State of Integration & AI 2026 — the European benchmark

Thesis

The 2026 benchmark for integration & AI maturity in Europe. Independent research with 600+ senior IT leaders across Europe — what's working, what's stalling, and the architecture move that separates the leaders from the rest.

Headline numbers

€10.7M

Annual cost of unnecessary manual work. Mean per organization with 1,000 employees, European sample.

7.6 hrs

Per employee, per week, lost to manual data work. Almost a full working day, every week.

34%

Of organizations have AI in production. Only 7% have deployed it widely; 33% are still piloting.

25.5%

Of AI projects deliver measurable P&L impact. Three in four still fail to move a number that matters.

36%

Cite integration complexity as a #1 AI blocker. Tied with skills gaps and security as the top barrier.

64%

Want centralized AI & integration governance. Yet only 12% have built integration as the governance layer.

79 hrs

Person-hours to deliver an average integration. Two working weeks per connection, before anything ships.

43%

Faster project delivery for integration-first organizations. Compared to peers — same teams, same budgets.

Executive summary

2026 is the year European organizations stop asking whether AI works and start asking why theirs doesn't. The gap between leaders and laggards is widening — and the dividing line is no longer ambition or budget. It is architecture. With the EU AI Act, GDPR and CSRD now reshaping the operating environment, the era of the experiment is closing. The era of integration-led, governed, in-production AI is opening.

1. Manual work is the silent line item. European organizations are losing €10.7M per year (per 1,000 employees) to manual work that integration could eliminate. The cost is invisible because no department owns it.
2. Most AI is still pre-production. Only 34% have AI in production and just 7% have deployed it widely. Just 25.5% of AI projects deliver measurable P&L impact. The bottleneck is not the model — it is everything around the model.
3. Integration is a top AI blocker, tied with skills and security. 36% of IT leaders name integration complexity as their #1 obstacle to AI — tied with skills gaps and security at 36% each. Most organizations face all three simultaneously.
4. Governance urgency outpaces governance infrastructure. 64% want centralized governance for AI and integration. Only 12% have actually built integration as that governance layer. The gap is the regulatory exposure.
5. Sovereignty is now a procurement filter. 61% of European leaders treat EU data sovereignty as already critical in operations. The EU AI Act, GDPR and CSRD have made the architecture question a regulatory one.

Key insights

• The manual labor tax is bigger than anyone admits. When you total up the unnecessary copy-paste work happening across European IT estates, you get a €10.7M annual cost for a 1,000-employee organization. It doesn't show up on a single line item, which is exactly why it persists. (€10.7M / year, per 1,000-employee organization)
• Two-thirds of Europe is stuck before production. Only 34% have AI in production. 33% are piloting and 30% are still investigating. Just 7% have deployed AI widely. The pattern is consistent across markets and sectors. Pilots are easy. Shipping is the hard part. (Only 34% in production, 7% widely deployed)
• Three out of four AI projects fail to move a number. Just 25.5% of AI initiatives deliver measurable P&L impact. The other 74.5% produce demos, dashboards and deck slides — but no euros. The honest reading: most AI investment is still pre-revenue. (25.5% measurable P&L impact)
• Integration is the AI bottleneck — tied at the top with skills and security. Asked to name the single biggest blocker to AI in production, 36% of leaders name integration complexity — tied with skills gaps and security at 36% each. Most organizations face all three at once. AI without plumbing is just a chatbot. (36% — tied at the top with skills and security)
• The governance pendulum is swinging back to the center. After a decade of 'let a thousand experiments bloom', 64% of leaders now want centralized governance for AI and integration. But only 12% have actually built integration as a governance and control layer. The urgency exists; the infrastructure does not. (64% want centralized governance — only 12% have built it)
• Integration-first organizations ship 43% faster. When you split the sample by who built integration as a deliberate design decision vs. who chased AI first, the integration-first cohort delivers 43% faster, with measurably higher AI success rates. Same teams. Same budgets. Different sequence. (+43% project delivery speed)
• The flywheel is real, and most aren't on it yet. Clean integration → clean data → working AI → recovered time reinvested into more AI. 40% would reinvest recovered time into accelerating AI adoption — the top answer. Cost or headcount reduction came last (28%). (40% would reinvest recovered time into more AI)
• European sovereignty is moving from talking point to procurement filter. 61% of organizations say EU data sovereignty is already critical in their operations. The EU AI Act, GDPR and CSRD have made architecture decisions a regulatory question, not just a technical one. (61% say sovereignty is already critical)

Country briefings

Denmark — The clear deployment leader

Denmark is further into production AI than any other market in the survey, with 17% widely deployed (vs 7% European average) and the highest AI success rate at 32.6%. The integration-first discipline came first; the deployment lead followed. The barrier has shifted from connectivity to AI transparency.

