What AI needs

to succeed
1,620 enterprise leaders on processes,
use cases, and ROI

EDITION
AI
· ·
N A

2025
IO

I
ED
IT
ED

IT
ION
AI

Process
·

Optimization
ION

AI E

Report
IT

DI
ED

IO T

I N
A
· ·
AI
EDITION
Contents Processes give AI what it needs to succeed 03

Section one: 04
What enterprises are doing with AI

Ubiquitous AI: GenAI is already everywhere 05

AI ROI expectations: A pivotal year 06

Section two: 09
Departmental perspectives on AI

Process Improvement and Operations: 10

A focus on intelligent automation

Supply Chain: A foothold in the warehouse 11

IT: Security concerns are paramount 12

Finance and Shared Services: A mandate 13

for investment

Section three: 14
There’s no AI without PI

AI success needs process understanding 15

and business context

What Process Intelligence makes possible 21

What’s next? 22

Survey sample 24

2 What AI needs to succeed

Processes We all understand the transformative power of artificial
intelligence (AI). The vast majority of enterprises already
give AI what use it, with 99% of the 1,620 business leaders we surveyed
for our 2025 Process Optimization Report saying their
it needs to organizations deploy some form of AI. What’s more, 73%
say their AI investments are delivering the expected ROI,
succeed so we seem to be off to a good start.

But most of today’s use cases are just scratching the

surface of what AI can do. Using GenAI to make developers
more productive for example, or create customer support
chatbots is undeniably useful. But the ROI from these types
of activities is unlikely to be game-changing.

The real ROI from AI will come when businesses apply

it to what they do every second of the day, day in and
day out. From paying, to collecting, to shipping, potential
AI use cases aren’t just expansive, they hold game-
changing promise. Right now, enterprise AI largely lacks
the process understanding and business context it
needs to be effective in these and other more strategic
applications of AI. (Jump to page 19 to learn about
process understanding and business context.)

Surprise, surprise the Process Intelligence company

is telling you processes and AI go hand-in-hand. But
the vast majority of business leaders agree. More than
half (58%) are concerned the current state of their
processes may limit their ability to make AI work
moving forward. And almost nine-in-ten (89%) say it’s
crucial AI has the context of how their business runs to
be effectively deployed. This all means that businesses
must understand and optimize their processes to make
the most of the AI opportunity which, incidentally, AI can
actually help them to do.

In this – the AI edition of the 2025 Process Optimization

Report – we’ll take a deeper look at how businesses are
using AI across IT, Supply Chain, Finance and Shared
Services, and Process Improvement and Operations. We’ll
also explore how enterprises are taking action to improve
process understanding, so they can give AI what it needs
to succeed.

Let’s get started.

3 What AI needs to succeed

Section one: What
enterprises are
doing with AI
Ubiquitous AI: A tiny 1% of business leaders say their organizations are
not deploying AI in one form or another. Four-in-five are
GenAI is already already using GenAI foundational models for use cases
everywhere like developer productivity, knowledge management, or
customer service. Three-in-five have deployed GenAI
chatbots and assistants for business users. If anyone
thinks the AI age is yet to start, this is your friendly
wake-up call.

What enterprises are doing with AI

81% Leveraging GenAI foundational models

Deploying GenAI chatbots and

assistants for business users

61%
Not using GenAI but using
Machine Learning

16% No AI deployments currently

1%

There are some interesting regional variations in the use of

AI. The use of GenAI chatbots and assistants, for example, is
far more popular in the US than in the DACH region.

Use of GenAI chatbots and assistants by region

75%
63% 57% 51%

US Australia, France, Austria,

Japan, Spain, UK Germany,
South Korea Switzerland
5 What AI needs to succeed
AI ROI expectations: Almost three-quarters (73%) of business leaders feel
their AI investments are currently delivering the expected
A pivotal year ROI. And a strong 86% say they clearly understand how
AI is going to help their department deliver more value.
Once again this confidence in AI ROI and value delivery
varies by region, and is strongest in the US.

Regional confidence in AI ROI and value delivery

Understand how AI will Say AI is delivering

help deliver value the expected ROI

92%
84% 86%
83%
82%

70% 69% 69%

US Australia, Austria, France,

Japan, Germany, Spain, UK
South Korea Switzerland

6 What AI needs to succeed

 So what about expectations for the future?

2025 is a pivotal year for enterprise AI, with 64% of

business leaders saying AI will deliver significant ROI,
and 73% saying there will be established use cases in
their departments, along with associated policies and
usage guidelines, in the next 12 months. Almost three-
quarters (74%) expect AI budgets to increase, although
58% say AI spend will come from a central pot rather
than each department having its own budget.

These expectations vary slightly between departments.

