State of the Route to Live 2025 Report PDF Free Download
1 / 52/52
100%
State of the Route to Live Report 2025
OVERVIEW
State of the Route to Live
RUNNING THE NUMBERS
End-to-end Route to Live
Team Dependencies
Individual Bottlenecks
EXPERT INSIGHTS
Myth #1
Myth #2
Myth #3
Myth #4
Myth #5
QCE Framework
THE IMPACT OF AI
What does the future looklike?
Strategic Risk Mitigation
The New RtL Paradigm
12 MONTH ROADMAP
Ideation and Planning
Design
Development
Build
Testing
Environments
Deployment
Monitoring & Observability
A Seismic Shift
Prioritisation Matrix
WHERE NEXT
The software delivery
performance gap is widening
01
02
05
08
10
12
16
17
18
19
20
21
22
23
25
28
30
31
33
34
34
36
37
38
39
40
41
42
43
44
WHAT’S INSIDE
THE ABILITY TO
EFFICIENTLY MOVE IDEAS
FROM CONCEPTION TO
PRODUCTION - WHAT WE
CALL THE “ROUTE TO LIVE”
(RtL) - HAS BECOME A
CRITICAL COMPETITIVE
DIFFERENTIATOR.
TODAY’S ORGANISATIONS
FACE UNPRECEDENTED
PRESSURE TO DELIVER
SOFTWARE FEATURES AND
PRODUCTS AT AN EVER
ACCELERATING PACE.
THE STATE OF THE
ROUTE TO LIVE REPORT
PROVIDES A DATA-DRIVEN
EXAMINATION OF THE
CURRENT STATE OF
SOFTWARE DELIVERY.
01
ClearRoute Limited
What are the most common
software delivery bottlenecks
across the Route to Live?
And how can resolving these
increase deployment frequency
by 12x and reduce production
defects by 75%?
The State of the Route to Live
report provides a data-driven
examination of the current
state of software delivery.
The ndings offer both
strategic guidance for C-suite
executives and actionable
recommendations for
engineering teams, with a focus
on practical, proven solutions to
common delivery challenges.
Sarndeep Nijjar
Chief Technology Ocer
02
State of the Route to Live Report 2025
Over the last three years, ClearRoute have
run more than 30 deep-dive Route to Live
assessments across financial services, retail,
healthcare, media and tech.
Together they represent £2 billion in annual
engineering spend and over 5000 developers
and almost every organisation wrestles the
same chokepoints: weeks of lead-time lost
to manual gates, brittle test suites masking
defects, widespread issues with environment
availability, and delivery teams cut o from
the business they’re meant to serve.
At the same time, we’re seeing breakneck
AI evolution transforming how organisations
think about software delivery. As this report
will show you, AI presents a generational
opportunity: not just to overcome entrenched
delivery challenges but to fundamentally
reshape how you conceive, build, and
deliver software.
This report works hard to cut through the
hype, to showcase tangible use cases to
integrate intelligence, address legacy issues,
and transform the end-to-end Route to Live.
Benchmarks, bottlenecks,
and breakthroughs in
enterprise software delivery:
a data-driven roadmap to
ship faster, better and safer.
State of the Route to Live 2025
WHAT IS THE ROUTE TO LIVE?
The Route to Live (RtL) encompasses the end-to-end journey of getting ideas into
customers’ hands, including the processes, tools, and cultural practices that enable
or impede delivery excellence.
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 03
ClearRoute Limited
12x
6x
75%
85%
70%
higher deployment frequencies
faster lead times for changes
fewer production defects
less time spent on manual activities
improvement in engineer satisfaction scores
TRANSFORMING SOFTWARE DELIVERY
PERFORMANCE IS A PROVEN PATH
TO IMPROVE ORGANISATIONAL
PERFORMANCE. OUR ANALYSIS REVEALS
THAT ORGANISATIONS WITH OPTIMISED
ROUTE TO LIVE CAPABILITIES ACHIEVE.
For leaders, understanding the current state
of RtL gives context for evaluating your own
capabilities and identifying improvements
with the highest ROI.
For teams battling mountains of toil, friction,
and rework, understanding how teams like
yours tackle the same challenges can reveal a
path out of the fog.
Benchmark your eorts against
cross-industry real-world data analysis
See which five big myths do most to
block performance improvement
Learn how to embed AI into your RtL
(Hint: it’s not just GitHub Co-Pilot)
Get an actionable roadmap for the next
12 months with concrete targets
These metrics translate
directly to business outcomes:
faster time-to-market,
improved product quality,
higher customer satisfaction,
and more ecient use of
engineering resources.
WHATEVER YOUR CURRENT
STATE OF RTL MATURITY,
AI ADOPTION, AND AI
PREPAREDNESS, READ THIS
REPORT TO:
04
State of the Route to Live Report 2025
RUNNING
THE
NUMBERS
05
ClearRoute Limited
Running the
numbers
We conducted more than 30 Route to Live
assessments across (mostly) mid-large
enterprises in financial services, retail,
healthcare, media, and technology.
These assessments collectively represent
over £2 billion in annual technology
investment and involve more than 5,000
engineering professionals.
This section turns those findings into a
public benchmark. We combine hard
numbers — lead-time, change-failure, pipeline
reliability — with qualitative evaluation of
organisational capabilities and processes.
You’ll see from the data there’s typically
significant variance, but many shared pain
points and opportunities too.
These challenges don’t just impede today’s
performance. They also derail tomorrows, by
critically impacting organisations’ readiness
to harness the generational opportunity of
AI – which we’ll explore later.
Nonetheless, the success of higher
performers should give hope to organisations
on the lagging end of these benchmarks.
These issues often feel entrenched and
impossible to solve but the data proves
there’s light at the end of the tunnel.
Throughout the rest of this report, we’ll show
you some of the explicit levers you can pull to
move the needle.
WE SLICE THE DATA THREE WAYS:
End-to-end RtL: the bird’s eye path
from idea to production
Team intersections: where hand-os,
siloes, and governance collide
Individual teams: into the reeds of toil,
tooling, and developer eciency
06
State of the Route to Live Report 2025
LEAD TIME TO PRODUCTION
Fastest 3.74 days
Longest 104
Median 30-60 days
DEFECT RATES
11-300 annually
9.15%+ damages reliability & eciency
CHANGE FAILURE RATES
Failure rate up to 98%
PIPELINE RELIABILITY
Highest 24.58 per day
Median 1 every 1-2 weeks
Lowest 1 per month or longer
DEPLOYMENT FREQUENCY
Automated 4 to 12 hours
Manual 1–2 days typical (highest 55 days)
Mixed 5-10 days average
REGRESSION TESTING
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 07
ClearRoute Limited
End-to-end Route to Live
Lead time to production
Lead time to production varies widely
from 4 days to 104 weeks, indicating
stark contrasts in SDLC maturity.
