AI enablement partner for German SMEs

We make your software AI-ready

61% of German companies run on outdated software. AI can drastically reduce your development costs — but only if your codebase is prepared for it.

Book a Quick Check — 990 € Learn More
61%1 of companies on
outdated platforms
43%2 of SMEs without
an AI strategy
83%3 see enormous potential
in AI development
78 AI-Ready
from Score 22

Already trusted by

Procter & Gamble Proaut Technology Magma Rock / Lanovo LAT

The Problem

AI tools are here — but your software isn't ready

The Trust Crisis

Only 29%4

of developers trust AI-generated code. AI PRs have 1.7x more bugs5 than human ones. The problem isn't the AI — it's the codebase.

The Productivity Lie

-19%6

Developers feel 20% faster with AI. But on complex tasks they are actually 19% slower. Without a clean codebase, AI creates more problems than it solves.

The Legacy Standstill

53%7

of companies report problems steering their digitalisation projects. Outdated systems slow down innovation, drain budgets and make AI use impossible.

Our Solution

Not just modernise — enable

We make your codebase AI-Ready

EPSIA modernises your existing software so that AI assistants like Claude Code, Cursor or GitHub Copilot can work with it reliably. After that, we enable your team to build features independently.

  • Modular Architecture AI needs manageable, isolated units of code
  • Strict Typing Types are the most important context for AI assistants
  • Comprehensive Testing The safety net for AI-generated changes
  • AI Context Files CLAUDE.md, .cursorrules — the AI understands your domain
  • CI/CD Pipeline Automated quality assurance for every commit
Before
22
Not AI-ready
  • Monolithic code
  • No typing
  • 0% test coverage
  • No documentation
  • Manual deployments
After
78
AI-Ready
  • Modular architecture
  • Fully typed
  • 82% test coverage
  • CLAUDE.md + API docs
  • CI/CD pipeline

EPSIA AI-Readiness Score™

Data-driven. Reproducible. Comparable across industries.

Our analysis software scans your codebase automatically and rates it on a scale of 0–100. Static code analysis, architecture metrics and hands-on AI tests feed into 8 weighted dimensions — for an objective, reproducible result.

20%

Architecture & Modularity

Can the AI work on isolated modules?

15%

Code Quality & Consistency

Does the AI learn from clean, consistent code?

15%

Typing & Data Structures

Does the AI have enough type context?

15%

Test Coverage

Can the AI verify its changes?

15%

Documentation & AI Context

Does the AI understand the domain?

10%

DevOps & Infrastructure

Can the AI run and test code?

5%

Dependencies & Stack

Does the AI know the technologies?

5%

Security & Configuration

Is the system protected against AI errors?

0–20 Not AI-ready
21–40 Limited
41–60 Conditionally AI-ready
61–80 Largely AI-ready
81–100 Fully AI-ready
View the methodology in detail

Full transparency: all 8 dimensions explained and how the score is calculated

Our Services

From analysis to ongoing support

01

Quick Check

990 €

Our scoring software analyses your codebase automatically. Complemented by an expert review, you receive your AI-Readiness Score and the three most impactful measures.

  • Automated code analysis (tool-supported)
  • AI-Readiness scorecard with radar chart
  • 30-min results call
1 day
02

AI-Readiness Assessment

4,500 – 8,500 €

Full assessment in 4 phases: automated analysis by our scoring engine, manual architecture review, hands-on AI test on your code and a detailed report with a prioritised roadmap.

  • 8-dimension analysis
  • Hands-on AI test (3 tasks)
  • Prioritised roadmap
  • Effort estimate
3–5 days
Core Service
03

AI-Ready Modernisation

15,000 – 120,000 €

Hands-on refactoring of your codebase. We raise your score to 70+ and make your software fit for AI-assisted development.

  • Refactoring & modularisation
  • Testing & typing
  • CI/CD pipeline
  • CLAUDE.md & AI context
4–24 weeks
04

AI Enablement Training

3,500 – 6,000 €

We train your team in working with AI assistants. Hands-on workshops with your own code.

  • Workshop (2–3 days)
  • Set up AI workflows
  • Practical guide
2–3 days
05

AI-Ready Retainer

1,500 – 4,000 € / month

Ongoing support: code reviews of AI-generated changes, architecture consulting and quality assurance.

  • Monthly quality report
  • AI code reviews
  • Architecture hotline
from 6 months
06

AI-Assisted New Development

25,000 – 200,000 €

When modernisation doesn't make sense: complete new development with a modern stack, AI-ready from the start (Score 85+).

  • Modern architecture
  • Complete test suite
  • AI-optimised codebase
2–6 months

How It Works

Five steps to AI-ready software

1

Quick Check

Our scoring software analyses your codebase and delivers your AI-Readiness Score within 24 hours.

Result: scorecard with radar chart + top 3 measures
2

Assessment

Our scoring engine analyses all 8 dimensions in detail. Complemented by hands-on AI tests on your code and an expert review.

Result: report + roadmap + effort estimate
3

Modernisation

We refactor your codebase step by step — architecture, testing, typing, documentation, CI/CD.

Result: AI-Ready codebase (Score 70+)
4

Training

Your team learns to develop features independently with AI assistants on the modernised codebase.

Result: enabled team + AI workflows
5

Support

With the retainer we safeguard quality, review AI-generated code and stand ready for architecture questions.

Result: long-term quality + a dedicated contact

Up to 50% Funding

Your modernisation can be subsidised

Example Calculation

Assessment 6,500 €
Modernisation 45,000 €
Training 5,000 €
Total cost 56,500 €
Possible funding depends on programme

Which grants and loans are available in your federal state we assess individually — type and amount depend on the respective programme and its eligibility requirements.