Process Improvement and Operations leaders have
the highest AI expectations for the coming year, while
Supply Chain leaders are a little more skeptical.

Departmental expectations of AI in next 12 months

Say AI budgets Expect established use Say AI will deliver

will increase cases in department significant ROI

50% 80%

Process
76%
and Operations
76%
66%

IT
75%
73%
66%

72%
Finance and
Shared Services 75%
62%

72%
Supply Chain 68%
61%

7 What AI needs to succeed

 Something most business leaders from all departments
and regions agree on is the need to use AI for process
improvement. A strong 81% say AI will be used to directly
improve business processes over the next 12 months.

Looking a little further ahead, AI is seen as one of the

greatest strategic opportunities for businesses over
the next two years. Implementing AI, automation,
and analytics is on par with improving productivity,
and placed above improving profitability, increasing
revenue, and improving customer experience.

Top strategic opportunities over the next two years

Percentage who
included the opportunity
in their top three

35%
Implementing AI,
automation, analytics

35% 32%
Improving productivity
Improving profitability

31%
Increasing revenue

30% 30%
Improving customer Optimizing
experiences operations /
processes
29%
Enhancing
reputation /
brand
25%
28% Boosting
Reducing costs employee skills and
engagement
25%
Improving
ESG impact

8 What AI needs to succeed

Departmental
perspectives on AI
While AI experimentation has already begun across the
enterprise, the major investment is still to come, according
to our 1,620 survey respondents. Let’s take a deeper look at
how leaders from Finance and Shared Services, IT, Process
Improvement and Operations, and Supply Chain perceive
the use of AI in their departments.
Process Process Improvement and Operations teams are
already using AI in a variety of ways, with more than
Improvement half currently using AI for both continuous improvement
and Operations: and process automation. Even in areas where use of
A focus on intelligent AI is currently low, such as identifying improvement
opportunities, there are plans to harness it in
automation the coming year.

How Process Improvement and Operations teams use AI

Already using Plan to use in

next 12 months

52% 43% 51% 44% 46% 47%

Continuous improvement Automating processes Insights to enable

with real-time monitoring data-driven decisions
of operational data

43% 49% 39% 52% 38% 48%

Supporting digital Modeling and Quality control through

transformation initiatives redesigning processes anomaly detection

37% 53% 29% 64% 24% 65%

Predictive analytics to Intelligent task Identifying improvement

inform actionable automation opportunities
strategies

Looking ahead, AI-powered automation will be a key

focus, with 89% of Process Improvement and Operations
leaders saying intelligent automation will unlock more
value than anything else in the next five years.

Read the Process Improvement

and Operations Edition

10 What AI needs to succeed

Supply Chain: Supply Chain leaders are experimenting with AI across
multiple use cases. Automated inventory management
A foothold in and warehouse automation top the list and are used by
the warehouse around half of supply chain teams. This indicates AI is
most firmly established in warehouse environments, but
other use cases like demand forecasting and supplier
relationship management aren’t far behind.

How Supply Chain teams use AI

Already using Planning to use in

the next 12 months

55% 40%
Automated inventory management

49% 49%
Warehouse automation

48% 49%
Demand forecasting

48% 45%
Supplier relationship management

48% 51%
Logistics automation

40% 48%
Automated quality checks

40% 56%
Transport network and route planning

The intention to increase AI adoption in the near future

is strong. For example, only 40% of Supply Chain leaders
are currently using AI for transport network and route
planning, but 96% say they plan to use it in the next
Read the Supply Chain Edition
12 months.

11 What AI needs to succeed

IT: Security concerns IT leaders understand the significance of AI in technological
development, with 87% saying they expect the technology
are paramount platforms they use to leverage the latest AI developments.

But despite this acceptance of AI’s pivotal role, 90%

have concerns over the increasing use of AI within
their organizations. Security and the increased risk
of data breaches is their biggest worry. The ability to
demonstrate ROI, and the skills gap between what they
have and what they need, tie for second place.

IT leaders’ concerns around enterprise AI use

Percentage who put the concern in their top three

Security and data breach concerns 45%

Skills gap between what we have and what we need 37%

Ability to demonstrate ROI 37%

Data availability and quality 36%

Unrealistic expectations about what is possible 31%

Time spent managing application sprawl and complexity 31%

Deprioritizing other investment areas to focus on AI 28%

Process shortcomings 24%

Process shortcomings come relatively low on the list,

which indicates IT leaders don’t yet fully appreciate
how important optimized processes are to effective
AI deployment (more on that in section three). Many
of the bigger concerns identified, like data availability
Read the IT Edition
and quality, can be addressed through process
improvement, but IT leaders may not yet be
12 What AI needs to succeed
connecting the dots.
Finance and Shared Finance and Shared Services leaders have a positive
outlook on the use of AI in their departments, with 70%
Services: A mandate saying they have a mandate to invest in AI-powered
for investment technologies to drive business efficiency.

