For the past couple of years, “AI strategy” in many German boardrooms meant running a few pilots, watching dashboards light up, and waiting to see what the technology could actually do. That phase is over. The latest wave of Infor’s Enterprise AI Index for Germany tells a clear story: companies are moving AI out of the sandbox and into their core operations, and finding out in the process what it really takes to make it work.
Scaling AI, it turns out, isn’t just a technical challenge. It’s a process challenge. Nearly 30% of respondents say legacy or disconnected systems are slowing their deployments. Around a quarter struggle with data quality or governance issues. And 69% (more than two in three) say that off-the-shelf AI tools simply don’t cut it for their specific industry needs.
You can’t bolt a generic AI model onto a complex manufacturing or distribution operation and expect it to understand your supply chain logic, your compliance requirements, or your production constraints. It doesn’t work that way.
That’s not a loss of faith in AI agents. Quite the opposite. Comfort with autonomous agents operating fully independently rose from 14% to 19% between spring and August. What’s changed is the expectation of how they work. Decision-makers no longer want agents that operate in isolation. They want agents that are woven into their existing processes and connected to their ERP data, governed by clear approval rules, and auditable at every step.
This is precisely the direction Infor has taken with the Velocity Suite. Rather than offering AI as a separate layer that sits on top of existing systems, Velocity Suite embeds industry-specific AI agents directly into Infor’s CloudSuites. An agent working on production planning, for example, doesn’t just analyse data in the abstract, it operates with full awareness of the specific process logic of that manufacturing environment. It understands what a deviation in a production run means in context, what the downstream implications are, and what corrective options exist within the system. No manual handover between tools, no integration gap.
That kind of precision only works if the AI understands the industry it’s operating in. A generic model can flag anomalies. An industry-trained agent can tell you why they matter and what the most logical next step is, within the specific constraints of your production environment.
Infor’s approach with Velocity Suite is built around this distinction. Industry AI Agents, the Agentic Orchestrator, and the Value+ Solutions aren’t designed to replace planners or procurement managers. They’re designed to make those people faster, less reactive, and better informed while keeping governance and traceability intact. Every agent action is registered, monitored, and auditable. For regulated industries, that’s not optional.
Yet only 19% actually have a comprehensive understanding of what AI adoption and integration will cost them long-term. The rest are working from rough estimates, or know the upfront price but have no clear picture of what the running costs will look like a year or two down the line.
This is precisely where a lot of AI projects run into trouble. Companies start with a pilot budget, scale up, and suddenly face consumption charges, integration costs, and consultant fees they didn’t anticipate. Velocity Suite’s approach - bundled, role-based, with clearly defined use case packs organized by industry and process - addresses this directly. Predictable pricing isn’t just a commercial preference; for most German mid-market and large enterprises, it’s what makes the investment case possible in the first place.
The concern that AI would make skilled workers redundant, or push their expertise to the margins, isn’t playing out that way. What’s emerging instead is a more productive split: AI handles the repetitive, data-intensive analysis; human experts focus on judgment calls, exception handling, and the decisions that genuinely require domain knowledge. The Velocity Suite model is built for exactly this kind of working relationship with agents that prepare and recommend, humans that approve and override.
This post is based on findings from the Infor Enterprise AI Adoption Impact Index, conducted by YouGov in Germany. The study surveyed 266 decision-makers in spring 2026 and 258 in August 2026, covering a range of industries including manufacturing, retail, and transportation & distribution.
From pilot to production: the numbers tell the story
Back in spring 2026, roughly one in ten German companies surveyed was still stuck in pilot mode, testing AI without committing to a real rollout. By August, that number had dropped to just 5%. Meanwhile, the share of businesses running AI in prioritized, live business processes jumped from 32% to 45% in the same period.Scaling AI, it turns out, isn’t just a technical challenge. It’s a process challenge. Nearly 30% of respondents say legacy or disconnected systems are slowing their deployments. Around a quarter struggle with data quality or governance issues. And 69% (more than two in three) say that off-the-shelf AI tools simply don’t cut it for their specific industry needs.
You can’t bolt a generic AI model onto a complex manufacturing or distribution operation and expect it to understand your supply chain logic, your compliance requirements, or your production constraints. It doesn’t work that way.