Available Funding Programmes

Bayern Digitalbonus

Up to 50% grant, max. 30,000 € (Digitalbonus Plus)

BW Digitalisation financing

Grant and loan variants via L-Bank

KfW-ERP Digitalisation loan

Interest from 3.5%, up to 25 million €

BAFA consulting funding

Grant up to 1,750 € for consulting projects

We support you with the application or connect you with specialised funding consultants.

Is your software ready for AI?

Our scoring software delivers the answer. Within 24 hours you receive your AI-Readiness Score, a radar chart of all 8 dimensions and the three most impactful measures.

Quick Check — fixed price 990 € plus VAT

We will get back to you within 24 hours with all the details.

Why EPSIA?

Experience meets innovation

EPSIA is a Berlin-based software company. With over 15 years of experience in developing and modernising custom software, we work for companies such as Procter & Gamble, Fluke Process Instruments and Proaut Technology.

Today we focus exactly this modernisation expertise on AI enablement: we know what an established codebase needs to make AI-assisted development truly work — because we use AI tools ourselves every day.

Absolute transparency Unrestricted access to all source code
Collaboration as equals Your project management integrated into every step
Measurable results Tool-supported AI-Readiness Score before/after — not an estimate
EPSIA GmbH Based in Berlin, active throughout the DACH region
Managing Director: Daniel Erbert Over 15 years of experience in software development and IT consulting
References Procter & Gamble, Fluke, Proaut Technology, Magma Rock
Specialisation Legacy modernisation, custom software, IoT, embedded

What our project partners say

Voices from our collaboration

„EPSIA GmbH is an officially approved supplier for development services and software solutions at Procter & Gamble.“

Simon Klotz
Simon Klotz Project Manager, Procter & Gamble

„Both we and our international customers greatly value the high quality and flexibility.“

Peter Krause
Peter Krause Site Manager, Proaut Technology GmbH

„We look forward to continuing our collaboration with our project partner.“

Sebastian Rescha
Sebastian Rescha Manufacturing Engineering Manager, Fluke Process Instruments

„We greatly value Mr Erbert and his team and look forward to continuing our successful collaboration.“

Rico Finster
Rico Finster Managing Director, Magma Rock Estate GmbH

Frequently Asked Questions

Everything you should know before our first conversation

AI-Ready means your codebase is structured so that AI assistants like Claude Code, Cursor or GitHub Copilot can work with it reliably. This includes modular architecture, strict typing, comprehensive testing, good documentation and a CI/CD pipeline. We measure this with our AI-Readiness Score on a scale of 0–100.

Our scoring software supports PHP, JavaScript/TypeScript, Python, Java and C#/.NET. Most legacy systems in SMEs are based on PHP or older JavaScript — exactly our area of expertise. For other languages, we discuss the options in the initial conversation.

The Quick Check takes 1 day, the full assessment 3–5 days. The actual modernisation depends on the scope: a single module can be AI-Ready in 4–6 weeks, a complete application takes 3–6 months. We work in phases — you see a measurable score increase after each phase.

AI assistants don't replace developers — they make existing staff significantly more productive. After modernisation, a technically skilled employee with AI support can implement features that would previously have required an experienced developer. For complex architecture decisions we recommend our retainer.

Yes. Depending on the federal state, programmes such as the Bayern Digitalbonus (grant up to 50%, max. 30,000 €), the BW Digitalisation financing via L-Bank or the KfW-ERP Digitalisation promotional loan may apply. The type and amount of funding depend on the respective programme and its requirements — we assess this individually and support you with the application or connect you with specialised funding consultants.

Absolute security. Your source code is never uploaded or stored externally at any point. Our scoring software runs locally on our systems. The analysis is carried out under an NDA. You retain full control and full access to all source code at all times — this is one of our core principles.

The Quick Check costs 990 € plus VAT at a fixed price. Within 24 hours you receive your AI-Readiness Score, a radar chart of all 8 dimensions and the three most impactful measures. Based on that, you decide whether and how you would like to continue — with no obligation.

News & Insights

Articles on AI-Readiness, legacy modernisation and AI-assisted development

AI-Readiness

Why 81% report more production problems with AI code8

The trust crisis in AI-assisted development has a cause that most overlook: it's not the AI — it's the codebase.

July 2026
Modernisation

PHP Modernisation 2026: From Legacy to AI-Ready in 5 steps

A practical guide for SMEs who want to make their PHP application fit for AI assistants — with concrete measures and effort estimates.

July 2026
Funding

NIS-2 and the EU AI Act: What SMEs must do now

Two new regulations are putting companies under pressure. What this means for your software — and how funding programmes reduce the costs.

July 2026

Contact

Let's talk about your software

Address

Siegfriedstraße 152
10365 Berlin

Sources

  1. Slalom, "Global AI Insights Survey" (August 2025), German subset (161 respondents in Germany). slalom.com
  2. KI-Index Mittelstand 2025, Deutscher Mittelstands-Bund & Salesforce (526 companies). salesforce.com
  3. Lünendonk study 2025 on the IT services market in Germany. luenendonk.de
  4. Stack Overflow Developer Survey 2025. survey.stackoverflow.co
  5. CodeRabbit, "State of AI vs Human Code Generation" (2025), analysis of 470 GitHub pull requests. coderabbit.ai
  6. METR, "Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity" (2025). metr.org
  7. Bitkom, "Digitalisierung der Wirtschaft 2025" (603 companies with 20+ employees). bitkom.org
  8. CloudBees, survey of 200+ technology leaders (2026). globenewswire.com