Executive Summary.
This study shows: the DACH region is on the move. But not evenly everywhere. The findings from 380 structured interviews with managing directors, owners and top executives between September and November 2025 paint a nuanced picture — one that differs markedly from what typical vendors' marketing materials claim.
Finding 1: The silent majority is already experimenting. 42 percent of respondents had at least one AI agent in productive use by the end of 2025. Another 29 percent are in a pilot phase. That means more than two-thirds of all surveyed companies are no longer just watching — they're acting. What makes this number interesting: only 18 percent talk about it publicly. Most are working on it quietly.
Finding 2: The biggest hurdle isn't technology — it's clarity. 64 percent of respondents named a lack of internal clarity about use cases as the main barrier — not data protection, not cost, not skill gaps. If you don't know what to automate, you don't move forward. This insight has consequences for the consulting landscape: whoever helps create clarity, wins.
Finding 3: Mid-sized companies are the quiet pioneers. Contrary to expectations, it's not large corporations but companies with 25 to 100 employees that are furthest along. They combine the agility of small companies with the resources for structured implementation. Corporations get stuck in approval loops, solo founders often end up testing tools without a strategy.
Finding 4: Budget growth is explosive. The average AI agent budget per company, by self-report, will rise from 14,300 euros in 2025 to 38,700 euros in 2026 — an increase of 170 percent. More important: the spread is widening. Front-runners are investing six figures, laggards are only just getting started with 5,000 euros. The gap is opening up.
Finding 5: Trust is the new currency. 78 percent of respondents say trust in the vendor matters more than technical specifications. Whoever can show reference customers in the same industry wins — whoever argues only with features, loses. That fundamentally changes the rules of the game in the vendor market.
2026 will be the year the gap between pioneers and laggards becomes unbridgeable. Whoever hasn't seriously started by the end of 2025 will, by the end of 2026, be competing against rivals running two or three production agents. The lead won't come from technology — it will come from organizational learning, which takes time.
Methodology & Sample.
Trust comes from transparency. Here's exactly how this study came together — including where it hits its limits.
How we asked
Between September 12 and November 28, 2025, we surveyed 380 managing directors, owners and C-level executives in structured interviews. 62 percent of interviews were conducted by phone, 24 percent by video call and 14 percent in writing (via online questionnaire). The average duration of the direct interviews was 24 minutes. The questionnaire covered 31 questions across five topic blocks: status, use cases, hurdles, success factors, plans.
Who answered
The sample is deliberately focused on operational decision-makers. 71 percent of respondents were managing directors or owners, 18 percent C-level (CTO, COO, CIO), 11 percent operational management with a direct reporting line. We deliberately did not survey pure IT-department staff — the study is meant to capture the strategic view, not the technical implementation perspective.
Geographic distribution
Germany 76 percent (290 respondents), Austria 14 percent (54), Switzerland 10 percent (36). This slightly overrepresents Austria and underrepresents Switzerland — an effect of our network. We weighted the regional analyses accordingly.
Industry distribution
The sample covers twelve industries. Key areas: B2B services (24%), manufacturing (18%), software/IT (15%), retail/e-commerce (11%), finance/banking (8%), consulting (7%), other industries (17%). This distribution isn't representative of the DACH economy as a whole, but it's broad enough for meaningful cross-industry comparisons.
What this data isn't
Three important caveats: First, this is not a representative survey. We set no quotas for company size or region. Second, all data is self-reported. We could not externally verify the figures given to us. Third, we drew on our own network plus its referrals. That creates a positive-selection bias: our sample tends to be more open to AI topics than the DACH economy as a whole.
With these caveats in mind: this is the most robust data we can produce. It shows tendencies — not absolute truths. If you can work with that nuance, what follows are the most solid insights we know of from the DACH region.
Adoption Stages by Industry.
We assigned each of the 380 companies to one of five adoption stages — from "no contact with the topic" to "strategically embedded." Here are the aggregated findings, broken down by the most important industries in our sample.
What the distribution tells us
The software industry leads, unsurprisingly — those who build AI use AI. But the gap to second-place consulting is smaller than expected. Marketing agencies and consultancies have long since caught up with software, because their core business (writing, research, structuring) benefits directly from agents.
