermediaries? Track it quarterly. If it's not improving, neither is your ROI.

The Bottom Line

Two years into the AI ROI plateau, the data is clear enough to stop treating this as a measurement problem. The enterprises generating returns aren't using better models—they're doing two things differently: closing the last-mile gap with purpose-built business applications, and building governance infrastructure before scaling autonomous agents.

The 57% stuck in the plateau aren't failing at AI. They're succeeding at the wrong problem. Getting models into production was the right problem in 2023. Getting business users to act on model outputs—safely, at scale, with governance built in—is the right problem in 2026.

The enterprises winning right now started solving that second problem 18 months ago. The question for every CIO and CTO reading this is: when do you start?


Source: Fifth Annual Domino Enterprise AI Report, Domino Data Lab / BARC Research, July 2026 (n=639 senior enterprise AI leaders, Director+, organizations with $100M+ revenue, North America / UK / Europe)


Rajesh Beri writes THE D*AI*LY BRIEF — enterprise AI insights for technical and business leaders. Follow on LinkedIn | Follow on X

Continue Reading

THE DAILY BRIEF

Enterprise AI insights for technology and business leaders, twice weekly.

beri.net

Subscribe at beri.net/subscribe for twice-weekly AI insights delivered to your inbox.

LinkedIn: linkedin.com/in/rberi  |  X: x.com/rajeshberi

© 2026 Rajesh Beri. All rights reserved.

Why 57% of Enterprises Waste AI Budgets—And How to Fix It

Photo by fauxels on Pexels

Here's the number that should be on every CIO's dashboard right now: 57%. That's the share of enterprises whose AI return on investment still isn't outpacing their spending—identical to last year, unchanged after two full years of record AI investment. This isn't a temporary trough. It's a plateau that 639 senior enterprise AI leaders just confirmed in Domino Data Lab's Fifth Annual Enterprise AI Report, released July 21, 2026.

The uncomfortable part? These same organizations report that 93% have improved their ability to move AI from experimentation to production—up from 88% in 2025. They're building better. They're deploying more. And they're still not making money on it.

If you're a CTO wondering why your AI stack keeps expanding while the CFO keeps asking for ROI proof, or a CFO wondering why AI line items keep growing without a matching revenue impact, this report explains why—and more importantly, what the enterprises actually succeeding are doing differently.

The Production-to-Profit Disconnect

The enterprise AI playbook for the last three years looked something like this: get models into production, show the board impressive deployment metrics, and let the ROI follow. That playbook is broken.

The Domino study identifies what researchers are calling the last-mile gap: the space between a model running in production and a business user actually deriving value from it. Getting a model deployed and getting a business user to act on what that model found are two entirely different problems—and most enterprises have only solved the first one.

Look at how business users actually access AI-generated insights today:

  • 34% of organizations report a mix of AI access methods that varies by business unit—no consistent delivery layer, no standard experience
  • 40% still rely on at least one fully mediated access method: a scheduled report from a data science team, or a request submitted to an analyst who runs the model and returns results manually

Think about what that second number means in practice. A VP of Sales wants to know which accounts are most likely to churn. The AI model that could answer that question in seconds is running in production. But to get the answer, she has to submit a ticket to the data science team, wait two to five business days, and receive a static PDF. By the time she acts on it, three of those accounts have already churned.

The model worked. The ROI didn't happen.

Why agentic AI Makes This Worse Before It Gets Better

Here's where the stakes escalate. The top organizational priority for enterprise AI leaders in 2026 is expanding agentic AI—tied at 38.5% with upskilling business users, and ahead of every other investment category.

But agentic AI is accelerating the last-mile gap problem, not solving it. When an AI agent takes autonomous actions—updating records, sending communications, triggering workflows—the distance between the model decision and the business outcome collapses to near-zero. There's no human mediating the loop. That's the upside.

The downside is governance. And the governance picture is alarming:

  • 43% of organizations have agentic AI running in governed production
  • 41% are piloting (12%) or scaling (29%) agentic AI without governance in place

Let that second number sink in. Nearly half of enterprises running agentic AI at scale today have no governance framework managing it. Those actively scaling without governance outnumber those merely piloting by more than two to one.