• AI in production or widely deployed: 44%
• AI widely deployed: 17%
• Integration-first by design: 38%
• Mean AI project success rate: 32.6%
• Annual manual-work cost (1,000 employees): €14.08M
• Sovereignty critical to operations: 73%

"Denmark proves the playbook works at scale. The barrier elsewhere is not the data — it's the willingness to recentralize."

Finland — Caught in the pilot trap

Finland has the most AI-first orientation in the survey but only 2% have deployed AI widely. The most consolidated integration platform landscape in Europe (just 3% run more than five platforms) is an underused asset. Cost (40%), not integration, is cited as the top AI blocker — which explains why integration foundations aren't being built first.

• Still in piloting or PoC: 45%
• AI in production or widely deployed: 28%
• AI widely deployed: 2%
• Top AI blocker — cost: 40%
• Run >5 integration platforms (lowest): 3%
• Pursuing AI with no integration plans (highest): 8%

"It feels that, in Finland, having one integration platform has already been the standard for a longer time." — Jenni Aliisa Koskinen, Integration Platform Lead, GlobalConnect

Germany — High burden, structural caution

German workers carry the heaviest individual manual burden in the survey at 8.5 hours/week — 102 minutes of every working day. The integration posture is the weakest of any country: 16% integration-first (lowest) and 8% with no integration strategy at all. But governance urgency is high (69%) and the appetite to reinvest is real.

• Mean manual work per week (highest): 8.5 hrs
• Integration as #1 AI blocker (highest): 40%
• AI in production rate: 33%
• Integration-first by design (lowest): 16%
• No integration strategy at all: 8%
• Annual manual-work cost (1,000 employees): €11.43M

"Germany has strong process discipline. But in many companies, that has not yet translated into true end-to-end digital execution. The cultural strengths — quality orientation, precision, strong awareness of risk and compliance — become a problem when they turn into over-caution, perfectionism or decision paralysis." — Prof. Dr. Moritz E. Behm, Fresenius University Munich

Netherlands — The high-stakes paradox

Dutch workers spend the least time on manual tasks (6.6 hrs/week) yet 97% report human error risk as a negative impact — the highest in the survey. The Netherlands has the lowest integration-first rate (12%), the lowest AI project success rate (21%) and the highest internal resistance to AI (38%). The easy automation is done; what remains is high-stakes and harder.

• Mean manual work per week (lowest): 6.6 hrs
• Manual work increases human error risk (highest): 97%
• Integration-first by design (lowest): 12%
• Mean AI project success rate (lowest): 21%
• Internal resistance to AI (highest): 38%
• Annual manual-work cost (1,000 employees): €7.93M

"The Netherlands has a strong innovation culture, but we also have a tradition of consensus-based decision-making: the 'Polder Model.' We are very good at experimenting. We like to do proofs of concept, to try things out. But going all-in on a strategic direction without full alignment is not typical Dutch, and that can slow down large-scale transformation." — Robert Daniëls, Managing Director, Harmony

Norway — High ambition, fragmented architecture

Norway has the most fragmented integration landscape in Europe (21% run more than five platforms) and the highest manual-work proportion (59% spend 7+ hours/week). And yet it leads the survey on combined time savings from automation (22.3 hrs/employee/month) and has the highest rate of organizations using integration as a governance layer (18%). The capacity is there; the architecture is not.

• Run >5 integration platforms (highest): 21%
• Spend 7+ hrs/week on manual tasks: 59%
• AI in production rate: 30%
• Lack of integration governance as #1 deployed-AI barrier: 45%
• Combined monthly time savings (highest): 22.3 hrs
• Annual manual-work cost (1,000 employees): €12.59M

"Norway's strength — local autonomy — is also the source of its biggest integration tax."

Sweden — The balanced operator

Sweden runs the highest parallel AI-and-integration adoption in the survey (43%), matches the European integration-first rate exactly (23%) and treats integration as strategic discipline. But the AI success rate (21.2%) lags the European average — likely the execution friction of doing both at once without completing either. Plus 22% still write custom point-to-point integration code (highest in survey).

• AI in production or widely deployed: 38%
• Working on AI + integration in parallel (highest): 43%
• Integration-first by design: 23%
• Mean AI project success rate: 21.2%
• Manual point-to-point coding (highest): 22%
• Annual manual-work cost (1,000 employees): €8.26M

"The whole transformation journey is troublesome when you have an unknown integration landscape — it's the kind of thing that takes the longest to figure out and change. If people continue with whatever quick fix they did in the past, maybe it'll be okay for that particular time, but we're just pushing the technical debt ahead of us." — Amanda Lindén, Head of Common Systems IT, GlobalConnect

