KrambergAI
Whitepaper · 2026 Edition
A decision-making guide for small and mid-sized companies
Local AI in the
SME Sector
Opportunities, limits and prerequisites for secure and cost-effective operation.
How companies can run language models and generative AI on their own infrastructure, in a private cloud or in hybrid architectures, without losing sight of data protection, information security, quality and cost-effectiveness.
KrambergAI GmbH
AI solutions for the SME sector
krambergai.com
As of July 2026
For management, IT, data protection
and information security
GDPR-compliant data protection Made in Germany Vendor-neutral
Contents

What you will find in this whitepaper

The guide moves from the market context through operating models, architecture, security and law to cost-effectiveness, decision aids and three practical examples from the SME sector.

01Management Summary
02Market context: AI adoption in the SME sector
03Five operating models for local AI
04Model, application, knowledge, operations
05Seven sound reasons to go local
06The most common false expectations
07Well-suited applications
08Less suitable applications
09Model selection and testing
10Model sizes and infrastructure
11RAG or fine-tuning
12Architecture of a production-grade AI
13Information security
14Data protection and the AI Act
15An operating role model
16Operations and the model lifecycle
17Cost-effectiveness and TCO
18Decision matrix
19A pilot project in ten steps
20Readiness check
21Case study: technical service
22Case study: construction and projects
23Company Brain and conclusion
24Studies, standards and sources
Guiding principle

Local AI is not a product decision but an operating decision. An installed language model is not yet an enterprise AI. Only a knowledge base, permissions, security, quality and durably owned operations turn it into reliable business value.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 1
01 · Management Summary

Local AI is not a product decision,
it is an operating decision

A growing number of small and mid-sized companies are examining whether to run generative AI on their own infrastructure. The reasons are understandable: sensitive company data should not leave the organization, dependence on individual cloud providers should fall, AI should remain available without an internet connection, usage costs should become more predictable, and company-specific knowledge should be integrated in a controlled way.

A local installation does not solve these requirements automatically. Installing a language model on a server does not yet create a usable enterprise AI. What is also needed is a robust knowledge base, roles and permissions, interfaces, logging, quality testing, security measures, updates and responsible operations.

The six key messages of this whitepaper

1

Local is not automatically better or safer

Running AI in-house shifts responsibility from the provider to your own organization.

2

A smaller model is often enough

What matters is not model size but performance in the specific business process.

3

Current knowledge does not belong in the model

It should be supplied through a controlled knowledge layer such as RAG.

4

Local does not automatically mean GDPR-compliant

Legal basis, purpose limitation, data minimization and data subject rights remain relevant.

5

The hard part begins after the pilot

Models, knowledge, dependencies and security have to be operated permanently.

6

Hybrid is often better than a pure in-house build

Sensitive tasks run locally; compute-intensive or fluctuating tasks run externally, under control.

“Installing a model is comparatively easy. Building a reliable enterprise system is the real task.”

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 2
02 · Market context

AI adoption is rising, the barriers remain operational

Artificial intelligence has arrived in the SME sector, but it is not yet embedded across the board in day-to-day operations. The Federal Statistical Office and the KfW paint a consistent picture, even though their figures are not directly comparable because of differing definitions, periods and survey methods.

26%
of companies with 10 or more employees used AI in 2025 (Destatis). Small: 23%, medium: 36%, large: 57%.
~780,000
SMEs used AI between 2022 and 2024, a share of 20% (KfW). Construction only 8%.
+ Momentum
Destatis: 12% (2023), 20% (2024), 26% (2025). KfW: up from 4% (2016 to 2018) to 20%.

Why companies have not yet started

Among the companies that had examined but not yet deployed AI, the barriers cited in 2025 were:

BarrierShare
Lack of knowledge72%
Uncertainty about the legal implications62%
Data protection and privacy concerns60%
Incompatibility with existing systems45%
Insufficient data availability or data quality44%
Costs too high32%
What this means for decision-makers

The core problem rarely lies in the AI model alone, but in the interplay of technology, company data, accountability and the ability to operate. The relevant question is therefore not “Which language model should we install?” but: “For which process do we need which AI capability, which data may be processed, and who takes durable responsibility?”