The intention to push forward with adoption is evident

when we look at how they plan to use AI in the next 12
months. Right now, automating routine tasks is the most
common AI use case, already being tackled by over half
(54%) of Finance and Shared Services departments. Teams
are also starting to invest in AI for a range of additional use
cases from fraud detection and cash flow optimization
to regulatory compliance. But over the next 12 months at
least 90% of Finance and Shared Services leaders intend
to implement AI for every one of these use cases – even
those where AI is currently less commonly deployed.

How Finance and Shared Services teams use AI

Already using AI Planning to use AI in

the next 12 months

54% 41%

Automate routine tasks

43% 50%

Financial reporting and analysis

43% 49%

Fraud detection

40% 50%

Financial forecasting and planning

36% 56%

Cash flow optimization

35% 56%

Expense management
Read the Finance Edition

31% 59%

13 What AI needs to succeed Regulatory compliance

There’s no AI
without PI
AI success A lack of process understanding may already be
hindering business adoption and deployment of AI.
needs process When we asked business leaders to indicate the three
understanding biggest warning signs that process understanding in
and context their organization isn’t quite where it should be, difficulties
adopting new tech like AI was one of their most common
answers, included in the top three by 30%.

Ten signs organizations don’t understand

their processes

Percentage who included

the sign in their top three

Increasing
31%
operational costs

Difficulties adopting
30%
new tech like AI

Employee morale and retention

24%
challenges

Transformation initiatives falling

24%
short of expectations

Inability to respond effectively

23%
to regulatory changes

Increasing operational
22%
throughput time

Silos and tensions between

21%
different business departments

Inability to respond effectively

21%
to supply chain disruptions

Productivity
20%
decline

Increasing
20%
customer churn

15 What AI needs to succeed

 What’s more, the majority (58%) of business leaders
are concerned that the current state of their processes
may limit their ability to make AI work effectively for
their organization. Almost a quarter (24%) strongly feel
that this is the case.

Process shortcomings may limit value from AI

Statement: We’re concerned the state of our processes

may limit the value we can achieve with AI

24%
Strongly agree

34%
Tend to agree

26%
Tend to disagree

15%
Strongly disagree

1%
Don’t know

16 What AI needs to succeed

 So leaders clearly see the need to understand their
processes, and to improve the way those processes run,
to get AI working for the enterprise.

More specifically, they also understand the need for

AI to have the context of how their business runs,
including how they calculate KPIs, what their policies
and procedures are, and how their organization is
structured. The vast majority (89%) say it’s crucial AI has
this business context to be effectively deployed. What’s
more, 43% strongly believe this is the case, while only 2%
strongly disagree.

Effective AI deployment needs business context

It's crucial that AI has the context of how our business

runs to be effectively deployed

43%
Strongly agree

46%
Tend to agree

9%
Tend to disagree

2%
Strongly disagree

17 What AI needs to succeed

 There are some regional variations in these perceptions.
Business leaders in APAC countries are most likely to
be concerned that a lack of process understanding will
prevent them making the most of AI. And those in the US
are most likely to say effective AI deployments require
specific business context.

AI, context, and concerns by region

Are concerned processes Say it's crucial AI has

may limit value from AI the context of how our
business runs

93%
90% 89% 86%

60% 62%
55% 55%

US France, Australia, Austria,

Spain, UK Japan, Germany,
South Korea Switzerland

The bond between processes and effective AI is

undeniable. Read on to discover how Celonis Process
Intelligence makes AI smart and successful for
the business:

18 What AI needs to succeed

 AI without PI isn’t smart enough
For AI to be relevant and effective for the enterprise, it
needs both process understanding and business context.

1. Process understanding: how activities are sequenced

as a process, including how they relate across
systems, what’s happening upstream, downstream,
and in parallel – as well as knowing if activities are
happening on time and in the right order.

2. Business context: institutional knowledge including

business-specific rules, benchmarks, and KPIs that
can be codified and made accessible to AI so it can
be deployed for strategic use cases.

Celonis Process Intelligence (PI) uses process mining

to create a digital twin of a business’s end-to-end
processes, and uses AI algorithms to show teams where
value opportunities lie and how to capture them. In
addition, PI codifies and enriches business context to
make it accessible to AI tools, such as agents, co-pilots,
and assistants.

Consider the GPS you use every day to find your way
around. It’s an incredibly useful technology, but it’s only
as effective as the map data that feeds it. No system
provides this kind of input for the enterprise today. But
PI can. By creating a map of the enterprise, connecting
data across your systems, and making sense of it, PI
enables AI to understand how your organization works
and take action to run it effectively.