The biggest surprise is finance: 31 percent sounds low, but it's remarkably high for such a heavily regulated industry. Banks and financial service providers have caught up massively over the past twelve months — mainly internally (compliance pre-checks, reporting automation), while customer-facing applications continue to be handled restrictively.
The other end is just as telling: construction and trades at only 9 percent. This is the biggest untapped lever in the DACH economy. Any vendor who manages to crack this industry — through radically simplified onboarding and industry-specific language — is sitting on a huge market.
Distribution by adoption stage
Across all industries, the 380 companies break down across our five stages as follows:
The largest group is Stage 4 at 31 percent — companies running at least one production agent but without a strategic roadmap yet. This group is the main market for 2026: they know it works, but they're still looking for the path to the next stage.
Top 5 Hurdles to Adoption.
We asked: "What's currently stopping you from using AI agents more?" Multiple answers were allowed. The order of the responses is surprising — and debunks a few myths about the main brakes on DACH adoption.
What this ranking really means
Most market reports and vendor marketing materials focus on hurdle 3 (data protection) and hurdle 4 (ROI). Our data shows, however: the real killers are hurdles 1 and 2 — lack of clarity and lack of skills. That has two implications.
First, for companies: if you're stuck on one of these hurdles, solve clarity first, then skills. Data protection and ROI are downstream concerns. If you get into GDPR discussions before it's clear what you actually want to automate, you're optimizing the excuse instead of the approach.
Second, for vendors: whoever produces clarity, wins. Consulting and audit processes matter more than technology demos. That's a fundamental shift in what vendors should actually be selling.
Among the pioneers (stage 4–5), the hurdles are different. For these companies, the top hurdles are no longer "clarity," but "scaling" and "multi-agent complexity." That shows: hurdles change with every stage. Solving stage 1 doesn't mean you've solved stage 4.
Success Factors of the Pioneers.
We took a closer look at the 42 companies in Stage 5 (strategically embedded). What do these pioneers do differently from the average? Six factors stand out.
What these factors have in common
On closer inspection it becomes clear: four of the six success factors are organizational, not technical. Ownership, backing, scope, metrics — these are all management decisions, not IT decisions. Pioneers aren't successful because they have better tools — they're successful because they treat the initiative with the discipline of a normal business project.
This insight is both liberating and revealing: liberating, because the success factors are reproducible without hiring tech geniuses. Revealing, because it shows that many companies that fail, fail because they work without discipline — not because they have too little money or talent.
The pioneers have an unfair advantage. But that advantage isn't a secret and it isn't magic — it's discipline. Any company that's serious about it can copy that.
Budgets & Investments.
We asked: "How much did you spend on AI agents in 2025? What are you planning for 2026?" The answers paint a dramatic growth picture — and show where the gap between pioneers and laggards is opening up.
The gap is widening
The average doesn't tell the whole story. Looking at the distribution, it becomes clear: the gap between pioneers and laggards is opening up dramatically.
In our data we see the following spread for 2026: the bottom quartile (i.e. the 25 percent with the lowest budgets) is planning 5,000 to 12,000 euros. The top quartile (the top 25 percent) is planning 60,000 to over 200,000 euros. The ratio between the bottom and top quartile has widened from 1:7 (2025) to 1:18 (2026 planned).
This isn't just a money question. Behind these numbers: pioneers learn faster, scale more aggressively, accumulate more experience. In two years, the gap will no longer just be financially unbridgeable — it will also be organizationally too big. Whoever doesn't invest seriously in 2026 won't be able to catch up in 2027.
What the money is spent on
Broken down by budget line, companies spend their AI agent budget as follows:
Notably: external consulting is the biggest line item at 38 percent. That fits our finding from Chapter 04 — clarity and skills are the main bottlenecks, and companies bring in external help for exactly that. LLM API costs are surprisingly low (22%), which shows: the models aren't the cost driver. Setup and implementation are more expensive than operation.
DE / AT / CH Compared.
The DACH region is often treated as a single market — but it isn't. Our data shows three clearly distinct adoption profiles that matter for vendors and investors.
