adSCAILE

Agent-based software development

What is adSCAILE? How adSCAILE works in practice How much longer can you afford reactive AI? Where does your organisation stand today? Four characteristics of the agent-based era Proprietary tools from everyday project work adSCAILE in productive operation adSCAILE in any project context Your benefits and journey with adesso

The future is now.

Software development is now agent-based.

adesso creates software throughout the entire lifecycle using AI agents. To this end, we have developed our own tried-and-tested process. adSCAILE opens up entirely new possibilities in software production. Scalable, compliant and proven.

adSCAILE – adesso Smart Cycle for AI-enhanced Lightweight Software Engineering

The Framework

What is adSCAILE?

adSCAILE is our technology-agnostic process for agent-based software development. Not a theoretical construct – but the result of over a year of intensive project experience and the consolidation of best practices from almost 30 years at adesso.

AI agents handle the operational implementation autonomously, whilst humans focus on business requirements and architectural decisions.

This results in ultra-short development cycles of two to four days – with faster iterations, better quality and sustainable integration of agent-based AI approaches.

adSCAILE is completely technology-agnostic: independent of tools, models or platforms – and works in greenfield projects as well as in complex brownfield scenarios and when modernising legacy systems.

The acronym


ad – adesso: 29 years of project experience as a foundation

S – Smart: Intelligent, goal-oriented process control

C – Cycle: Ultra-short, highly iterative development cycles

AI – AI-enhanced: AI agents as the primary implementation unit

L – Lightweight: Lean processes, small teams

E – Engineering: Engineering thinking remains at the core

Agents need structure. Our industry expertise and best practices from almost 30 years at adesso form the framework within which AI can work safely and efficiently. Without strong input, there can be no strong output – no matter how good the assembly line is. With adSCAILE, we create the conditions for excellent software through the collaboration of our developers with agent-based AI.
Frank Dobelmann, CTO, adesso SE

The process

How adSCAILE works in practice

adSCAILE covers the entire software development lifecycle – not an isolated AI application, but a seamless end-to-end process.

Step 1 – System requirements & architecture (Human Activity)

The team defines requirements, architectural principles and guidelines. Humans specify the “why” and “what” – the foundation on which agents can act with confidence. Without this step, the process goes off the rails.

Step 2 – Component Specification (Human & Agent Artefacts)

Specs and guidance are created as structured artefacts – the context that AI agents need for precise, focused implementation. Clear inputs prevent hallucination.

Step 3 – Implementation (Agent Activity in Long-running Loops)

AI agents autonomously handle operational implementation: generating code, running tests, fixing bugs, creating documentation – in continuous feedback loops. Cycle: 2–4 days.

Step 4 – Review & Quality Assurance (Human Oversight)

Automated review pipelines with AI quality gates ensure code quality. The team handles final validation, merge decisions and acceptance.

Step 5 – Operation & Further Development (Continuous Improvement)

adSCAILE does not end with the release. Specialised agents handle troubleshooting and optimisation. Agent skills are refined through a continuous improvement process.

The new reality

How much longer can you afford reactive AI?

The era of AI as a tool is over. Anyone still relying exclusively on chatbots and coding assistants today is optimising the wrong lever – and falling behind competitors who are already delivering agent-based solutions.

2026 marks a turning point: AI agents plan independently, break down tasks, execute them and improve themselves in autonomous feedback loops – around the clock.

Typical weaknesses of reactive AI approaches:

Coding assistants

Reactive, input-dependent, no independent action

Isolated AI tools

Stand-alone solutions with no end-to-end impact on the SDLC

Lack of compliance

Data protection, IP protection and governance unresolved

Software development is undergoing a fundamental shift

We are witnessing the transformation from a primarily human activity to a hybrid process. From now on, humans and AI agents will design software together.

This is not an evolution. Agile working methods are reaching their limits. In this context, humans are more than just the ‘human in the loop’. We orchestrate, control, train and direct the agents and, thereby, the process. This is AI-orchestrated software engineering. This is adSCAILE.

Contact us

Proof of concepts and initial pilots? That’s a thing of the past. 2026 is the year when agent-based AI moves from the lab into production. You can buy the tech – but you have to build the mindset. And anyone who fails to act now will pay dearly in two years’ time: to the competition.
Benedikt Bonnmann, Member of the board, adesso SE

Where does your organisation stand today?

Software development is undergoing a transformation of unprecedented scale. adesso distinguishes between four levels of AI involvement.

Level 01

AI-assisted SWE

AI as a reactive tool: coding assistants, autocomplete, ChatGPT in a browser tab. AI provides support on request – the developer remains the sole actor.

Typical examples: GitHub Copilot, chat tools

Level 02

AI-enabled SWE

AI is systematically integrated into the development process. First automated sub-steps – testing, documentation, code reviews. Humans retain full control.

Typical: AI-supported CI/CD pipelines

Level 03

AI-driven SWE

AI independently takes on larger task sets. Agents operate within defined loops. Humans orchestrate, AI delivers. First agent-based approaches in productive use.

Typical: Agent workflows, Claude Code

Level 04

AI-native SWE - adSCAILE

Agent-based AI is not an add-on, but the foundation. Small teams, ultra-short cycles, autonomous implementation. Humans think as architects and orchestrators – not as coders.

adSCAILE operationalises this stage.