• Fastest observed: 3.74 days
• Longest observed: 104 weeks
• Median range across assessments:
30–60 days
Defect rates
Defect rates remain high, with
production incidents ranging from
11 to over 300 annually.
• Production incidents range from
11 to over 300 annually
• Incidents often cluster in lower
priority categories but remain a
persistent delivery issue
Change failure rates
Change failure rates of up to 9.15% hurt
reliability and operational eciency.
However, this stat belies some complexity.
Our average here is still under the industry
benchmark of 15%. But that benchmark is
for organisations of all sizes. We’ve looked
predominantly at mid-large enterprises,
which are typically more risk-averse
anyway. Less innovation naturally correlates
to a lower change failure rate. Plus, this
lower rate is demonstrably coming at the
cost of delivery speed.
• Change failure rates up to 9.15%
The first portion of our
data analysis considers the
end-to-end RtL, looking
across the full software
delivery lifecycle from
requirement to production.
For this analysis we started measuring only
once work was accepted by the delivering
team. That means it doesn’t account for
time spent in any planning or strategy
processes before that point, which can also
cause significant delays even before the RtL
ocially kicks o.
08
State of the Route to Live Report 2025
Deployment frequency
Deployment frequency varies significantly from high-frequency deployments several times daily to
low-frequency releases once every few weeks.
• Highest observed: 24.58 deployments/day
• Lowest observed: monthly or longer intervals
• Median frequency: 1 deployment every 1–2 weeks
Pipeline reliability
Broken builds and flaky pipelines quietly sponge capacity and force rework, indicating poor
CI/CD maturity.
• High pipeline failure rates common – up to 98% in extreme scenarios
Regression testing
Regression testing remains largely manual across many teams, contributing significantly to cycle
times. Automated testing is minimal.
• Automated packs: 4 to 12 hours
• Manual testing: 1–2 days typical, with one case at 55 days for a single tester
• Mixed methods: combined eort ranges from 5 to 10 days on average
WIDER INDUSTRY DATA SHOWS HIGH-PERFORMING
ORGANISATIONS TYPICALLY SHOW
• Deployment multiple times per day
• Change failure rates under 15%
• Lead times under one week
• Incident recovery in under an hour
It seems clear that larger organisations continue to err on the side
of caution, consciously sacrificing speed to minimise change failures.
Elite organisations don’t make this trade-o.
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 09
ClearRoute Limited
Team dependencies
across the RtL
The next portion of our
analysis looks at the
intersections of teams
across the Route to Live.
Team intersections are
critical to speed and
stability: handovers, siloes,
and manual checkpoints
add considerable friction.
The gold-standard model for fast,
reliable delivery means empowered,
cross-functional engineering teams who
own changes holistically through to
production, protected by guardrails like
policy-as-code. But our analysis shows
many companies aren’t here yet.
Team involvement per change
The friction from cross-team collaboration
significantly aects operational eciency.
Our analysis shows that many RtLs
involve a slew of dierent teams including
feature development, QA/testing, release/
governance, DevOps or platform engineering,
infrastructure support, cybersecurity,
change advisory boards, plus downstream
systems, compliance, and enterprise
architecture functions.
• Average of 6 teams contributing
to the lifecycle of a change
• In more complex organisations,
10+ teams contribute to change
Siloed structures
Siloed structures were a common pain
impeding collaboration, particularly around
data migration and shared infrastructure.
We saw:
• Data migration teams often operate
independently of core delivery crews
• QA, DevOps, and engineering strategies
are often misaligned with each other,
hindering cohesive execution
10
State of the Route to Live Report 2025
Lengthy governance and approvals
Processes involving penetration tests and governance approvals introduce substantial delays.
• Penetration testing lead times: 4 weeks
• Change Request processes from 4 to 8 weeks
• Teams often unable to deploy to production without approvals from multiple
peer reviewers, causing 2 to 4 day delays
Shared resource constraints
We consistently saw how shared environments and shared roles exacerbate dependency
management and create bottlenecks.
• Shared test environments lead to scheduling conflicts and test contention
• Specialist roles assigned over multiple delivery teams split time and priorities,
creating pinch points
• Sandbox refresh requests take multiple days to process
• Infrastructure managed by central teams isn’t owned by delivery teams,
impacting autonomy, lengthening lead times, and undermining eciency
INDUSTRY BENCHMARKS FROM THE
LATEST DEVOPS REPORTS SHOW THAT
CROSS-FUNCTIONAL TEAMS WITH
END-TO-END OWNERSHIP TYPICALLY
ACHIEVE HIGHER OPERATIONAL
EFFICIENCIES AND REDUCED DELAYS.
OUR DATA SHOWS THAT’S RARELY
HAPPENING – AND THE COST IS BOTH
VELOCITY AND CONFIDENCE.
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 11
ClearRoute Limited
Our analysis spotlights three
significant automation gaps
that are hurting speed,
stability, and developer sanity.
Testing automation
Our analysis showed some use of automated
testing but it wasn’t the norm and results
were often unreliable.
• Extensive manual regression testing
still happening
• Manual testing takes 1 to 2 days to
over 50 days
• Automated tests often unreliable: E2E
test success rates as low as 45%
• Some test pipelines require disabling
key functions to run successfully
• Coverage metrics often not
actively tracked
The final portion of our data
analysis considered the
process bottlenecks inside
individual teams across the
Route to Live. If hand-os
are the visible trac jams on
the Route to Live, manual toil
inside each team represents
crater-deep potholes that
derail delivery and drive
engineers to despair.
Individual team RtL bottlenecks
12
State of the Route to Live Report 2025
Deployment processes
Manual deployment and documentation practices significantly delay RtL cycles and
increase operational overhead. Where organisations had automated the release process,
the time-dierence was from weeks to minutes.
• Manual deployments to production environments are still used
• Documentation-heavy change processes were common
• Fully automated total release process time: 68 minutes
• Manual and gated release process time: multiple weeks
Environment and test data management
Environment and test data management were pain points for almost everyone, with teams
often facing unreliable environments and unavailable test data. Automated provisioning and
configuration solutions were either insuciently implemented or absent altogether.
The most common issues:
• Test environments experience frequent outages or inconsistencies leading to recovery
time of 2-3 weeks in some instances.
• Poor test data management drives unreliable test outcomes and delivery timelines
• Developers lack tools to autonomously provision or restore test environments, causing
frustration and delay
THE LATEST INDUSTRY DATA SHOWS
ORGANISATIONS THAT HAVE EMBRACED
AUTOMATED, SCALABLE SOLUTIONS
ACROSS THESE AUTOMATION GAPS
TYPICALLY REPORT SHORTER CYCLE
TIMES, IMPROVED QUALITY, AND HIGHER
DEVELOPER SATISFACTION.
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 13
ClearRoute Limited
4.5 DAYS PER FEATURE CHANGE,
WAITING FOR CODE REVIEWS
35% OF TIME ON TOIL
23 TOOLS PER DAY NAVIGATED
IN SEVERAL CASES, MORE THAN
HALF OF DELIVERY TIME WAS
SPENT ON MANUAL ACTIVITIES.