Sector briefings

• Public Sector — Furthest behind, biggest mandate. Zero public-sector organizations have AI widely deployed. 49% cite 'no clear business case' as the top AI barrier — nearly double the European average. Only 7% use integration as a governance layer, while AI piloting and planning combined account for 82% of cases. AI in production 14%; integration-first 21%; success rate 17%.
• Energy & Utilities — Governance urgency, deployment paradox. Energy leads on centralized-governance urgency (77%, highest in the survey) — yet zero organizations have AI widely deployed and 13% have no AI plans at all. The lowest manual-work burden in the survey (5.3 hrs/week). High awareness without a path to action. AI in production 26%; integration-first 38%; success rate 24%.
• Manufacturing — ERP gravity meets AI ambition. Manufacturing has the most legacy gravity and the most concrete AI use cases. Supply-chain data management is the second-largest bottleneck at 28%. The winners are the ones treating their ERP as one node in an integration mesh, not the center of the universe. AI in production 29%; integration-first 30%; success rate 26%.
• Healthcare — The integration gap is the AI gap. Healthcare has the lowest AI project success rate in the survey at 17.5%. 50% of respondents name integration challenges as the primary barrier to AI deployment. Compliance pressure is forcing clearer governance — the leaders treat that as a velocity feature, not a brake. AI in production 26%; integration-first 28%; success rate 17.5%.
• Financial Services — Highest automation ambition among regulated sectors. Financial Services expects a 51.4% mean reduction in manual work through full automation — the highest ambition of any regulated sector. The challenge is now consolidation, not capability. AI in production 31%; integration-first 32%; success rate 27%.
• Retail — Highest AI success rate, highest wide-deployment rate. Retail has the highest AI project success rate in the survey (35.2%) and the highest proportion of organizations with AI widely deployed (22%). Clear customer-facing ROI metrics demand connected data — the integration discipline followed the commercial logic. AI in production 35%; integration-first 32%; success rate 35.2%.
• Professional Services — Knowledge work, kept manual. Professional services firms report the highest manual hours in the survey at 8.4/week — and 92% say manual work slows decision-making. The opportunity is enormous; the cultural barrier is bigger. AI in production 23%; integration-first 26%; success rate 22%.
• Technology — Builds for itself, but the technical-debt signal is loud. Tech firms have above-average internal AI deployment — but 23% still write custom point-to-point integration code, the highest of any sector. Speed culture is producing the same integration debt the rest of the market is now paying off. AI in production 38%; integration-first 40%; success rate 31%.
• Transport, Logistics & Construction — The AI deployment leader, hidden in plain sight. Transport, Logistics and Construction has the highest AI production rate of any sector at 56%, with only 23% still in proof-of-concept. The reason is operational: integrating vehicle data, logistics systems and supply chains is not optional in this sector. AI in production 56%; integration-first 40%; success rate 30%.

The four phases of automation maturity

1. Manual & reactive — People are the integration. ~22% of organizations sit here today.
2. Piloting & experimenting — AI in proofs of concept, integration considered during development. ~46% of organizations — the largest cohort.
3. Scaling with integration — Flows are governed, observable, reusable. ~25% of organizations have made it here.
4. AI-native & agentic — Integration is the governance and control layer. ~7% — the frontier; 97% factor agents into governance design.

Expert quotes

"It feels that, in Finland, having one integration platform has already been the standard for a longer time." — Jenni Aliisa Koskinen, Integration Platform Lead, GlobalConnect.

"Germany has strong process discipline. But in many companies, that has not yet translated into true end-to-end digital execution. The cultural strengths — quality orientation, precision, strong awareness of risk and compliance — become a problem when they turn into over-caution, perfectionism or decision paralysis." — Prof. Dr. Moritz E. Behm, Fresenius University Munich.

"The Netherlands has a strong innovation culture, but we also have a tradition of consensus-based decision-making: the 'Polder Model.' We are very good at experimenting. We like to do proofs of concept, to try things out. But going all-in on a strategic direction without full alignment is not typical Dutch, and that can slow down large-scale transformation." — Robert Daniëls, Managing Director, Harmony.

"The whole transformation journey is troublesome when you have an unknown integration landscape — it's the kind of thing that takes the longest to figure out and change. If people continue with whatever quick fix they did in the past, maybe it'll be okay for that particular time, but we're just pushing the technical debt ahead of us." — Amanda Lindén, Head of Common Systems IT, GlobalConnect.

"There is a gap between what organizations know they need and what they have actually built. Most leaders will tell you that AI is a strategic priority. Most IT teams will tell you, quietly, that their systems are not ready for it. Both are right." — Jukka Rautio, CEO, Frends.

Methodology

The State of Integration & AI 2026 is based on an independent survey of 611 senior it and business decision-makers at organizations with 201+ employees, conducted by Sapio Research in April 2026. Results weighted by country and sector to reflect the European IT estate population. Contact: research@frends.com.

About Frends

Frends is a European integration platform (iPaaS) headquartered in Finland, used by enterprises and public-sector organizations across Europe to connect systems, automate processes and put AI into production. Frends commissioned and funded this research independently; Sapio Research designed and fielded the survey. Frends is the sponsor of the report, not the subject.