How adSCAILE works

Four characteristics of the agent-based era

adSCAILE combines four design principles that make agent-based software development scalable and manageable.

  • 01 – Beyond Agile
    Agent-based development requires leaner processes. Small teams, lightweight coordination cycles – beyond the classic sprint model.
  • 02 – Fluent Process
    Highly iterative approach with permanent, continuous cycles – like an assembly line running 24/7. Agents work through the night too.
  • 03 – Discrete Building Blocks
    Highly specialised agents with a clearly defined focus and precise context. Clear handover points minimise misalignment.
  • 04 – Guided Automation
    We define, control and set the course – the agent implements. This is how the transition from one-off production to a production line is achieved.

Technological Foundation

Proprietary tools from everyday project work

adSCAILE is not just a process – it is underpinned by a proprietary technological foundation that adesso has developed and tested in its own practice.

streb – Operating Framework

The operating framework for agent-based software delivery. Set up a fully production-ready development environment, including Claude Code, in under five minutes. The new AI workspace – ready for immediate use.

adesso AI Hub – LLM Gateway

Standardised, legally compliant access to a wide range of Large Language Models – both on-premises and in the cloud. With full cost transparency, governance and compliance. LLM costs are invoiced to projects.

In practice

adSCAILE in productive operation

adSCAILE is not a proof of concept. It has been used in real customer projects since 2025 – with measurable results.


Project: einfachnetz.de
Client: TransnetBW & VNBs Baden-Württemberg
Industry: Utilities / Energy
Technology: AWS · GitLab · React · MCP · RAG


adSCAILE in production
Classic Scrum team vs. adSCAILE Agentic Team – running in parallel, on the same production system. 2 releases per week.


Results & learnings:

  • Frontend rewrite Angular → React: fully agent-driven, 0 lines of manually written code
  • Parallel operation confirmed: agentic team delivers faster and more consistently than classic Scrum team
  • 5,752 pages of technical documentation embedded as RAG domain knowledge
  • Multi-agent SDLC: Analysis · Implementation · Testing · Review · Documentation – fully agent-supported


adSCAILE delivers measurably faster, with higher quality – and without traditional headcount growth.

Use cases

adSCAILE – in any project context

adSCAILE is not a greenfield-only approach. The framework is deliberately designed for both scenarios.

Scenario A

Greenfield

New systems are built using an agent-based approach from the outset. No legacy debt, no refactoring – adSCAILE defines the process from the initial requirement right through to the production release.

  • Small teams (4–6 people), ultra-short cycles from the outset
  • streb enables a production-ready environment in under 5 minutes
  • Full efficiency potential from day one – no migration burden
  • Architecture designed for agent-based implementation from the outset

Typical for: new digital products, platform development, SaaS development

Scenario B

Brownfield

adSCAILE integrates into existing systems and mixed teams. Traditional and agent-based development run in parallel – without mutual interference, with clearly defined handover points.

  • Two-speed model: classic team and adSCAILE team in parallel
  • Pull methodology (e.g. Kanban) prevents conflicts between teams
  • Classic: 2–3 week cycles | adSCAILE: 2–4 days – both within the same system
  • Legacy modernisation and new development possible simultaneously

Typical for: complex, organically grown infrastructure and heterogeneous systems

Traditional approach

Sprint: 2 - 3 weeks


vs.

adSCAILE Agentic Team

Cycle: 2 - 4 days

Your benefits

Measurable added value – for business and IT

For business decision-makers & C-level

  • Drastically shorter time-to-market – releases within 2–4 days instead of 2 weeks. More releases, faster response to market demands.
  • Secure competitiveness – AI integration is not an option, but a prerequisite. Those who act now secure the future viability of their IT.
  • Scaling without increasing headcount – Greater output with small teams (4–6 people). Significant efficiency gains across the entire SDLC.
  • Transparent costs – LLM usage costs fully billable to projects – whilst total development costs fall.

For IT managers & CTOs

  • Technology agnosticism – adSCAILE works independently of specific LLMs, tools or platforms – maximum flexibility.
  • Compliance & Governance by Design – adesso AI Hub ensures legally compliant LLM access, IP protection and full cost transparency.
  • Brownfield & Greenfield – Deployment in existing systems and in new projects – without mutual interference.
  • Quality assurance through structure – automated review pipelines and AI quality gates minimise hallucinations and ensure code quality.

Your journey with adesso

AI@SWE Roadmap


adSCAILE is not a one-off measure. adesso supports you along a comprehensive roadmap – from assessing your current situation to organisation-wide implementation.

Phase 01

Optimised Software Development

From initial pilot projects using AI to scaled efficiency gains.

  • Kick-off & Baseline Assessment
  • Establishing the AI foundation in software engineering
  • Project implementation with adSCAILE
  • Agent-based development pipelines

Phase 02

Future-proof Applications

Application modernisation with AI – from legacy replacement to AI-ready core systems.

  • AI readiness of the application landscape
  • Implementation of migration projects
  • AI-supercharged applications

Phase 03

Scaled transformation

Embedding agent-based software development processes across the entire organisation.

  • Consultancy from adesso AI experts
  • Competence building & change management

Ready for agent-based software development?

Let’s work together to identify where adSCAILE offers the greatest leverage within your organisation – in ongoing projects, when modernising legacy systems, or when establishing new development pipelines.