ON AVERAGE,
INDIVIDUALS SPEND
14
State of the Route to Live Report 2025
INSIGHTS
FROM THE
EXPERTS
15
MYTH MYTH MYTH MYTH MYTH
More process =
more safety
More QA =
higher quality
REALITY REALITY REALITY REALITY REALITY
You need better
governance, not more
Focus on reliability,
not volume
Microservices are often
distributed monoliths
Microservices are a silver
bullet for sluggish delivery
Business alignment
takes eort
Engineers develop what
the business needs
DevEX is a performance
lever, not a perk
Developer experience
doesn’t matter
12345
ClearRoute Limited
The problems organisations experience
across the Route to Live are many and
varied, but they’re also common and
consistent. We see teams struggling with
the same bottlenecks again and again.
We believe many of these issues stem
from five pervasive myths that impact how
engineering happens across the organisation.
Challenging these myths helps organisations
start to take control over the Route to Live:
to ship better products, faster and safer.
Insights from the experts
16
State of the Route to Live Report 2025
REALITY:
YOU NEED
BETTER
GOVERNANCE
,
NOT
MORE.
You don’t need more governance for
governance’s sake. You need governance
that’s fit for purpose. Governance that’s
actual risk management, not documentation.
• Organisations with compliance
embedded as code into their pipelines
cut release delays by 4 weeks and
reduce change failure rates by 64%.
• Organisations with automated
operational readiness checks instead
of service transitions have 58% fewer
production incidents
• Organisations with security practices
embedded within delivery teams
identify vulnerabilities 4x faster and
have 84% fewer security incidents.
The safety dividend comes from fast,
deterministic feedback loops, not more
paperwork. And it’s the foundation for
trustworthy AI integration.
MYTH #1:
MORE PROCESS = MORE SAFETY
For most enterprises governance is still a
manual paper-based process, complete with
endless documents, checkpoints, serial sign-
os, and external CABs. Processes snowball
to include drawn-out service transitions and
checks from centralised cyber teams.
Everyone knows these controls slow delivery
but teams believe they’re a necessary evil in
the name of safety, stability, and security.
But the data shows that’s just not true.
In practice, governance layers pile up
with no clear owner, and the process
becomes incrementally slower and more
bureaucratic, without adding more value.
Delivery timelines are pushed back weeks,
if not months. Mountains of documentation
covers everyone’s back. But the process
fundamentally does little to reduce risk.
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 17
ClearRoute Limited
REALITY:
FOCUS ON RELIABILITY,
NOT VOLUME
Quality comes from a coherent testing
strategy and an embedded, holistic quality
engineering approach, not just slapped-on
automation and bulked-up QA teams.
When quality is truly embedded into
the development process and automated
eectively, organisations need fewer
dedicated QA resources and quality
outcomes improve. These reliable systems
are the essential foundation to begin
embedding AI eectively, to augment and
accelerate testing.
Our data showed that leaders in quality
engineering achieve:
• 100% reliability through proper
isolation and data management
• 81% reduction in post-deploy defects
• 64% faster delivery cycles
• 78% improvement in first-time quality
• 92% reduction in time spent on manual
regression testing
MYTH #2:
MORE QA = HIGHER QUALITY
Bloated QA functions feel like an insurance
policy, so enterprises keep hiring or keep
throwing resources blindly at test automation.
But these are both blunt levers: they feel
rigorous on paper but don’t actually correlate
to quality.
• Teams with 1:1 QA:Dev ratios had
43% more defects
• 62% of test automation suites have
running times exceeding 5 hours
• 45% of automated tests are unreliable,
creating false positives/negatives
Each extra QA engineer adds another
hand-o, another queue, another meeting.
Coordination overhead balloons but the
real root-causes stay buried. More QAs are
good for spotting more defects – not for
reducing them.
Meanwhile, test automation suites often swell
into flaky five-hour marathons that generate
false results and need rerunning manually.
18
State of the Route to Live Report 2025
REALITY:
MICROSERVICES ARE OFTEN
DISTRIBUTED MONOLITHS
Instead of the promised fleet of independent
services, many teams stitch together a
distributed monolith with dozens of tightly
coupled APIs, version mismatches and
cross-team dependencies that amplify rather
than reduce coordination costs.
Microservices are an architectural tool, not
a silver bullet. They unlock speed only when
teams invest in clear platform boundaries,
robust tooling, automated contract tests
and observability that makes service
interactions obvious.
These elements are essential for
driving eciency - and moving towards
leveraging agentic AI - within complex
service landscapes.
MYTH #3:
MICROSERVICES ARE A SILVER
BULLET FOR SLUGGISH DELIVERY
That’s the promise peddled across the
industry. Break it down, speed it up.
A compelling narrative, but one that often
crumbles under real-world pressures when
implemented naively. Microservices can
enable delivery improvement. But as many
have discovered, implementing microservices
and implementing microservices well are two
very dierent things.
The truth is, after adopting microservices:
• 67% of organisations report
increased complexity
• Only 23% achieve the expected
speed increase
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 19
ClearRoute Limited
REALITY:
BUSINESS-ALIGNMENT
TAKES EFFORT
Alignment isn’t osmosis. Unless leaders
broadcast clear objectives, wire business
metrics into the pipeline, and keep the
dialogue two-way, engineering will keep
shipping technically solid solutions to poorly
understood problems. And engineering
teams will continue delivering unpredictable,
inecient value against the goals they exist
to serve.
• Organisations with clear,
communicated objectives
achieve 47% higher feature
adoption rates
• Teams with business-aligned
metrics deliver 3.2x more
value-generating features
• Organisations with advanced
observability practices connecting
delivery metrics to business
outcomes achieve 45% higher
customer satisfaction
And as we’ll explore in the next sections,
this alignment is even more powerful when
you embed AI, to analyse and act on these
integrated data streams.
MYTH #4:
ENGINEERS DEVELOP WHAT
THE BUSINESS NEEDS
Most organisations believe engineers
understand what they’re working on and why,
trusting them to develop features that deliver
value against strategic goals. Engineers are
close to the product day-in, day-out. There’s
a concordant assumption that this deep
technical knowledge comes hand-in-glove
with commercial context.
But our data shows that’s not the reality.
Actually, only 23% of engineering teams
can articulate how their work connects to
business objectives. The rest optimise for
technical outcomes not value, investing
resources into nice-to-have dead-ends.
20
State of the Route to Live Report 2025
MYTH #5:
DEVELOPER EXPERIENCE
DOESN’T MATTER
Leadership often treats DevEx as an
afterthought, if it’s on radar at all. But the
primary constraint on software delivery is
not usually technical capability but cognitive
capacity. Every friction point in a developer’s
day bleeds speed, hurts reliability, and
hampers productivity.