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 3
03 · Operating models

What “local AI” actually means

The term “local AI” is often used loosely. For a sound decision, at least five operating models should be distinguished. They differ in control, effort, scalability and suitability for sensitive data.

1. AI on a single workstation

The model runs on a notebook or workstation. Suitable for personal research in approved documents, confidential draft texts, and development and testing. Limits: hard to administer centrally, no shared knowledge base, limited performance, and the risk of uncontrolled point solutions.

2. Central on-premises operation

The model runs on servers in your own building or data center. Suitable for sensitive internal data, stable user groups, high and predictable usage, offline or production environments, and companies with their own IT operations.

3. Dedicated private cloud

The system runs in an isolated cloud environment or on dedicated servers in a data center. Suitable for companies without their own server room, multiple sites, controlled scaling, external operational support, and requirements for EU hosting or defined data locations.

4. Edge AI

The model runs directly on a machine, in a vehicle, at a branch or on a mobile device. Suitable for speech recognition in the field, image analysis at equipment, local assistance without a stable connection, and time-critical evaluation.

5. Hybrid AI

Local models and external AI services are combined through shared control. Internal documents and personal data are processed locally, general market information via external models. A routing layer decides based on the task, the data class and the required model performance.

Recommendation

For many SMEs the hybrid architecture is the most practical target architecture: it combines data sovereignty for sensitive processes with performance headroom for demanding tasks.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 4
04 · Core concepts

Model, application, knowledge and operations

Four terms are frequently confused. Distinguishing them is the prerequisite for realistically assessing effort, value and responsibility.

The AI model

Generates text, analyzes input, classifies content. But it does not automatically possess current company knowledge, access rights to ERP or DMS, awareness of the approved version of a document, traceability, or professional responsibility.

The AI application

Provides the interface and the workflow, for example as an internal knowledge assistant, service assistant, quotation support, site documentation, maintenance assistance or tender analysis.

The knowledge layer

Connects the model to approved data sources: document management, ERP, CRM, ticketing, project files, operating instructions, bills of quantities, maintenance records, rulebooks and product catalogs.

AI operations

Keeps the system usable over time: updates, permissions, logging, quality testing, backups, monitoring, incident handling, version changes, security reviews and support.

“The model produces answers. Only the application, the knowledge and the operations produce business value.”

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 5
05 · Reasons to go local

When local AI offers a genuine advantage

Seven reasons justify local or largely local operation. In each case they should be demonstrable, not merely plausible in principle.

1

Protecting particularly sensitive information

Technical drawings, costings, customer files, personnel and contract documents, price lists. Prerequisite: the local system itself is adequately secured.

2

Operation without a reliable internet connection

Site containers, production halls, workshops, vehicles, ships and port facilities, remote locations, temporary venues.

3

Short and predictable response times

With suitable infrastructure, lower and more predictable response times, for example for voice assistance, machine operation, image inspection or real-time classification.

4

Predictable, sustained usage

With high, stable load, in-house operation can become cost-effective, especially with many short requests or large internal document collections.

5

Integration into isolated systems

Production, security or administrative networks without internet access can host a local model within the environment.

6

Control over model versions

An approved version can be kept stable for longer. In return comes the duty to organize security updates and later migrations yourself.

7

Reducing strategic dependencies

Open-weight models and interchangeable runtimes reduce lock-in to a single provider. Full independence still rarely results: dependencies remain on hardware, licenses, operating systems, drivers, open-source components, partners and support.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 6
06 · Managing expectations

What local AI does not do automatically

Five widespread assumptions lead to poor decisions in practice. They should be clarified before any investment.