19 What AI needs to succeed

 Imagine you want to build an AI solution to automatically
handle credit blocks on orders. Credit blocks are set
for a good reason, but if they aren’t addressed quickly
enough orders get delayed and customers become
less than satisfied.

For AI to unblock the order, it needs to know to look at the

order in the first place. Then it has to understand why the
block was set, if this is normal or not, if the order is at risk
of a late delivery, how to prioritize it, and how to handle
the reason for the block.

Combining process understanding and business

context, Celonis makes it possible for AI to do all this. AI
is made smart enough to look at the order, understand
the reason behind the block, assess the risk of late
delivery, take customer priority into account, and
determine the right steps to move the order forward.

20 What AI needs to succeed

What Process Businesses are exploring a range of tools and tactics
to increase process visibility, which is the first step in
Intelligence makes gaining the Process Intelligence (PI) required to make
possible AI work. It’s particularly encouraging to see 39% of
businesses are already using process mining – the
technology that enables PI – and a further 52% plan
to adopt it in the next 12 months.

How enterprises are gaining process visibility

Currently use Process model reviews

Plan to use in next

12 months 67% 31%

Process mapping workshops BI tools

66% 29% 61% 34%

Process mining Task mining

39% 52% 32% 55%

Process mining visualizes, analyzes, and optimizes

business processes. It reveals why processes aren’t
working the way they are designed to, provides the
information needed to improve them, and enables PI.

As businesses across the world get comfortable with

AI, they’ll want to move past early deployments of
GenAI models and chatbots, and use the technology to
generate substantial ROI. With PI on their side, leaders
can understand their processes and give AI the business
context it needs for more advanced, strategic use cases
that will deliver a potentially game-changing impact.
21 What AI needs to succeed
What’s next?
 Right now, enterprises like yours, and people like you,
are using Celonis Process Intelligence to gain a better
understanding of processes, act on opportunities to
improve how their business runs, create value, and make
AI work.

Looking for inspiration and information to help you

do the same?

• Get up close and professional with Celonis

Process Intelligence.

• Check out Process Mining For Dummies, Celonis

Special 2nd Edition.

• Discover how PI works with AI copilots, assistants,

and agents.

Plus check out the full 2025 Process Optimization Report,

‘Making processes work’, to find out more about what
business leaders are doing in pursuit of productivity,
profitability, and effective AI.

23 What AI needs to succeed

Survey The research was conducted by Insight Avenue, an
independent, third-party, specialist B2B and technology

sample
research consultancy. A total of 1,620 interviews – spread
between four regions and four departments – were
conducted during July and August 2024. Business
leaders were interviewed from organizations with
revenue of $500m+ across a range of sectors.

The sample used in this report is made up as follows:

 Department

Supply Chain 25%

Finance / Shared Services 25%

Operations / Process Improvement 25%

IT / Digital 25%

Level of seniority

Board / C-level 6%

VP level 10%

Head of Department / Director 34%

Senior manager 50%

Organization revenue

$500m - $2bn 15%

$2bn - $10bn 80%

More than $10bn 5%

25 What AI needs to succeed

 Industry sector

Life sciences 4%

Pharma 3%

Oil & gas 10%

Retail 9%

CPG / FMCG 8%

Manufacturing 8%

IT / technology 11%

Automotive 9%

Chemicals 5%

Banking 10%

Insurance 7%

Logistics 6%

Public sector 5%

Utilities 5%

26 What AI needs to succeed

 Region

US 25%

Europe (France, Spain, UK) 25%

DACH (Austria, Germany, Switzerland) 25%

APAC (Australia, Japan, South Korea) 25%

Country

Australia 10%

Austria 5%

France 7%

Germany 15%

Japan 7%

South Korea 7%

Spain 5%

Switzerland 5%

UK 14%

US 25%

27 What AI needs to succeed

About Celonis makes processes work for people, companies,
and the planet. The Celonis Process Intelligence Platform

Celonis
uses industry-leading process mining and AI, and
augments it with business context to give customers
a living digital twin of their business operation. It’s
system-agnostic, without bias, and provides everyone
with a common language for understanding and
improving businesses, and enabling AI to be effective
and relevant for the enterprise. Celonis empowers its
customers to continuously realize significant value
across the top, bottom, and green line.

Celonis is headquartered in Munich, Germany, and New

York City, USA, with more than 20 offices worldwide.

Find out more at celonis.com

© 2025 Celonis SE. All rights reserved. Celonis and the Celonis “droplet” logo
are trademarks or registered trademarks of Celonis SE in Germany and
other jurisdictions. All other product and company names are trademarks
or registered trademarks of their respective owners.

28 What AI needs to succeed