In conversations with operations leaders across regulated industries, I keep hearing the same thing: "We know we need governance, we just haven't had time to build it yet." That answer is acceptable when you're running a proof of concept. It's not acceptable when your agents are autonomously making pricing decisions, approving credit applications, or routing customer escalations.

The Governance Advantage Is Real and Measurable

Here's what makes this data more than a cautionary tale: the study identifies exactly what separates the enterprises generating ROI from the 57% that aren't. It's governance—not a bigger model, not a better vendor, not more data.

The numbers are stark:

  • Among organizations whose governance is fully keeping pace with AI activity: 67.5% have agentic AI running in governed production
  • Among organizations where governance is only partially keeping pace: that figure drops to 17.2%
  • Organizations with fully integrated AI governance are 3.9 times more likely to have reached governed agentic deployment

And the velocity impact is equally clear:

  • 75% of organizations with fully integrated governance report significantly improved AI delivery velocity
  • Compare that to 23% among organizations where governance is falling behind

The enterprises with the strictest governance policies aren't moving slower—they're moving three times faster and generating better results. Financial services, banking, and insurance organizations lead every vertical in both governance maturity and production velocity, despite being among the most heavily regulated. They built governance infrastructure first. Then they scaled.

This inverts the conventional wisdom that governance slows AI delivery. The data says the opposite: governance is the accelerant.

The Regional Split Worth Watching

The ROI plateau is global, but the severity varies significantly by geography—and the patterns reveal something important about where the implementation gaps are deepest.

On ROI failure:

  • 51.1% of North American organizations report ROI growing at the same pace as investment or slower
  • 66.9% in the UK
  • 67.0% in Europe

North America is faring better on ROI, but worse on something else: 12.8% of North American organizations report that business users have no direct access to AI-generated insights at all, compared to 1.4% in the UK and 6.4% in Europe. North American enterprises are deploying more but delivering less to the people who actually need it.

Europe has the most acute governance gap for agentic AI. European organizations report the lowest rate of fully integrated governance at 42.6%, compared to roughly 51% in both North America and the UK. Nearly half of European organizations are piloting or scaling agentic AI without governance—a compliance exposure that will become a regulatory liability as the EU AI Act enforcement matures.

What the CFO Needs to Hear

If you're presenting AI budget justification in the second half of 2026, the Domino data gives you a clear narrative:

The issue isn't your AI models. In 93% of cases, the production infrastructure is working. The issue is the delivery layer—the last mile between the model output and the business decision it should be informing.

The ROI math changes completely when you close this gap. Consider the difference between:

  • A customer success team that receives a weekly static report on churn risk (current state for 40% of enterprises)
  • A customer success team with a purpose-built application that shows real-time churn probability scores, recommended interventions, and tracks whether interventions worked

The underlying model is identical. The business impact is not.

Build for the business user, not the data scientist. The enterprises generating ROI have recognized that AI infrastructure is table stakes. The competitive advantage is in the application layer—purpose-built interfaces that let non-technical users act on model outputs without submitting tickets or waiting for analyst handoffs.

What the CTO/CIO Needs to Do Next

The technical implications of this data are straightforward, even if the execution isn't:

1. Audit your last-mile gap immediately. Map every production AI model to its business delivery mechanism. How does a VP of Marketing access campaign performance predictions? How does a logistics manager get supply chain disruption alerts? If the answer is "they submit a request to our data team," you've found your ROI leak.

2. Treat governance as a velocity tool, not a speed bump. The 3.9x deployment advantage for fully governed organizations isn't theoretical—it's measured across 639 organizations in regulated industries. Every week you delay governance is a week you're building technical debt that will slow you down later.

3. Prioritize agentic governance before agentic scaling. The 41% of enterprises scaling agentic AI without governance are not ahead of the curve—they're accumulating risk. The top risks cited by respondents: 44% cite regulatory sanctions from ungoverned AI outputs, 42% cite revenue loss from competitors moving faster with governed AI products.

