What is a forward deployed engineer?

A software engineer who works at the customer, builds the solution on their real systems and is responsible for it until it runs in daily work. Palantir coined the term; today OpenAI and Anthropic build such teams for their largest accounts.

Why not just buy ChatGPT or Copilot?

A tool for the individual saves the individual time. A workflow for the whole team only changes with a solution that is connected to your systems and that someone is responsible for.

Do you work on site or remotely?

Workshops take place at your site, building happens remotely. That keeps travel costs low, and you see us when decisions are due. Clients in Germany, Austria and Switzerland.

Which AI providers do you use?

OpenAI, Claude by Anthropic, Google Gemini or open models, depending on data protection, cost and quality. We are not tied to any provider and switch when another model fits better.

The bottleneck workshop costs €390 net and is credited. Then you receive a written fixed price. For an ongoing monthly allotment of hours we quote after a conversation.

What if the solution does not deliver?

We agree the success criterion beforehand. If the pilot does not reach it, we stop, and you only pay the agreed fixed price.

Can you complement our IT instead of replacing it?

Yes. If your company has an IT department, we work with it and hand over as much as it wants to take on.

The engineer who comes to you.And stays until it runs.

In Silicon Valley this is called a forward deployed engineer: someone who sits with the customer and puts AI into production where the work actually happens. OpenAI and Anthropic do this for their largest accounts. We do it for small and medium-sized businesses: on your systems, with the model that fits your data, at a fixed price.

Request bottleneck workshop€390 · 90 minutes · credited

What exactly is a forward deployed engineer?Read the article (German)

Does one of these sound familiar?

“The PoC was done quickly. Then it went quiet.”

Tested, approved, and then the results fade away because nobody owns the path into everyday work.

“Management does not decide on continuation in time.”

The pilot expires before anyone says yes or no. Without measured benefit there is no basis for the decision.

“The benefit is hard to communicate.”

The result was technically there, or not satisfying, but nobody could put a number on what it brings the company.

In all three cases the technology is not what is missing. What is missing is someone who defines the benefit beforehand, measures it afterwards and makes the solution run at your company.

How an engagement works

Four steps. After each one, you decide whether to continue.

Bottleneck workshop€390 · 90 minutesWe find the workflow that costs your team the most time and check whether software or AI can solve it. You receive a written fixed price and a success criterion against which the benefit is measured later.

Build, at your placeFixed price · in weeksWe build on your real data and systems, together with someone from your team who approves it later. The workshop takes place at your site, building happens remotely.

Put into operationRollout · measurementRollout with the users, approvals, monitoring. We stay until the workflow runs without prompting and the agreed savings are measured.

Hand over or stayYour choiceCode, data, access and documentation belong to you. If you want us on the team for longer, you get a monthly allotment of hours at a fixed price.

Request bottleneck workshop€390 · 90 minutes · credited

OpenAI, Claude or an open model on your own server.

We belong to no vendor. Three questions decide which model fits.

Data protectionMay the data leave the building? If not, an open model runs on your server or in a German data centre.

CostReading a receipt does not need a frontier model. Checking a contract does. For each task we use the smallest model that is reliable.

QualityBefore rollout we measure how often the model is right, using your real cases. Only then does it become everyday work.

EXAMPLE

DCM Lumina, our AI inbox, runs locally at the customer with their own AI key. They choose between Anthropic Claude, Google Gemini and Fireworks AI.

OpenAI, Claude, Gemini and the other names are trademarks of their owners. We name them because we work with them, not because we belong to them.

Fixed price. Or longer on the team.

Two options that can be combined.

Per projectDefault

Fixed price

After the workshop, cost and go-live date are in writing. No open-ended consulting days.

Scope and success criterion agreed beforehand
Price and date in writing
Workshop fee is credited

Option

Engineer on retainer

You have several projects or want someone on the team who keeps improving workflows. Then we agree a monthly allotment of hours at a fixed price.

Monthly allotment of hours, flexible to call on
Fixed monthly price instead of day rates
Quote after a conversation

What you keep

How we run software (German)

Source code in a repository you have access to
Your data in an open format, exportable at any time
Access to servers and services in your name
Documentation and operations manual
The freedom to switch model or provider

Where the experience comes from

We have worked this way since 2016, long before the term existed.

SAP and energy utilities

IT consulting in projects where we sat on the customer team until interfaces and portals ran. Regulated environments and approval processes are everyday work.

Our own software in production

DCM Lumina, orderplix and nucompliance are our own AI products running at customers. We know what AI looks like in operation, not just on slides.

One contact

From the first call to operation you talk to the same person who writes the code.

About LEANOFY See references

Frequently asked questions

Which workflow costs you the most time?

The bottleneck workshop finds out: 90 minutes, €390, credited towards implementation. Afterwards you know what the solution costs and when it runs.

Request bottleneck workshop Prefer to talk first? +49 800 589 5433, free of charge.