AssumptionWhy it is not quite true
“Our data does not leave the company.”Only if telemetry, updates, logs, interfaces, front ends and add-on services are configured accordingly. Otherwise error reports, update services, external embedding or search services, or monitoring will still transmit data.
“Then we are GDPR-compliant.”The storage location alone does not decide this. Purpose, legal basis, permitted data categories, data minimization, deletion concept, access and rectification processes, roles, logging and, where applicable, a data protection impact assessment remain to be checked.
“A local model does not hallucinate.”Incorrect or invented answers are a property of generative models, not a consequence of cloud operation. Local systems also need source citations, thresholds, professional sign-off, refusal to answer on a thin data basis, and human control points.
“After installation there are hardly any costs.”Beyond hardware there are administration, power and cooling, updates, security reviews, backups, model testing, interface maintenance, support, contingency provisions and technical renewal.
“Open source means free and unrestricted.”Often only the model weights are available. License terms can restrict use, redistribution, modification or certain scenarios. Every model therefore needs a documented license review.
Note

The German Data Protection Conference recommends defining fields of use, purposes, legal bases, responsibilities, technical safeguards and the review of results before deployment.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 7
07 · Fields of application

Which applications are particularly well-suited

Local AI delivers its value where your own data and process knowledge are decisive. Every professional answer should be returned with its source, document version and approval status.

Company knowledge and technical documentation

An internal assistant searches work instructions, assembly manuals, product data sheets, maintenance documents, risk assessments, quality specifications and approved templates.

Service and maintenance

The technician describes a fault by voice. The system structures equipment, fault pattern, actions taken, spare-part needs and next step, and suggests relevant documents or earlier cases.

Construction, projects and assembly

Minutes from voice notes, daily site reports, deviations from project files, open points from meetings, bills of quantities. Without replacing professional, technical or legal sign-off.

Traffic safety and access control

Connects traffic management plans, traffic orders, conditions, setup and inspection records, material lists and internal check notes. The decision on securing a specific site stays with qualified staff.

Production and quality management

Structure shift handovers, summarize deviation reports, classify inspection reports, locate work instructions, match fault patterns against approved cases.

Sales and quotation support

Classify enquiries, extract requirements from emails, find references, propose quotation building blocks, prepare CRM entries. Prices and commitments remain subject to approval.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 8
08 · Where to draw the line

Not every task belongs on your own server

For some tasks an external service or a hybrid solution is more cost-effective or simply better suited. These cases should be deliberately excluded.

Highly variable peak loads

If very high compute is only needed occasionally, external infrastructure is usually cheaper, for example for one-off analysis of very large document collections, extensive image or video generation, or infrequent bulk processing.

The highest demands on general knowledge

Smaller local models can lag behind leading external models on demanding strategy, research or problem-solving tasks.

Constantly changing external information

A local model does not automatically know current laws, new product information, market prices, news, security advisories or amended standards. Such content requires controlled data sources or external research functions.

High-risk automated decisions

Particularly critical are decisions affecting recruitment and personnel assessment, creditworthiness, insurance benefits, medical treatment, access to essential services or security clearances. Additional legal, technical and organizational requirements apply here.

Fully automated external communication

Quotations, complaints, service commitments or legally relevant statements should not be sent without defined approval rules.

Regulatory note

The EU rules for certain high-risk AI systems were staggered by the Digital Omnibus adopted in 2026: for standalone high-risk applications (Annex III) they apply from 2 December 2027, and for AI in regulated products (Annex I) from 2 August 2028.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 9
09 · Model selection

The biggest model does not win

The capability of smaller models has risen sharply. The Stanford AI Index reports that the gap between open and closed models on selected benchmarks narrowed from eight to 1.7 percent within a single year. At the same time, the price of inference at the former GPT-3.5 level fell by more than a factor of 280 between November 2022 and October 2024. Hardware costs recently fell by around 30 percent and energy efficiency rose by around 40 percent per year.

Guiding principle

The best AI is the smallest one that reliably fulfils the process. Do not oversize infrastructure for years, keep models interchangeable, do not decide on public benchmarks alone, reassess regularly, and test with smaller model classes first.

Five evaluation dimensions

DimensionKey question
Professional qualityAre tasks solved correctly using your own documents?
Response timeIs the speed sufficient for the process?
ConcurrencyHow many employees use the system at the same time?
Resource requirementsWhat hardware is needed under normal and peak load?
OperabilityCan the model, license and software be operated over time?

Your own test cases, not a vendor demo

A robust model test comprises at least 50 to 100 real tasks: typical questions, difficult edge cases, incomplete input, contradictory documents, unanswerable questions, confidential content, manipulated documents and factually wrong assumptions by the user. Approve only once defined minimum scores are met.