Note that second risk. Your competitors who govern first are going to move faster, not slower. The organizations running ungoverned agentic AI today are setting up regulatory exposure that will force a governance retrofit later—which costs far more in time and money than building governance correctly from the start.

4. Measure business user access as a primary KPI. Right now, most AI programs measure model performance, deployment velocity, and cost per token. Add a fourth metric: what percentage of your business users can access AI-generated insights directly, without intermediaries? Track it quarterly. If it's not improving, neither is your ROI.

The Bottom Line

Two years into the AI ROI plateau, the data is clear enough to stop treating this as a measurement problem. The enterprises generating returns aren't using better models—they're doing two things differently: closing the last-mile gap with purpose-built business applications, and building governance infrastructure before scaling autonomous agents.

The 57% stuck in the plateau aren't failing at AI. They're succeeding at the wrong problem. Getting models into production was the right problem in 2023. Getting business users to act on model outputs—safely, at scale, with governance built in—is the right problem in 2026.

The enterprises winning right now started solving that second problem 18 months ago. The question for every CIO and CTO reading this is: when do you start?


Source: Fifth Annual Domino Enterprise AI Report, Domino Data Lab / BARC Research, July 2026 (n=639 senior enterprise AI leaders, Director+, organizations with $100M+ revenue, North America / UK / Europe)


Rajesh Beri writes THE D*AI*LY BRIEF — enterprise AI insights for technical and business leaders. Follow on LinkedIn | Follow on X

Continue Reading

Share:
THE DAILY BRIEF
Enterprise AIAI ROIAgentic AIAI GovernanceAI Strategy
Why 57% of Enterprises Waste AI Budgets—And How to Fix It

57% of enterprises still can't outpace AI spend with ROI—unchanged for 2 years. New data reveals the last-mile gap separating losers from winners.

By Rajesh Beri·July 25, 2026·8 min read

Here's the number that should be on every CIO's dashboard right now: 57%. That's the share of enterprises whose AI return on investment still isn't outpacing their spending—identical to last year, unchanged after two full years of record AI investment. This isn't a temporary trough. It's a plateau that 639 senior enterprise AI leaders just confirmed in Domino Data Lab's Fifth Annual Enterprise AI Report, released July 21, 2026.

The uncomfortable part? These same organizations report that 93% have improved their ability to move AI from experimentation to production—up from 88% in 2025. They're building better. They're deploying more. And they're still not making money on it.

If you're a CTO wondering why your AI stack keeps expanding while the CFO keeps asking for ROI proof, or a CFO wondering why AI line items keep growing without a matching revenue impact, this report explains why—and more importantly, what the enterprises actually succeeding are doing differently.

The Production-to-Profit Disconnect

The enterprise AI playbook for the last three years looked something like this: get models into production, show the board impressive deployment metrics, and let the ROI follow. That playbook is broken.

The Domino study identifies what researchers are calling the last-mile gap: the space between a model running in production and a business user actually deriving value from it. Getting a model deployed and getting a business user to act on what that model found are two entirely different problems—and most enterprises have only solved the first one.

Look at how business users actually access AI-generated insights today:

  • 34% of organizations report a mix of AI access methods that varies by business unit—no consistent delivery layer, no standard experience
  • 40% still rely on at least one fully mediated access method: a scheduled report from a data science team, or a request submitted to an analyst who runs the model and returns results manually

Think about what that second number means in practice. A VP of Sales wants to know which accounts are most likely to churn. The AI model that could answer that question in seconds is running in production. But to get the answer, she has to submit a ticket to the data science team, wait two to five business days, and receive a static PDF. By the time she acts on it, three of those accounts have already churned.

The model worked. The ROI didn't happen.

Why agentic AI Makes This Worse Before It Gets Better

Here's where the stakes escalate. The top organizational priority for enterprise AI leaders in 2026 is expanding agentic AI—tied at 38.5% with upskilling business users, and ahead of every other investment category.