Concrete differences, in numbers
A few data points that sharpen the profile: time to first production agent is a median of 87 days in Germany, 52 days in Austria, 124 days in Switzerland. Average setup budget in DE: 14,500 euros, in AT: 11,200 euros, in CH: 23,800 euros. Share with multi-agent setups in DE: 17 percent, AT: 24 percent, CH: 11 percent.
These numbers are explained by the respective business cultures. Austrian companies tend to be more action-oriented and smaller, hence faster. Swiss companies have higher margins, more resources, but also more conservative cultures. Germany sits in the middle — cautious, but thorough.
What this means for vendors
Anyone expanding into DACH shouldn't try to serve all three markets the same way. Successful vendors usually start in one market, perfect their approach there, and then move on sequentially. The most commonly observed order: AT → DE → CH. Austrian speed allows fast iteration, the German market brings scale, Switzerland cements the margins.
What 2026 Will Really Bring.
Based on the survey — specifically the "What are you planning for 2026?" answers — and our own observations from 150+ production setups, here are four predictions we can make with high confidence.
Prediction 1: Adoption rate doubles
71 percent of respondents plan to have at least one production agent by the end of 2026. We assume an effective realization rate of 60 to 70 percent of these plans — meaning a real adoption rate of 65 to 80 percent within our sample group. Extrapolated to the DACH economy as a whole, we expect around 35 to 45 percent adoption by the end of 2026.
Prediction 2: Multi-agent becomes a mass phenomenon
Currently, 17 percent of production setups have more than one agent. Among pioneers (Stage 5), it's 89 percent. We expect that by the end of 2026, more than half of all production setups will be multi-agent architectures — simply because pioneer companies are expanding their setups and new setups are already being planned as multi-agent from the start.
Prediction 3: Consolidation in the vendor market
The number of vendors exploded in 2025. 2026 will be the first year of consolidation — many small vendors without differentiation will disappear from the competition. Three clusters will remain: global platform players (OpenAI, Anthropic, Google), industry-specific specialists, and local DACH consultancies with real industry know-how. Generic "AI agency" business models will have a hard time in 2026.
Prediction 4: Regulatory maturity
The EU AI Act is coming into force in stages. 2026 will be the first year in which compliance requirements for AI agents become concrete — especially for "high-risk" applications in finance, health and HR. Companies operating without audit trails and documentation will have to retrofit in 2027. Whoever factors this in now has an advantage.
2026 is the consolidation year. Any company that starts now still has a chance to ride the wave. Whoever waits will enter a market that's already consolidated — one where the easy ROI wins have been picked over. Any vendor active in this space has to choose between volume platform and specialist consulting. The middle is disappearing.
Conclusion & Recommendations.
The DACH region is no longer in the "should we use AI agents?" phase. It's in the "how do we do this right?" phase. Anyone still in the first phase should hurry — the second phase won't wait.
Here are the three most important recommendations from 380 interviews:
Recommendation 1: Clarity before technology
The top hurdle isn't technology, it's a lack of internal clarity. Before you think about tools, models or vendors, clarify: which process is actually costing you time? Who would be the owner? What would "successful" mean? These three questions solve 64 percent of all problems before they arise.
Recommendation 2: Copy the pioneers, don't reinvent
The six success factors of the pioneers from Chapter 05 are reproducible. You don't need a tech genius, a million-euro budget, or your own AI department. You need discipline in applying six simple principles. Most failed setups didn't fail because of technology — they failed because of at least one of these six principles.
Recommendation 3: Start with a small scope
76 percent of pioneers started with exactly one task. Not three. Not the most important one. The clearest one. If you take that to heart, you'll be in production after 30 days. If you ignore it, you'll be frustrated after 30 days. The choice is yours.
In 2026, the DACH economy won't be deciding whether AI agents catch on. It has already decided that. All it's deciding now is who comes along — and who gets left behind.
If you want to use this study to place your own situation — book a 30-minute strategy call. We'll compare your position against the benchmarks from this study and give you an honest assessment of where you stand — and which two or three next steps would make sense.