On average:
• Engineers navigate 23 tools in their
daily workflow
• 72% report significant context switching
• Engineers wait 4.7 hours each week for
builds, tests or environments
• Teams navigate ~7 dierent approval
processes to deliver a single change
• Engineers wait 4.5 days per feature
change for code reviews
• Teams spend 35% of their time on toil
REALITY:
DEVEX IS A PERFORMANCE LEVER,
NOT A PERK
Developer experience is the most undervalued
but highest-impact dimension of delivery
performance. Organisations invest heavily in
tools and technologies to optimise delivery
but often overlook the human factors that
determine how eectively those tools are used.
DevEx is especially critical given the
proliferation of AI tooling. Developers without
cognitive capacity can’t eectively engage
with and guide the advanced tools that
organisations must learn to grapple with.
Our data makes a compelling case for
DevEx investment:
• Developers in high-performing
organisations spend 81% of their time on
value-adding activities compared to just
51% in low-performing organisations
• Teams with optimised developer
experiences deliver features 3.7x faster
than those with significant friction points
• Organisations with dedicated DevEx
teams see a 42% reduction in
onboarding time and a 54% decrease
in context-switching costs
• Organisations with a dedicated tooling
team achieve 3.2x higher deployment
frequency and productivity returns of
£5.7 to every £1 spent
• When organisations adopt and govern
Golden Paths eectively, lead time is
reduced 78%.
• Developers report 63% higher job
satisfaction in organisations that prioritise
eliminating unnecessary cognitive load.
• Organisations that systematically address
developer cognitive load see:
• 68% reduction in lead time for changes
• 75% decrease in self-reported frustration
• 57% improvement in code quality metrics
• 89% increase in innovation activities
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 21
ClearRoute Limited
Why the Quality Cloud
Engineering Framework?
Understanding these myths can help
organisations identify discrete improvements
to improve software delivery performance;
like automating release governance, or
improving observability.
But we’ve seen time and again that the
true transformation happens when you
look through one lens, to consider the
Route to Live as a whole. That lens is the
Quality Cloud Engineering framework.
A
u
t
o
m
a
t
i
o
n
&
G
o
v
e
r
n
a
n
c
e
R
e
l
e
a
s
e
A
u
t
o
m
a
t
i
o
n
O
b
s
e
r
v
a
b
i
l
i
t
y
M
o
n
i
t
o
r
i
n
g
&
G
o
l
d
e
n
P
a
t
h
M
e
a
s
u
r
i
n
g
E
n
g
i
n
e
e
r
i
n
g
D
e
v
e
l
o
p
e
r
H
u
b
T
e
s
t
F
i
r
s
t
C
o
n
t
i
n
u
o
u
s
T
e
s
t
i
n
g
T
e
s
t
S
t
r
a
t
e
g
y
S
E
C
U
R
I
T
Y
·
C
U
L
T
U
R
E
·
A
I
A
G
E
N
T
S
E
C
U
R
I
T
Y
·
C
U
L
T
U
R
E
·
A
I
A
G
E
N
T
S
E
C
U
R
I
T
Y
·
C
U
L
T
U
R
E
·
A
I
A
G
E
N
T
C
L
O
U
D
P
L
A
T
F
O
R
M
D
E
V
E
L
O
P
E
R
E
X
P
E
R
I
E
N
C
E
Q
U
A
L
I
T
Y
E
N
G
I
N
E
E
R
I
N
G
QUALITY
CLOUD
ENGINEER
QCE
The QCE framework gives you a holistic view
across the end-to-end Route to Live, which
is the foundation to systematically resolve
fragmentation, decrease friction and unblock
complex barriers to production.
It’s a mirror for the complex, dependent
ways friction actually builds up – snowballing
momentum from incremental change into
radical compound advantage that takes full
advantage of evolving AI.
22
State of the Route to Live Report 2025
THE IMPACT
OF AI
23
ClearRoute Limited
The Impact of AI
The AI landscape is
evolving at a breakneck
pace, heralding a profound,
generational shift for the RtL.
This isn’t just about adding
new tools. AI is fundamentally
reshaping how high-quality
software is conceived, built,
and delivered.
But we observe a large spectrum of AI
adoption, from basic LLM assistance at
isolated touchpoints to pioneering
non-deterministic multi-agent ecosystems
that collaborate and continuously improve
across the end-to-end RtL.
But this advanced AI use is - so far - rare.
There’s a systematic disconnect between
what’s possible and what enterprises can
actually achieve. There’s AI tooling flying
around everywhere and huge hype, but little
meaningful insight on how large enterprises
can turn pipedreams into ROI-driving realities.
AI’s transformative potential can’t be
unlocked in isolation. Our analysis shows
a recurrent AI Preparedness Gap, where
organisations grapple with foundational RtL
weaknesses that mean they’ll struggle to
eectively scale advanced AI.
Slapping AI ad-hoc onto poor processes
and messy architecture won’t work. Legacy
issues like manual gates, brittle test suites,
environment issues, and siloed teams
create an unstable foundation, hampering
AI adoption.
Organisations with optimised Route to Live
capabilities already achieve better software
delivery outcomes by several orders of
magnitude. This also means they’re better
placed to take advantage of AI as it continues
its lightspeed evolution.
Elite performers that already boast
exceptional software delivery performance
can expect AI to deliver more advanced
optimisation, intelligently automate complex
tasks, and unlock new innovation capabilities.
Not just boost speed.
But AI isn’t only transformative for the
Googles of the world. Organisations
struggling with foundational issues can
still leverage AI to tighten and refine and
will see big leaps forward by doing so. For
organisations with less mature DevSecOps
practices, for example, well-targeted AI
initiatives around specific bottlenecks will
stamp on the accelerator.
The key is thoughtful, strategic AI
implementation tailored to your unique
context. RtL “perfection” shouldn’t be the
gatekeeper to AI adoption. Indeed, enhancing
your core delivery pipelines and strategically
adopting AI can come hand-in-hand, mutually
reinforcing eorts and together shaping an
AI-driven future.
24
State of the Route to Live Report 2025
What does the future look like?
An agentic blueprint.
ORGANISATIONS MUST TAKE A THOUGHTFUL,
STRATEGIC APPROACH TO AI ADOPTION,
CONSIDERING WHERE YOU ARE TODAY AND
DEVELOPING A ROADMAP TO ADVANCE MATURITY,
WHILE PROACTIVELY MANAGING THE ESCALATING
COMPLEXITIES AND RISKS.
Our AI Maturity Framework maps five distinct levels of AI maturity:
PREDICTIVE ASSISTANCE
Foundational AI oers pattern-based suggestions like code autocompletion,
with minimal deep organisational context. Human oversight is paramount.
CONTEXT-AWARE GENERATION
AI leverages organisational context (via RAG) from codebases and
documentation to generate more relevant outputs, like tailored user stories
or initial test cases.