Factual accuracy 30% Source grounding 20% Completeness 15% No invented statements 15% Response time 10% Operating cost 10%
Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 10
10 · Infrastructure

Parameter count alone is not enough for planning

The memory required for the raw model weights can be estimated approximately:

Parameters × bit width ÷ 8
A model with eight billion parameters requires roughly four gigabytes for the weights at 4-bit quantization.

On top of this come requirements for the runtime, intermediate results, context memory, several concurrent users, embedding models, security components, and the operating system and reserves.

Model classTypical use
Small modelsClassification, extraction, short assistance, single workstation
Medium modelsKnowledge assistant, document analysis, service support
Large modelsMore demanding reasoning, complex generation, higher infrastructure cost
Several specialized modelsRouting by task, data class and required performance

This classification does not replace a load test. Quantization, context length, output length and concurrency can change the requirements considerably.

Not just “tokens per second”

For the workflow, other figures also matter: time to first output word, response time with 5, 10 or 20 users, behavior with long documents, stability after several hours, error rate on interface calls, load during indexing of new documents, and recovery after a failure.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 11
11 · Providing knowledge

RAG or fine-tuning?

Company knowledge usually belongs in a controlled knowledge layer, not in the model itself. With retrieval-augmented generation (RAG), the system first retrieves relevant content from an approved data basis and passes it to the model together with the question.

RAG – suitable for

Policies, project documents, product information, maintenance documents, contracts, rulebooks and frequently updated content.

Advantages: knowledge updatable without retraining, sources displayable, permissions at document level, outdated content removable, answers restrictable to approved sources.

Fine-tuning – suitable for

Recurring output formats, domain language, classification, specific dialogue patterns, structured extraction, standardized workflows.

Less suitable for frequently changing facts, current product data, document versions, permission control and content that must be deleted on demand.

“RAG provides knowledge. Fine-tuning changes behavior.”

In many enterprise solutions the combination makes sense:

1

a suitable base model,

2

a controlled knowledge base,

3

clear process rules,

4

and, where needed, a limited domain adaptation.

Data protection note

The German Data Protection Conference notes that RAG can influence the data protection assessment but still requires a case-by-case review. Legal bases, data subject rights, protection against data extraction and the security of the connected knowledge base remain relevant.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 12
12 · Architecture

More than a model server and a chat window

A robust enterprise architecture consists of eight coordinated building blocks.

1

User interface

Web, mobile, integration into line-of-business systems, voice or telephony interface.

2

Identity & access

Sign-in via existing accounts, roles and groups, document permissions, MFA.

3

AI gateway

Decides on model choice, permitted data, tool calls, logs and blocks.

4

Model runtime

Model management, resource control, parallelization, version changes, fallback.

5

Knowledge system

Import, text recognition, chunking, metadata, vector and full-text search, versioning, deletion.

6

Integration layer

ERP, CRM, DMS, file server, email, calendar, ticketing, production, project management, telephony.

7

Monitoring & logging

Errors, latencies, usage, model version, data source, quality and security events.

8

Governance layer

Approved use cases, data classes, control points, escalation, quality thresholds, accountability.

Interplay

No block carries the load alone. Only their interaction makes the AI productive and controllable.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 13
13 · Information security

Local operation shifts the target of attack

Generative AI brings additional attack surfaces. The German Federal Office for Information Security points, among other things, to risks from manipulated input, unintended disclosure of information, faulty results, and attacks on models and data. NIST recommends capturing, measuring and treating risks across the entire lifecycle, from selection and development through operation to decommissioning.

Prompt injection

A document or input contains hidden instructions intended to override the desired behavior.

Data leakage

The model discloses information the user is not authorized to access.

Manipulated knowledge sources

A compromised document influences later answers.

Supply chain

Model files, containers, libraries or extensions contain vulnerabilities or malicious parts.

Over-privileged agents

A system with write access to email, files, CRM or ERP performs unwanted actions.

Secrets in logs

Prompts, answers, credentials or personal data are logged without protection.