But agentic AI is accelerating the last-mile gap problem, not solving it. When an AI agent takes autonomous actions—updating records, sending communications, triggering workflows—the distance between the model decision and the business outcome collapses to near-zero. There's no human mediating the loop. That's the upside.

The downside is governance. And the governance picture is alarming:

  • 43% of organizations have agentic AI running in governed production
  • 41% are piloting (12%) or scaling (29%) agentic AI without governance in place

Let that second number sink in. Nearly half of enterprises running agentic AI at scale today have no governance framework managing it. Those actively scaling without governance outnumber those merely piloting by more than two to one.

In conversations with operations leaders across regulated industries, I keep hearing the same thing: "We know we need governance, we just haven't had time to build it yet." That answer is acceptable when you're running a proof of concept. It's not acceptable when your agents are autonomously making pricing decisions, approving credit applications, or routing customer escalations.

The Governance Advantage Is Real and Measurable

Here's what makes this data more than a cautionary tale: the study identifies exactly what separates the enterprises generating ROI from the 57% that aren't. It's governance—not a bigger model, not a better vendor, not more data.

The numbers are stark:

  • Among organizations whose governance is fully keeping pace with AI activity: 67.5% have agentic AI running in governed production
  • Among organizations where governance is only partially keeping pace: that figure drops to 17.2%
  • Organizations with fully integrated AI governance are 3.9 times more likely to have reached governed agentic deployment

And the velocity impact is equally clear:

  • 75% of organizations with fully integrated governance report significantly improved AI delivery velocity
  • Compare that to 23% among organizations where governance is falling behind

The enterprises with the strictest governance policies aren't moving slower—they're moving three times faster and generating better results. Financial services, banking, and insurance organizations lead every vertical in both governance maturity and production velocity, despite being among the most heavily regulated. They built governance infrastructure first. Then they scaled.

This inverts the conventional wisdom that governance slows AI delivery. The data says the opposite: governance is the accelerant.

The Regional Split Worth Watching

The ROI plateau is global, but the severity varies significantly by geography—and the patterns reveal something important about where the implementation gaps are deepest.

On ROI failure:

  • 51.1% of North American organizations report ROI growing at the same pace as investment or slower
  • 66.9% in the UK
  • 67.0% in Europe

North America is faring better on ROI, but worse on something else: 12.8% of North American organizations report that business users have no direct access to AI-generated insights at all, compared to 1.4% in the UK and 6.4% in Europe. North American enterprises are deploying more but delivering less to the people who actually need it.

Europe has the most acute governance gap for agentic AI. European organizations report the lowest rate of fully integrated governance at 42.6%, compared to roughly 51% in both North America and the UK. Nearly half of European organizations are piloting or scaling agentic AI without governance—a compliance exposure that will become a regulatory liability as the EU AI Act enforcement matures.

What the CFO Needs to Hear

If you're presenting AI budget justification in the second half of 2026, the Domino data gives you a clear narrative:

The issue isn't your AI models. In 93% of cases, the production infrastructure is working. The issue is the delivery layer—the last mile between the model output and the business decision it should be informing.

The ROI math changes completely when you close this gap. Consider the difference between:

  • A customer success team that receives a weekly static report on churn risk (current state for 40% of enterprises)
  • A customer success team with a purpose-built application that shows real-time churn probability scores, recommended interventions, and tracks whether interventions worked

The underlying model is identical. The business impact is not.

Build for the business user, not the data scientist. The enterprises generating ROI have recognized that AI infrastructure is table stakes. The competitive advantage is in the application layer—purpose-built interfaces that let non-technical users act on model outputs without submitting tickets or waiting for analyst handoffs.

What the CTO/CIO Needs to Do Next

The technical implications of this data are straightforward, even if the execution isn't:

1. Audit your last-mile gap immediately. Map every production AI model to its business delivery mechanism. How does a VP of Marketing access campaign performance predictions? How does a logistics manager get supply chain disruption alerts? If the answer is "they submit a request to our data team," you've found your ROI leak.