ENVIRONMENT-CONNECTED INTELLIGENCE
A critical shift where AI gains bidirectional integration with enterprise tools
(Jira, GitHub, CI/CD), enabling it to retrieve data and initiate actions (like
updating Jira epics or triggering test suites).
AUTONOMOUS REASONING LOOPS
Advanced AI agents reason, decompose tasks, and interact with
environments to craft solutions. At this collaborative stage, AI augments
human expertise (like in feasibility analyses or feature development) and
humans provide crucial oversight, guidance, and validation.
MULTI-AGENT ORCHESTRATION
At the pinnacle of AI maturity, collaborative AI ecosystems of
specialised agents achieve complex, end-to-end RTL objectives (such
as requirements negotiation, feature implementation, and deployment)
with minimal human intervention.
1
2
3
4
5
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 25
ClearRoute Limited
ROUTE TO LIVE
AI MATURITY FRAMEWORK
LEVELS OVERVIEW
LOW MATURITY
HIGH MATURITY
1
Predictive
Assistance
Basic generative AI providing autocompletion and
suggestions based on patterns, with minimal
context awareness.
2
Context-Aware
Generation
Generative AI with organisational context through
RAG, understanding codebases, documentation,
andorganisational guardrails.
3
Environment-
Connected
Intelligence
AI systems with bidirectional integration to
organisational tools (Jira, GitHub, CI/CD), able to
both retrieve real-time data and make updates to
these systems.
4
Autonomous
Reasoning Loops
AI agents that can reason through complex
problems, break down tasks, and interact with
development environments to implement, test,
and refine solutions.
5
Multi-Agent
Orchestration
Collaborative AI systems where specialised agents
work together under orchestration, delegating
tasks among themselves to achieve complex
objectives with minimal human intervention.
ROUTE TO LIVE
AI MATURITY FRAMEWORK
LEVELS OVERVIEW
26
State of the Route to Live Report 2025
ULTIMATELY, ELITE ORGANISATIONS WILL
WEAVE INTELLIGENT AGENCY INTO THE FABRIC
OF SOFTWARE DELIVERY. AS ORGANISATIONS
ADVANCE, AI EVOLVES FROM ASSISTANT TO
PROACTIVE PARTNER, AND FINALLY, TO AN
ORCHESTRATED ECOSYSTEM OF SPECIALISED
AGENTS. THIS TRANSFORMATION TOUCHES
EVERY STAGE OF THE ROUTE TO LIVE.
Requirements and Design:
AI evolves from basic assistance like
suggesting user stories (Level 1-2) to
data-driven trade-o analysis (Level 3)
to AI agents collaboratively negotiating
requirements and dynamically evolving
system architecture (Level 4-5).
Development:
AI matures from code completion
(Level 1) to context-aware generation
using internal libraries (Level 2), then to
direct IDE integration for automated fixes
(Level 3). Ultimately, AI enables agent-led
feature implementation and even full AI
development teams (Level 4-5).
Testing:
AI shifts from suggesting test cases
(Level 1-2) to connecting with CI/CD to refine
test suites (Level 3); performing autonomous
exploratory testing and root cause analysis
(Level 4); and finally culminating in
specialised AI agents (security, performance,
compliance) delivering continuous,
comprehensive quality assurance (Level 5).
Deployment & Operations (IAC):
AI advances from suggesting IaC
configurations (Level 1-2) to simulating
deployments and validating drift (Level 3).
The next step is managing full pipelines and
adapting infrastructure to load (Level 4); then
eventually a fully AI-led, self-healing delivery
mesh that autonomously coordinates rollouts
and responses (Level 5).
Feedback, Maintenance, Governance and
DevEx:
AI progresses from assistive generation
like summarising feedback and drafting
documentation (Level 1), to the emergence
of co-collaborative agentic development
environments (Level 3) - where engineers
gain access to LLMs and tool via self-service
workbenches. These tools support persistent
planning, context-aware documentation,
and in-repo workflows that evolve with the
codebase.
Finally, level 5 introduces fully orchestrated,
self-service agent ecosystems - where
agents not only triage and fix issues, but
also participate in the feedback loop itself,
providing autonomous workflows that can:
reduce pull request cycle times, maintain
system hygiene, optimise Developer
Experience, and embed continuous
governance across the development lifecycle.
This journey illustrates AI
becoming an indispensable,
intelligent, and increasingly
autonomous component of a
high-performing RTL. But the
critical point is, AI oers major
gains wherever you are now.
Even moving into Level 1 AI
adoption at key bottlenecks can
drive talked-about outcomes.
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 27
ClearRoute Limited
Strategic risk mitigation
is an essential lever
AI is transformative for the RtL and that
brings exciting, cutting-edge use cases.
But teams should be wary of
fly-by-the-seat-of-your-pants change.
Advancing AI maturity necessarily comes
with an evolving risk landscape, as each level
introduces new complexities that demand
proactive management.
Risks escalate from data governance and
bias concerns at early levels of AI maturity, to
operational risks with AI-system interactions,
then to challenges in ensuring explainability
and control for autonomous decisions.
As teams approach multi-agent
orchestration, risks from highly autonomous,
collaborative AI systems are most acute.
These require stringent inter-agent
contracts, continuous observability and
traceability, and the ability to short-circuit
control loops.
To navigate this journey safely and
successfully, organisations’ risk management
practices, governance, and ethical
considerations must mature in lockstep
with AI capabilities. Strategic risk mitigation
tailored to where you are and where you’re
moving is a critical enabler of sustainable,
responsible, safe AI adoption.
28
State of the Route to Live Report 2025
LLM + Prompt
LLM + Retrieval
LLM + Retrieval + Actions
Multi Agent Systems
Many Tools & Reasoning Loop+
LLM
LLM RAG
LLM Tools
RAG
LLM
Tools
RAG
AN EVOLUTION
AI AGENTS
AN EVOLUTION
AI AGENTS
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 29
ClearRoute Limited
The new RtL paradigm: intelligence,
adaptive, and autonomous
AI is redefining the Route
to Live, charting a course
from simple co-pilot to a
fully integrated system of
collaborative agents.
This creates a new RtL paradigm. One that’s:
• More intelligent, learning from
vast datasets
• More adaptive, responding
dynamically to changes
• More autonomous, freeing human
talent for strategic innovation
Organisations that master foundational
delivery practices and strategically climb
the AI maturity ladder will experience a
RtL that’s lightyears faster, more secure,
and more resilient.
But the stakes are higher than ever to
improve software delivery performance.
Where organisations are struggling with
RtL fundamentals - as our analysis shows
so many still are - these legacy issues
create a shaky foundation that’s not able
to accelerate AI maturity at pace and scale.
For these organisations, your perspective
must be two-fold: achieve core excellence
and embark on strategic AI maturation.
Waiting for the former before considering
the latter isn’t just an unaordable speed
restriction to capitalising on the most
monumental innovation since the
dot-com boom. It’s also ignoring the
mutually reinforcing benefits of
championing these priorities in tandem.