Minimum measures

Network segmentationRole-based accessEncryption Vetted model sourcesHash and version checksSoftware bills of materials Vulnerability managementPrompt-injection protectionSandbox for tools Backup and restore testingSecurity loggingIncident response
Principle for agents

Read access can be broad; write access must remain tightly scoped and traceable.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 14
14 · Data protection and law

Local does not mean outside the law

Data protection

Before productive use, the following should be documented: processing purpose, affected groups of people, data categories, legal basis, storage and deletion periods, recipients and interfaces, technical and organizational measures, options for rectification and deletion, and the need for a data protection impact assessment. The European Data Protection Board notes that models trained on personal data cannot be treated as anonymous by default; the possibility of extraction must be assessed case by case.

Employee data

For performance assessment, recruitment, staff scheduling, time recording, behavioral or qualification analysis, the data protection officer, HR and, where applicable, the works council should be involved early.

EU AI Act

The obligation to ensure an adequate level of AI literacy among the employees involved has applied since 2 February 2025. The Digital Omnibus, finally adopted in 2026, adjusted several deadlines:

DateWhat applies
since 02 Feb 2025Duty to ensure sufficient AI literacy among staff (Art. 4)
from 02 Aug 2026Official supervision and most transparency obligations (Art. 50), such as labelling of chatbots and deepfakes
from 02 Dec 2026Provider marking of AI-generated content (Art. 50(2)) for existing systems; new prohibition of abusive image content
from 02 Dec 2027Obligations for standalone high-risk systems (Annex III)
from 02 Aug 2028Obligations for high-risk AI in regulated products (Annex I)

For local AI this means: users must understand limits and failure modes, administrators need additional technical knowledge, and business units must be able to assess results; training should be documented. This whitepaper does not constitute legal advice.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 15
15 · Accountability

Responsibility must not end with “IT”

A productive AI operation distributes responsibility across clearly named roles. If one is missing, the typical weak points arise.

Management

Strategic objectives, risk appetite, budget, organizational anchoring, approval of major areas of use.

Use-case owner (business)

Process value, functional requirements, test cases, acceptance criteria, control points, professional sign-off.

IT system owner

Architecture, availability, interfaces, updates, backups, technical documentation.

Information security

Threat analysis, safeguards, vulnerability management, monitoring, incident response.

Data protection

Legal basis, purpose limitation, data minimization, data subject rights, deletion, impact assessment.

Knowledge owner

Approved sources, document versions, currency, archiving, functional ownership.

Quality owner

Test data sets, error patterns, approval thresholds, regular re-testing, quality reports.

What to avoid

Operation without a named system owner, a business unit that does not take part in testing, uncontrolled document imports, the administrator as the sole quality reviewer, unrestricted write access for AI agents, and productive operation without fallback and shutdown procedures.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 16
16 · Operations

A language model is not a one-off installation

Check before every new model version

Has the license changed? Is the required language supported? Does the answering behavior change? Do existing prompts and interfaces still work? Do resource requirements change? Are there new security risks? Are the functional minimum scores still met?

Document version management

Model name and origin, license, checksum, quantization, runtime, system prompt, security rules, connected knowledge sources, test results, approval date, responsible persons.

Useful monitoring metrics

AvailabilityResponse timeError rate Aborted requestsUnanswered questionsUser ratings Source coverageReported functional errorsSecurity events Cost per unit of use
Fallback strategy

A productive system needs a previous approved model version, a saved configuration and knowledge indexes, a documented restart, the ability to switch off individual functions, and a manual replacement process.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 17
17 · Cost-effectiveness

Hardware costs are only part of the calculation

A sound cost-effectiveness analysis captures the total cost of operation over the lifecycle, not just the initial purchase.

One-off

Server or workstation, GPU, storage, network changes, installation, integration, building the knowledge base, security review, piloting.

Recurring

Power and cooling, hosting, administration, monitoring, backups, updates, quality testing, support, replacement hardware, security management.

Hidden

Downtime, missing functional ownership, upkeep of outdated documents, model migration, dependence on individuals, low user acceptance.