2. Treat governance as a velocity tool, not a speed bump. The 3.9x deployment advantage for fully governed organizations isn't theoretical—it's measured across 639 organizations in regulated industries. Every week you delay governance is a week you're building technical debt that will slow you down later.

3. Prioritize agentic governance before agentic scaling. The 41% of enterprises scaling agentic AI without governance are not ahead of the curve—they're accumulating risk. The top risks cited by respondents: 44% cite regulatory sanctions from ungoverned AI outputs, 42% cite revenue loss from competitors moving faster with governed AI products.

Note that second risk. Your competitors who govern first are going to move faster, not slower. The organizations running ungoverned agentic AI today are setting up regulatory exposure that will force a governance retrofit later—which costs far more in time and money than building governance correctly from the start.

4. Measure business user access as a primary KPI. Right now, most AI programs measure model performance, deployment velocity, and cost per token. Add a fourth metric: what percentage of your business users can access AI-generated insights directly, without intermediaries? Track it quarterly. If it's not improving, neither is your ROI.

The Bottom Line

Two years into the AI ROI plateau, the data is clear enough to stop treating this as a measurement problem. The enterprises generating returns aren't using better models—they're doing two things differently: closing the last-mile gap with purpose-built business applications, and building governance infrastructure before scaling autonomous agents.

The 57% stuck in the plateau aren't failing at AI. They're succeeding at the wrong problem. Getting models into production was the right problem in 2023. Getting business users to act on model outputs—safely, at scale, with governance built in—is the right problem in 2026.

The enterprises winning right now started solving that second problem 18 months ago. The question for every CIO and CTO reading this is: when do you start?


Source: Fifth Annual Domino Enterprise AI Report, Domino Data Lab / BARC Research, July 2026 (n=639 senior enterprise AI leaders, Director+, organizations with $100M+ revenue, North America / UK / Europe)


Rajesh Beri writes THE D*AI*LY BRIEF — enterprise AI insights for technical and business leaders. Follow on LinkedIn | Follow on X

Continue Reading

THE DAILY BRIEF

Enterprise AI insights for technology and business leaders, twice weekly.

beri.net

Subscribe at beri.net/subscribe for twice-weekly AI insights delivered to your inbox.

LinkedIn: linkedin.com/in/rberi  |  X: x.com/rajeshberi

© 2026 Rajesh Beri. All rights reserved.

Frequently Asked Questions

Why do 57% of enterprises fail to get ROI from their AI investments?

According to Domino Data Lab's Fifth Annual Enterprise AI Report (639 senior AI leaders, July 2026), the problem isn't model deployment — 93% of organizations have improved at moving AI into production. The failure is the 'last-mile gap': business users can't directly act on model outputs. 40% of organizations still deliver AI insights through mediated channels like scheduled reports or analyst requests, so the model runs but the business decision it should inform never happens.

Does AI governance slow down enterprise AI delivery?

The data says the opposite. Organizations with fully integrated AI governance are 3.9 times more likely to have agentic AI in governed production (67.5% vs 17.2%), and 75% of them report significantly improved AI delivery velocity versus 23% where governance lags. Heavily regulated verticals like financial services and insurance lead in both governance maturity and production velocity — governance acts as an accelerant, not a speed bump.

What is the 'last-mile gap' in enterprise AI?

It's the space between a model running in production and a business user actually deriving value from it. A churn model may work perfectly, but if a sales VP has to file a ticket and wait days for a static report, the ROI never materializes. Closing the gap means purpose-built applications that give non-technical users direct, real-time access to model outputs — and measuring that access as a primary KPI.

How many enterprises run agentic AI without governance?

41% of surveyed organizations are piloting (12%) or scaling (29%) agentic AI with no governance framework in place, versus 43% with agentic AI in governed production. Respondents rank the risks clearly: 44% cite regulatory sanctions from ungoverned AI outputs and 42% cite revenue loss to competitors who move faster with governed AI products.

Newsletter

Stay Ahead of the Curve

Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.

Subscribe