AI doesn’t just oer use cases that everyone
at Google’s talking about. It’s also the
opportunity to bake-in the right intelligence
to tackle many of your thorniest historical
challenges with the RtL - miles easier and
faster than before.
Whatever the state of your
current RtL and whatever
your AI maturity, sleeping on
AI isn’t an option. The future
is radically dierent – and it’s
just around the corner.
30
State of the Route to Live Report 2025
31
ROADMAP
FOR THE
NEXT 12
MONTHS
ClearRoute Limited
32
Roadmap for the next 12 months
Over the next 12 months,
elite teams might not
look radically dierent
on the surface. They’ll
still write code, run tests,
and ship products.
But the dierence will be
how much toil disappears,
with AI, platform guardrails,
and tighter feedback loops
encasing every stage of
the SDLC.
Below we’ve mapped out
where enterprises typically
stumble today, what tools
they have (or are starting
to have) in play, and the
near-term bets that will
bend the curve on speed
and stability.
If your pipeline isn’t
where it could be, here’s
where to focus to start
driving progress.
State of the Route to Live Report 2025
Ideation and planning
THE BIG ISSUES
Ideation and planning are often a flurry of ideas but lack rigour. Scope is hazy, risk
assessment is late, and gut feel still green-lights too many non-viable projects.
HOW TEAMS ARE SOLVING NOW
Teams often engage with collaborative whiteboarding (physical and digital),
implement basic project management for ideation, and conduct rudimentary market
research. But data analysis is often high-level and anecdotal.
ACTIONS FOR THE YEAR AHEAD
Focus on improving data-driven ideation, using AI-agent market and feasibility
analysers that triage ideas and predict roadblocks before they hit Jira.
AI can also analyse contributions to facilitate better brainstorming, to empower
better products downstream.
Expected impact
• 10-15% reduction in time on non-viable ideas
• 5-10% improvement in correlation between scope and value
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 33
ClearRoute Limited
Design
THE BIG ISSUES
Traditional waterfall design often created a gulf between user needs and technical
feasibility, with lengthy, assumption-based upfront planning driving endless rework.
Accessibility and inclusivity were often an afterthought, and design and development
frequently became siloed. That said, there’s now growing recognition of the value of
robust upfront planning – “good input equals good output” – but with challenges in
making this crucial strategic thinking dynamic and responsive.
HOW TEAMS ARE SOLVING NOW
Teams usually lean heavily on UI/UX design tools (like Sketch, Figma, Adobe XD) and
prototyping platforms (like InVision and Marvel). Basic user testing is common. However,
without a framework for rapid iteration within the design phase, these tools can still be
constrained by longer, inflexible planning cycles. This causes delays and hampers true
understanding of user needs and technical viability.
ACTIONS FOR THE YEAR AHEAD
Embrace an evolved approach to upfront design. Supercharge design by
combining the thoroughness of waterfall planning with AI-driven iteration. Instead
of committing to a single design path for an extended period, the focus shifts to
rapidly creating multiple POCs and fostering a ‘test and learn’ culture, accelerating
time to innovation hugely.
This iterative, AI-enhanced upfront design phase allows teams to get as close to the
customer as possible, test multiple hypotheses quickly, and make data-informed
decisions before committing significant development resources. It empowers enhanced
accessibility and inclusivity through earlier, more frequent validation and improves
product-market fit.
Expected impact
• Significant improvement in time-to-innovation
• 15-25% reduction in rework
• 20-30% faster design iteration cycles
34
State of the Route to Live Report 2025
Development
THE BIG ISSUES
Development cycles still sag under poor quality code, weak standards, poor
feedback cycles and endless toil. Rush jobs breed spiralling tech debt.
Collaboration is an ongoing challenge.
HOW TEAMS ARE SOLVING NOW
Many engineering teams are actively integrating AI-powered code completion tools like
GitHub Copilot. Early observations often suggest an increase in perceived developer
eectiveness and positive sentiment.
But the promise of these tools is nuanced. Individual productivity might feel boosted
but emerging data and industry observations suggest that overall team throughput and
final code quality may not improve proportionally without careful management.
One significant factor is the increased time engineers might spend debugging,
understanding, and refactoring AI-generated code, which can introduce subtle or
complex bugs if not rigorously reviewed and validated.
ACTIONS FOR THE YEAR AHEAD
Harness AI code generation tools strategically. Don’t just “adopt” tools. Integrate
them fully with robust validation and quality assurance practices. Pair AI code
completion with clear guardrails and standards. Enrich AI-driven reviews with
static analysis for deeper insight on code quality and remediation. Develop
clear guidelines and training for engineers. Establish golden paths and integrate
organisational best practices into AI code generation and review. Integrate
AI-generated reviews for correctness and reliability, promoting early and frequent
feedback, freeing human reviewers to shift focus onto complex logic, architectural
soundness, and system impact.
Expected impact
• 10-15% reduction in time on non-viable ideas
• 5-10% improvement in correlation between scope and value
• ≈ 90% of security vulnerabilities caught pre-deployment
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 35
ClearRoute Limited
Build
THE BIG ISSUES
CI/CD concepts are mature but many build pipelines remain brittle. Tightly coupled tests
and dependency-heavy regression packs mask weak early-stage testing, creating slow,
flaky feedback loops that derail velocity and sap confidence.
HOW TEAMS ARE SOLVING NOW
• Orchestration tools (Jenkins, GitLab CI, GitHub Actions) automate
build/test/deploy, driving faster, more frequent releases with better
consistency and reliability.
• Artifact repositories (Nexus, Artifactory) store and manage build artefacts
centrally, acting as a single source of truth that powers better traceability,
simplifies version control, and makes collaboration and re-use easier.
• Containerisation (Docker) is becoming the standard for consistency,
guaranteeing consistency and eliminating the classic “works on my machine”
problem. To simplify deployment and facilitate truly scalable, portable
microservices architectures.
ACTIONS FOR THE YEAR AHEAD
Continue to relentlessly pursue loosely coupled builds. Integrate AI dependency
analysis to flag risky merges, intelligent test selection to trim runtimes, and anomaly
detection that parses build logs before humans triage.
Expected impact
• 15-20% faster build times
• 5-10% fewer integration-driven failures
• ≈80% of integration risks flagged pre-build
• 10-15% faster test execution
• 5-10% reduced MTTI for build failures
36
State of the Route to Live Report 2025
Testing
THE BIG ISSUES
Contract testing is on the rise but real-world roll-outs stumble. Frameworks get adopted
but service teams struggle to wire them into pipelines. API testing is under-used so
defects hide in the seams. And teams still wrestle static or scrubbed data sets - and
without realistic, dynamic data, critical issues aren’t caught until they cause chaos.
HOW TEAMS ARE SOLVING NOW
• Contract testing frameworks (like Pact) bolster microservices stability,
empowering teams to prove service compatibility early and deploy independently
without heavyweight end-to-end runs.