Example calculation – purely illustrative

Cost blockLocal / yearExternal / year
Hardware depreciation€7,000
Energy and infrastructure€3,000
Administration and updates€12,000€6,000
Support and security€5,000€4,000
Usage fees€30,000
Total€27,000€40,000

These example figures are not market prices. Lower usage, higher availability requirements or additional administration alone can reverse the result.

Rule of thumb

Local operation becomes more attractive when usage is high and stable, a suitable internal operating team is in place, sensitive data is processed, the required model class remains manageable, and there are no extreme load peaks.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 18
18 · Decision aid

Local, private cloud or external service?

Rate each criterion from 0 to 3 (0 = not relevant, 1 = low, 2 = relevant, 3 = business-critical) and multiply by the weight. Negative weights argue against in-house operation.

CriterionWeightRating
Particularly sensitive information30–3
Operation without internet required30–3
Constant high usage20–3
Very short response time20–3
Integration into isolated systems30–3
Own IT operating capability30–3
High demands on leading model performance−20–3
Highly variable load−20–3
Frequently changing external information−20–3
Limited operating budget−20–3

Above 20 points

Local or largely local operation should be examined in depth.

8 to 20 points

A hybrid architecture is often the best starting point.

Below 8 points

A controlled external service or a private cloud is likely more cost-effective.

The matrix does not replace a data protection, security or cost-effectiveness review. It serves as a pre-selection.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 19
19 · Implementation

A pilot project in ten steps

From the use case to a sound operating decision, without starting from “We need a local language model”.

1

Choose the business problem

A concrete bottleneck, not a technology wish.

2

Limit the user group

Often 10 to 20 users to start.

3

Define the data scope

Approved documents, test data, anonymized cases.

4

Determine protection needs

Public, internal, confidential, specially protected.

5

Create reference tasks

At least 50 real tasks and edge cases.

6

Compare several models

Two model classes plus an external reference.

7

Build the knowledge architecture

Owner, version, validity, permission, deletion rule.

8

Review security and data protection

Access, logging, backup, tamper protection.

9

Measure the pilot

Time saved, quality, error types, acceptance, cost.

10

Make the operating decision

Local, private cloud, hybrid or not yet.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 20
20 · Self-assessment

Is your company ready for local AI?

Each fulfilled point counts as one. The check assesses organizational maturity, not technology alone.

Business problem

Process described · value measurable · user group known · a business unit takes ownership.

Data and knowledge

Sources known · documents have owners · version and validity recognizable · permissions documented · outdated content removable · personal data identified.

Technology

Model performance tested · concurrency and response time measured · interfaces defined · backup and restore provided · monitoring planned · fallback procedure in place.

Security

Threats assessed · write access limited · vetted model and software sources · logs without unnecessary sensitive content · incidents detectable and manageable.

Governance

Recorded in the AI register · data protection and information security involved · users trained · control points defined · quality criteria documented · an owner for updates · errors and complaints reportable.

0–9
The organizational foundations are still missing.
10–18
A limited pilot is feasible.
19–27
A good basis for the detailed review.
Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 21
21 · Case study

Technical service: making knowledge from maintenance documents usable

Situation. A technical service provider with several field teams holds maintenance manuals, machine files, fault reports, spare-part information, email threads and the experience of individual technicians. The information is scattered across file servers, the DMS and various folder structures.

Target picture. A local service assistant should:

1

structure the reported fault pattern,

2

locate relevant documents,

3

show similar earlier faults,

4

suggest inspection steps,

5

prepare a service report.

Architecture

Local language model, RAG knowledge base, sign-in via company accounts, access based on existing permissions, no automatic ERP changes, and source display on every professional answer.

Human control point

The technician decides on diagnosis, spare parts and the work carried out. The AI provides suggestions and references, not binding determinations.

Pilot metrics

Time spent searching for documents, completeness of service reports, share of usable suggestions, functional errors, and acceptance in the field.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 22
22 · Case study

Construction and projects: generating documentation from the workflow

Situation. Site managers and project leads document site progress, obstructions, additional works, material use, acceptances, open points and photos. The information arises as voice notes, messenger messages, emails or handwritten notes.

