Trades 2026: What the DIHK digitalization survey reveals about AI in SMEs

The German economy is keeping pace with digitalization, but it is not catching up. That is the sober core message of the 2026 DIHK digitalization survey, which polled almost 5,000 companies across all sectors. The average digitalization score remains at a solid 2.8, but it is stagnating. For the trades sector the findings are especially relevant, because they show a dual reality: AI has arrived, but trades, construction and industry are nowhere near as productive with it as information and communication, or financial services.

Short version: AI is standard in German SMEs in 2026, but a productivity gap opens between digitally mature sectors and the trades, construction and industrial mid-market. Three levers bridge it.

What the DIHK numbers mean for trades in practice

According to DIHK, 78 percent of companies use generative AI to create text, images or code. That sounds like widespread adoption, but it is an average across all sectors. The branch-specific picture is different:

• Hospitality: 62 percent use AI primarily for personalized customer communication.
• Retail: 53 percent with a similar focus on customer communication.
• Financial services: 41 percent use AI intensively for risk analysis.
• Industry, retail, construction and hospitality: mostly rate the productivity effect as moderate.

The strongest productivity effect shows up in sectors where digital processes and data-driven decisions are already established:

In trades, construction and industry the potential is just as large, but implementation is more complex because physical processes, supply chains and on-site customer interactions all need to be integrated. More than one in three companies that already use AI or plan to roll it out within the next three years expect a strong impact on their own productivity. Among companies that already use AI in production today, 41 percent rate the productivity effect as high.

Why trades lag behind in productive AI use

DIHK names three structural reasons that show up again and again in the field:

1. Legal uncertainty remains the biggest challenge in data use. The EU AI Act, GDPR and a growing number of sector-specific requirements mean that many trades companies postpone the move into AI-supported automation until the legal framework is clear. That is understandable, but it hands a head start to competitors that begin early with clearly defined boundaries.
2. Missing digital maturity as a precondition. AI only delivers value if the processes underneath are structured. Often the same CRM database serves as a digital address book for years before it becomes the training and context dataset for automation. Without that maturation, AI remains a toy.
3. Technical hurdles such as data silos, missing infrastructure or software tools that do not talk to each other decrease as digital maturity grows, but in trades they are often the actual brake. A typical HVAC or plumbing company works with three to five different software tools for quotes, job scheduling, materials and invoicing. As long as those silos persist, every AI integration remains patchwork.

Three levers that bridge pilot to production

From the DIHK findings and the typical starting position in trades companies, three levers emerge that make the leap into the productive everyday easier.

Lever 1: start small, measure clearly

The DIHK survey confirms that companies that start with concrete, irritating tasks see productive results faster. Sensible entry points are:

• Email inquiry triage
• Automatic quote generation from CRM data
• Appointment confirmations with callback option
• Follow-up lists for old inquiries

Each of these should deliver a clearly measurable result, so ROI becomes visible in the first weeks.

Lever 2: automate with clear boundaries

An autonomous AI agent that sends quotes to customers on its own sounds tempting, but in trades it is high risk. Trust comes from gradual expansion:

1. The AI drafts the quote.
2. A person in the company reviews and approves it.
3. The customer receives the verified document.

This human-in-the-loop approach combines the speed of AI with the reliability of human oversight. Exactly this bridge between automation and control decides whether a company commits to AI long-term, or whether it switches the pilot back off after a failure.

Lever 3: keep data in-house

For many trades companies data protection is the decisive compromise point. When quotes, customer data and project documents run through non-European AI services, GDPR compliance is hard to secure. Locally operated or European-hosted AI models solve that problem, and in 2026 they are noticeably easier to obtain than they were two years ago. In this context DIHK calls for open interfaces, standards and open source as building blocks of digital sovereignty.

What the EU AI Act means for the workshop floor

Many trades companies ask whether the EU AI Act even applies to them. The answer depends on how deeply AI interferes with the business process:

• Not high risk: Pure text generation for internal notes.
• Potentially high risk: An AI system that creates quotes and sends them directly to customers, with obligations for risk assessment, human oversight and technical documentation.

The real relief comes from architecture decisions that bake those obligations in from the start. A system that requires every AI-generated output to pass through a human approval gate before it is sent out meets the oversight obligation structurally. Documentation becomes part of the approval workflow, not an extra bureaucratic mountain.

Realistic opportunities on the table in 2026

Trades companies that start with AI in 2026 have realistic levers within reach:

• Automate routine tasks: inquiry intake, appointment coordination, follow-up.
• Generate quotes faster: AI drafts based on historical quotes, the master craftsman reviews and sends.
• Retain knowledge in the company: document the know-how that would otherwise be lost when senior staff retire.
• Scale customer communication: stay reachable outside office hours without adding headcount.
• Optimize materials and purchasing: demand forecasts based on past projects.

What makes sense next

The DIHK survey shows: German companies are not falling behind, but they are not catching up either. For trades, 2026 is when we will see whether keeping pace turns into a real leap. The prerequisites are better than ever: mature tools, local hosting options, clear regulatory frameworks. What is missing is not the technology, but the willingness to automate the first three to five processes consistently and to bring the team along.

Trades companies that start now, with clearly defined use cases and human control, gain a competitive advantage that will be hard to catch up within the next two years.

If you want to find out which three to five processes in your operation have the biggest leverage, a short initial conversation is worthwhile. In 30 minutes it is usually possible to assess where the entry point makes sense and what effort realistically stands behind it.