Integration is the new competitive advantage
Ask German decision-makers what they need most from an AI solution to ensure long-term success, and the answer has shifted considerably since spring. ERP and business system integration has jumped from 25 to 36% percent, now sharing the top spot with data security and sovereignty. The fascination with autonomous agents operating freely, without human oversight, has cooled: it dropped from 30% to 21% as a top priority.That’s not a loss of faith in AI agents. Quite the opposite. Comfort with autonomous agents operating fully independently rose from 14% to 19% between spring and August. What’s changed is the expectation of how they work. Decision-makers no longer want agents that operate in isolation. They want agents that are woven into their existing processes and connected to their ERP data, governed by clear approval rules, and auditable at every step.
This is precisely the direction Infor has taken with the Velocity Suite. Rather than offering AI as a separate layer that sits on top of existing systems, Velocity Suite embeds industry-specific AI agents directly into Infor’s CloudSuites. An agent working on production planning, for example, doesn’t just analyse data in the abstract, it operates with full awareness of the specific process logic of that manufacturing environment. It understands what a deviation in a production run means in context, what the downstream implications are, and what corrective options exist within the system. No manual handover between tools, no integration gap.
Why industry-specific AI actually matters
Take production planning. The real value of AI in that environment isn’t producing a smarter dashboard, it’s detecting a likely disruption, say, a material shortage developing on a specific line, before it becomes a missed delivery. And it’s doing that while giving the human planner full control to approve, redirect or override whatever the system recommends.That kind of precision only works if the AI understands the industry it’s operating in. A generic model can flag anomalies. An industry-trained agent can tell you why they matter and what the most logical next step is, within the specific constraints of your production environment.
Infor’s approach with Velocity Suite is built around this distinction. Industry AI Agents, the Agentic Orchestrator, and the Value+ Solutions aren’t designed to replace planners or procurement managers. They’re designed to make those people faster, less reactive, and better informed while keeping governance and traceability intact. Every agent action is registered, monitored, and auditable. For regulated industries, that’s not optional.
The cost question is becoming impossible to ignore
One number from the latest survey stands out: 94% of German business leaders say fixed and predictable costs are important when evaluating AI solutions. That’s near-unanimous.Yet only 19% actually have a comprehensive understanding of what AI adoption and integration will cost them long-term. The rest are working from rough estimates, or know the upfront price but have no clear picture of what the running costs will look like a year or two down the line.
This is precisely where a lot of AI projects run into trouble. Companies start with a pilot budget, scale up, and suddenly face consumption charges, integration costs, and consultant fees they didn’t anticipate. Velocity Suite’s approach - bundled, role-based, with clearly defined use case packs organized by industry and process - addresses this directly. Predictable pricing isn’t just a commercial preference; for most German mid-market and large enterprises, it’s what makes the investment case possible in the first place.
The human factor: better than expected
One finding that rarely gets the attention it deserves is the impact on the people actually doing the work. In Germany, 70 % of surveyed leaders report that AI has had a positive effect on frontline workers in terms of workload, skills development, or the quality of decisions they can make day-to-day. Fewer than 4% report negative effects.The concern that AI would make skilled workers redundant, or push their expertise to the margins, isn’t playing out that way. What’s emerging instead is a more productive split: AI handles the repetitive, data-intensive analysis; human experts focus on judgment calls, exception handling, and the decisions that genuinely require domain knowledge. The Velocity Suite model is built for exactly this kind of working relationship with agents that prepare and recommend, humans that approve and override.
What comes next
The companies that pull ahead over the next 12 months won’t necessarily be those with the biggest AI budgets. They’ll be the ones that solve the integration problem: that connect AI tightly enough to their ERP backbone that it can actually act on real operational data, in real time, within the specific logic of their industry.
That’s a harder problem than running a pilot. But it’s also where the real gains are.
This post is based on findings from the Infor Enterprise AI Adoption Impact Index, conducted by YouGov in Germany. The study surveyed 266 decision-makers in spring 2026 and 258 in August 2026, covering a range of industries including manufacturing, retail, and transportation & distribution.