• API testing tools (Postman or RestAssured) hits endpoints directly, surfacing
logic- or contract-level bugs early to improve API reliability and usability. Essential
for robust API-first development.
• Synthetic data and data masking experiments address static data limitations
by generating fresh, representative, and privacy-safe datasets. So test runs stay
realistic, repeatable, and compliant.
ACTIONS FOR THE YEAR AHEAD
Adopt a single, strategic test architecture. Fold AI-generated, realistic test data into
your pipelines. Use AI test-prioritisation to run the most valuable cases first. Lean on
AI assistants to draft missing edge-case tests and keep coverage current.
Expected impact
• 10-15% increase in unit/integration coverage
• 5-10% drop in production-defect density
• 5-10% increase in critical bugs caught pre-prod
• 10-15% reduction in test-execution time
• 5-10% increase in edge cases identified
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 37
ClearRoute Limited
Environments
THE BIG ISSUES
Everyone agrees ephemeral, immutable infrastructure is the goal but execution
at scale is hard. Legacy estates block true ephemerality, leaving long-lived,
overburdened, shared stacks that drift from production and queue behind Ops tickets.
HOW TEAMS ARE SOLVING NOW
• Cloud platforms (AWS, Azure, GCP) give teams elastic, pay-as-you-go capacity,
swapping CapEx for on-demand scale and instant experimentation.
• Infrastructure-as-Code tools (Terraform, CloudFormation) encode environments
into version-controlled files, driving repeatability, peer review and push-button
rebuilds.
• Container orchestration (Kubernetes) automates deployment, scaling and
self-healing of containerised workloads, squeezing more from hardware and
keeping apps resilient.
ACTIONS FOR THE YEAR AHEAD
Pursue intelligent, predictive environment management. Use AI to forecast demand
and pre-provision and de-provision capacity. Embed anomaly-detection in IaC
pipelines so drift auto-reconciles. Provide self-service portals so developers can spin
up, clone, or nuke environments fast.
Expected impact
• 10–15% cut in wasted cloud spend
• 20–25% faster environment provisioning
• 15–20% reduction in manual environment management
• 5–10% drop in environment-related incidents
38
State of the Route to Live Report 2025
Deployment
THE BIG ISSUES
CI is mature but release orchestration lags. Many teams still rely on scripted
playbooks with manual checkpoints, so every push carries downtime risk and
rollback drama. The gap between “code merged” and “feature live” saps confidence,
slows cadence, and leaves product teams afraid to press deploy on a Friday.
HOW TEAMS ARE SOLVING NOW
• Deployment tools (like Ansible, Chef and Puppet) automate deployment,
trimming manual errors and enforcing consistent configs. It’s the automation
backbone for reliable, repeatable deployment.
• Emerging release-management platforms orchestrate the holistic
release process, adding visibility and workflow control and reducing risk
and complexity.
• Feature flags decouple code deployment from feature release, enabling
dark launches, A/B testing and rapid rollbacks without a full redeploy.
Risk down; control up.
ACTIONS FOR THE YEAR AHEAD
Move towards intelligent release control. Add AI risk-scoring that inspects changes
and predicts blast radius. Let AI-driven canary/blue-green logic scale trac on
live telemetry and auto-rollback. Surface deploy and release metrics into shared
dashboards so everyone sees the same real-time picture.
Expected impact
• 5–10% increase in deployment success rate
• 10–15% reduced deployment lead-time
• 10–15% faster MTTR for deployment failures
• 5–10% fewer user-reported issues post-release
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 39
ClearRoute Limited
Monitoring and observability
THE BIG ISSUES
Dashboards and alerts are table-stakes but most teams still fight fires reactively.
Mountains of logs and metrics flood Ops, but few signals predict trouble before
users feel it. MTTR stalls, and root-cause hunts eat sprint time.
HOW TEAMS ARE SOLVING NOW
• APM tools (like Datadog, New Relic, Dynatrace) give real-time visibility into
performance, identify bottlenecks, and raise proactive threshold alerts, giving
engineers proactive insight into prod behaviour and UX.
• Log aggregation and analysis stacks (like ELK or Splunk) centralise event data so
teams can search errors and patterns across services, giving critical insights into
app behaviour and troubleshooting.
ACTIONS FOR THE YEAR AHEAD
Evolve to intelligent AI-powered observability. Plug-in AI anomaly detectors to predict
failures and pinpoint root causes. Enable self-healing run-books that auto-trigger when
confidence is high. Expose a single, cross-service view so product, Ops and engineering
see the same leading indicators.
Expected impact
• 70–80% of incidents predicted proactively
• 15–20% faster MTTR
• 10–15 % faster MTTD
• 0.5–1 % uplift in application uptime/availability
40
State of the Route to Live Report 2025
A SEISMIC SHIFT
IS COMING
THE NEXT 12 MONTHS AREN’T ABOUT
SPRINKLING AI WIDGETS. THEY’RE
ABOUT TAKING A THOUGHTFUL,
STRATEGIC APPROACH; DEFINING
PROBLEM SPACES CLEARLY; AND THEN
CREATING SAFE-TO-FAIL USE CASES
THAT PROVE VALUE AND SECURE BUY-IN,
MAKING A REAL IMPACT FAST.
WE’LL SEE TEAMS WEAVE INTELLIGENT
AGENTS AND POLICY GUARDRAILS INTO
EVERY STAGE, SO VELOCITY COMPOUNDS,
RISK FALLS, AND DEVELOPERS FINALLY
SPEND THEIR DAYS BUILDING VALUE
RATHER THAN BATTLING THE PLUMBING.
Overview
Running the
Numbers
Expert
Insights
Impact
of AI
12 Month
Roadmap
Where
Next
41
ClearRoute Limited
ROUTE TO LIVE
12 MONTH INITIATIVE PRIORITISATION MATRIX
RtL Phase (Colour)
AI Integration (Size)
High AI Driven
Medium AI Augmented
Low AI Standard
DesignBuild Deployment Development Environments
Ideation &
Planning Monitoring &
Observability Platform Secure Te s t
Low High
Expected Business Value / Strategic Impact
HighLow Investment / Effort
Time Sinks / Reconsider
Low Value, High Eort
Big Bets
High Value, High Eort
Fill-ins / Incrementals
Medium Value, Medium Eort
Quick Wins
High Value, Low Eort
Hasty Microservices
Migration
AI Dev: Code
Gen & Validate
Refine Test
Strategy Doc
Header-base
Routing (Mocks)
Secret Vault Setup
Pilot DevSecOps
Agentic Use Cases
Focussed API
Testing Adoption
Define Core
Golden Paths
AI Build:
Analyse & Select
Optimise Key
CI Pipeline
AI Assisted
Code Review
AI Ideation:
Feasibility Scan
AI Design:
Rapid POCs
Automated
Release/Gov Docs
Ephemeral Test
Environments
Launch Internal Dev
Platform (MPV)
AI Testing: Data & Prioritise
Observable, Traceable Agentic Platform
AI M&O:
Predict & Heal
AI Deploy:
Risk & Canary AI Envs: Forecast
& Self-Serve
Manual Regression
for All Changes
Over-customising
COTS Tools
42
State of the Route to Live Report 2025
WHERE
NEXT
43
ClearRoute Limited
THE DATA
IS BLUNT.