Target picture. A local AI application should structure voice notes, assign entries to a project, flag missing mandatory information, prepare a daily site report, extract open points and deadlines, and locate relevant project documents.

What the system doesWhat the system explicitly does not do
Structure and assign voice notesno legal assessment of a variation
Flag mandatory entries and open pointsno sending of service commitments
Prepare the daily site reportno changes to billing data without approval
Locate project documentsno replacement of review by site or project management
Expected benefit

Fewer media breaks, more complete documentation, faster project handovers, better findability of facts, and less dependence on individual knowledge.

Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 23
23 · Bringing it together

Making company knowledge securely available

Situation. A mid-sized company runs a DMS, an ERP, several file servers and many process descriptions. Employees frequently search for the current template, the valid work instruction, the responsible person, the approved product information or an earlier project decision.

Target picture. A Company Brain connects the approved sources with a local or hybrid language model. Every answer includes the source used, the document version, the approval status, the responsible unit and the time of the last update. Where the data basis is missing, the system does not answer speculatively but points to the responsible office.

Conclusion

Local AI can be a sensible building block for small and mid-sized companies. It offers advantages in data sovereignty, offline capability, integration and predictable usage. These advantages only materialize, however, once five prerequisites are met: a clearly defined business use case, a maintained and permission-controlled knowledge base, suitable infrastructure, defined security and governance procedures, and durably owned operations.

Next step
Local AI Readiness Check

KrambergAI works with you to assess suitable use cases, data and protection needs, local, private and hybrid architecture variants, model and infrastructure requirements, operating and security requirements, the rough cost-effectiveness and the sensible scope of a pilot project. The result is a concise decision document with an architecture proposal, prerequisites, risks and a pilot plan.

Request the Readiness Check
KrambergAI GmbH · krambergai.com
GDPR-compliant data protection Made in Germany
Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 24
24 · Studies, standards and sources

The basis of this whitepaper

Federal Statistical Office (Destatis) – Use of ICT in enterprises 2025. 26% of companies with 10 or more employees used AI in 2025 (small 23%, medium 36%, large 57%); the main barriers were lack of knowledge, legal uncertainty and data protection.
destatis.de

KfW Research – Use of artificial intelligence in the SME sector (Focus on Economics No. 533, February 2026). Special analysis of the KfW SME Panel: around 780,000 SMEs using AI (20%), construction 8%, R&D-active companies 53%.
kfw.de

Eurostat – Use of Artificial Intelligence in Enterprises 2025. EU-wide data on adoption, areas of use and barriers.
ec.europa.eu/eurostat

Stanford Institute for Human-Centered AI (HAI) – AI Index Report 2025. Gap between open and closed models on selected benchmarks narrowed from 8% to 1.7%; inference cost at GPT-3.5 level fell by more than a factor of 280 between November 2022 and October 2024.
hai.stanford.edu

Federal Office for Information Security (BSI) – Generative AI models: opportunities and risks for industry and public authorities.
bsi.bund.de

German Data Protection Conference (DSK) – Guidance on the data-protection-compliant use of AI applications, including with reference to the RAG method.
datenschutzkonferenz-online.de

European Data Protection Board (EDPB) – Opinion 28/2024 on data protection aspects of AI models. Models trained on personal data cannot be treated as anonymous by default.
edpb.europa.eu

European Commission / Council of the EU – EU AI Act and Digital Omnibus. On 29 June 2026 the Council gave its green light to the staggered new application deadlines (Annex III from 2 December 2027, Annex I from 2 August 2028).
digital-strategy.ec.europa.eu · consilium.europa.eu

National Institute of Standards and Technology (NIST) – AI Risk Management Framework. A framework for governance, measurement and treatment of AI risks across the lifecycle.
nist.gov

Editorial note

The studies cited use different populations, company sizes, periods and methods. Percentages are therefore not directly comparable. Together they show growing and accelerating AI adoption alongside persistent gaps in knowledge, law, data quality and operations. This whitepaper is intended for professional orientation and does not constitute legal, data protection or information security advice.

KrambergAI
KrambergAI GmbH · krambergai.com
Local AI in the SME Sector · Opportunities, Limits and PrerequisitesKrambergAI · Page 25