Trusted by
You're not alone - 56% of CEOs report zero financial return from AI. The reason is always the same: individual usage, no delivery system, no path. We build that path with you, step by step, organisation-wide.
To talk about AI adoption clearly, we needed a shared language. That is why we built the Delivery Levels: a scale that shows where your team stands today and where the value is.
Over the past months we ran AI workshops with three companies and did the honest read with each of them - different sizes, different industries, different tools. All three landed on DL2. Because underneath, it's the same story: individual, isolated AI use. Everyone gets a little faster on their own, and nothing ever becomes a team standard.
The shape says it all: one spike where the typing happens, and almost nothing anywhere else. The AI helps every developer write code a little faster, but requirements, architecture, testing, review and documentation still run exactly the way they did before the licenses arrived.
Leadership approved GitHub Copilot for every developer in one procurement decision. Rollout in two weeks, adoption at 90%.
Copilot in every IDE. Nothing else: no rules files, no architecture docs fed to the AI, no review methodology. Every suggestion starts from zero project knowledge.
~1.25× individual gains, invisible at team level
This shape is jagged for a reason. Development sticks out because a handful of power users built brilliant personal setups, while testing and documentation barely register. The peaks belong to individual people rather than the team, so when those people are on holiday, the peaks go with them.
No central decision. Engineers brought their own tools: Cursor here, Claude there, ChatGPT everywhere. A few power users are visibly faster.
Five different AI tools across twelve teams. Some engineers wrote personal rules files, none are shared. The best setup in the company lives on one laptop.
Pockets of 2× next to teams at 1×
The neatest shape of the three, and that is exactly the problem. Every phase sits at the same modest level, because the platform made access equal without making the AI any smarter about the actual product. Even, controlled, and stuck.
Built an internal AI gateway: SSO, compliance logging, an approved-model list, even an internal chat UI. Eighteen months of platform work.
Impressive infrastructure around the AI, nothing inside it. The gateway routes prompts but carries no project conventions, no service map, no team standards. Secure access to a context-blind assistant.
Enterprise-grade plumbing, individual-grade output
Three roads in, one plateau. What all three are missing is identical: a shared context system the AI reads before every task. That's the single move from DL2 to DL3.
You don't need to hire a full-time task force for this. What you need is the capability itself - a few days a month, at a cost you can predict, for as long as it takes until the new way of working becomes routine. We've structured it as three stages that build on each other, and none of them commits you to the next one.
A 30-minute conversation plus a scored analysis of your six delivery phases.
We build the context system the AI reads before every task, on your own codebase.
Foundations through to leadership, taught in your codebase on your own work.
We typically work with product and engineering leaders in Mittelstand companies between 50 and 1,000 employees: manufacturing, software, financial services, logistics, retail.
The feedback from our workshops and trainings - and the outcomes they can point to a few months later.
A two-day AI Delivery Workshop in Munich with the SimonsVoss engineering teams: seven teams assessed on the Delivery Levels, a shared CLAUDE.md established as the team brain, and AI use structured across all six SDLC phases, from spec-first requirements to release notes generated from the repository.
“We came to Intertec with an app that was holding us back: slow to build on, slow to change, no AI anywhere in the process. They didn't just rebuild the product. They rebuilt how we build: a design system, shared components, AI in the ideation and delivery process itself. We went from a fully manual setup to a DL3-ready team.”
“We didn't have a product, we had spreadsheets. Every basketball program registration was manual, row by row. Intertec started at zero with us: the ideation, the planning, the product requirements, all of it AI-driven from day one. What we got is a custom platform, built DL3-ready. It changed what our small team can take on.”
Field notes from the practice: how the Delivery Levels work, and what it takes to move up one.
The complete guide to the five-level scale: definitions, productivity factors and the objective gates that decide when a team has cleared a level.
DL4 is where you hand your shared context to a fleet of agents and step back to review. The hard part is not the agents. It is the agent org chart.
The teams winning with agents did not start with agents. They started with context: shared rules, feedback loops and human gates the AI reads before every task.
In 30 minutes we place your team on the ladder, name the one bottleneck blocking the rest, and show what the next level is worth. Report in one week - ready for the board.