Most large organisations still bleed months,
and millions, from the Route to Live. Manual
interventions add weeks, pipelines are
unreliable, bloated QA catches bugs but fuels
rework, and shared environments keep Ops
buried in tickets.
Yet the same dataset shows a clear
escape route.
Teams that
• automate extensively,
• embed quality engineering,
• decouple everything, and
• attack developer-experience head-on
ship better products, faster and safer.
AI WIDENS
THE GAP.
Early adopters are moving towards AI
becoming a true strategic partner for the
Route to Live, with collaborative, predictive
and adaptive capabilities that free developers
to eliminate toil and repetition and focus
where it counts.
This more intelligent, more adaptive,
and more autonomous Route to Live will
consistently deliver faster, more reliable
products that drive better business outcomes.
Lagging estates - with brittle pipelines and
siloed teams - can’t simply bolt that future on.
The AI preparedness gap sees organisations
grappling with foundational RtL weaknesses
that mean they’ll struggle to scale advanced
AI - and the generational opportunities that
come with that.
The software delivery performance
gap is widening
44
State of the Route to Live Report 2025
AI DEMANDS
ACTION. FROM
EVERYONE.
For elite performers, the AI opportunity is
obvious and achievable. But AI should be
on every organisation’s agenda, even if
you don’t feel “ready” for this future.
Optimising foundational software delivery
practices and strategically advancing AI
maturity must come hand-in-hand, as
mutually reinforcing priorities.
For organisations that sleep on AI, the
performance divide is about to become
existential. In a world of autonomous,
self-healing pipelines and agentic
operations, shipping fast is no longer
an edge. It’s survival.
But for organisations that act now to
bake-in intelligence in a strategic,
thoughtful way that mitigates risk, there’s
a generational opportunity to move back
into the fast lane.
SO WHERE
NEXT?
1. Map your lifecycle and identify where
initial AI capabilities oer most impact.
Pinpoint the real constraints and
opportunities, not the surface-level pains.
2. Build the right foundations to harness
advanced AI.
Replace paperwork with policy-as-code;
replace eort with automation.
3. Free your teams to engage eectively
with sophisticated AI.
Treat DevEx as a product; cut cognitive
load before you bolt-on more tooling.
4. Turn incremental wins into compounding
advantage with the QCE framework,
progressing towards the intelligent and
adaptive RtL of the future.
5.
6.
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 45
ClearRoute Limited
A
u
t
o
m
a
t
i
o
n
&
G
o
v
e
r
n
a
n
c
e
R
e
l
e
a
s
e
A
u
t
o
m
a
t
i
o
n
O
b
s
e
r
v
a
b
i
l
i
t
y
M
o
n
i
t
o
r
i
n
g
&
G
o
l
d
e
n
P
a
t
h
M
e
a
s
u
r
i
n
g
E
n
g
i
n
e
e
r
i
n
g
D
e
v
e
l
o
p
e
r
H
u
b
T
e
s
t
F
i
r
s
t
C
o
n
t
i
n
u
o
u
s
T
e
s
t
i
n
g
T
e
s
t
S
t
r
a
t
e
g
y
S
E
C
U
R
I
T
Y
·
C
U
L
T
U
R
E
·
A
I
A
G
E
N
T
S
E
C
U
R
I
T
Y
·
C
U
L
T
U
R
E
·
A
I
A
G
E
N
T
S
E
C
U
R
I
T
Y
·
C
U
L
T
U
R
E
·
A
I
A
G
E
N
T
C
L
O
U
D
P
L
A
T
F
O
R
M
D
E
V
E
L
O
P
E
R
E
X
P
E
R
I
E
N
C
E
Q
U
A
L
I
T
Y
E
N
G
I
N
E
E
R
I
N
G
QUALITY
CLOUD
ENGINEER
QCE
46
State of the Route to Live Report 2025
Overview
Running the
Numbers
Impact
of AI
Expert
Insights
12 Month
Roadmap
Where
Next 47
ClearRoute Limited
Kouros Aliabadi
Managing Director, ClearRoute UK
kouros.aliabadi@clearroute.co.uk
Sarndeep Nijjar
Chief Technology Ocer
sarndeep.nijjar@clearroute.co.uk
Matt Lowry
Global Head of Technology & Solutions
matt.lowry@clearroute.co.uk
Justin Wilkin
Global Head of Engineering
justin.wilkin@clearroute.io
Jason Man
Global Head of Consulting
jason.man@clearroute.co.uk
ABOUT CLEARROUTE
CLEARROUTE IS A GLOBAL PLATFORM
ENGINEERING CONSULTANCY THAT HELPS
ENTERPRISES RELIABLY BRING HIGH-IMPACT
DIGITAL PRODUCTS TO MARKET FASTER,
CHEAPER, AND SAFER.
Report Authors
48
This report is provided for informational purposes only and
does not constitute financial, legal, or professional advice.
All views expressed are those of ClearRoute Limited unless
otherwise stated and are based on data available at the
time of publication.
All trademarks, logos, and brand names are the property
of their respective owners and are used in this report for
identification purposes only. Their inclusion does not imply
endorsement.
Where applicable, data has been anonymised or aggregated
to protect individual privacy and comply with relevant data
protection regulations, including GDPR. No personally
identifiable information (PII) is included.
All third-party data, statistics, and visuals cited in this
report are credited to their original sources. For detailed
benchmarking methodologies, please contact ClearRoute
Limited directly.
Any forward-looking statements are hypothetical and
subject to change based on future developments. They
should not be interpreted as guarantees of future outcomes.
This document is copyright © ClearRoute Limited, 2025.
All rights reserved. No part of this publication may be
reproduced or distributed without prior written consent
from ClearRoute.
For further information, legal questions, or to request
accessibility support, please contact: info@clearroute.com
or visit www.clearroute.io.
LONDON, UK
WeWork Holborn
1 Waterhouse Square, London
EC1N 2ST
NEW YORK, USA
WeWork SoHo
524 Broadway, New York
NY 10012
SOFIA, BULGARIA
myFlex Megapark
115G Tsarigradsko Shose Blvd, fl 9
1784 Sofia
SYDNEY, AUSTRALIA
WeWork ANZ Tower
161 Castlereagh Street, Sydney
NSW 2000
PUNE, INDIA
WeWork Raheja Woods
8, Central Avenue, Kalyani Nagar, Pune,
MH 411006
BENGALURU, INDIA
WeWork Roshni Tech Hub
Marathahalli Main Rd, Bengaluru
KN 560037
www.clearroute.io
