The governance playbook behind flaschenpost’s 120 minute delivery promise

Impact

• 30+ warehouse locations coordinating in real time
• 887 users building safely in a governed platform, with 26 builders across business units
• 20+ legacy pages replaced with 1 developer

“Retool is an execution multiplier. It helps us to move from a problem to a working solution very quickly, and helps us to iterate a lot faster.”

—Niklas Hoffmann, Head of Operations Solutions, flaschenpost SE

Flaschenpost delivers groceries and beverages to more than 40% of German households, with a promise of arrival within 120 minutes. Meeting that promise depends on coordination between dozens of warehouses, a large delivery fleet, and operations teams that need real-time visibility into all of it.

As flaschenpost has scaled, their sprawling toolchain of individual solutions couldn’t keep up. They needed a platform that could grow safely across a decentralized organization, that provides the governance to trust what is built, and the flexibility to let teams build it.

Company overview

• Industry: E-commerce, on-demand grocery and beverage delivery
• Regions: Germany
• Retool products: Apps, Workflows, Retool Database, Retool Agents (pilot)
• Use cases: Warehouse operations visibility (HubView), fleet management (Fleet Operations Suite), customer care, self-service operational tooling
• Data sources: Databricks, PostgreSQL, T-SQL, internal and external APIs

How do you govern growth across a decentralized organization?

Flaschenpost, a subsidiary of the Dr. August Oetker Group, runs more than 30 warehouse hubs across Germany. Each one operates with its own staff, local conditions, and pace of growth. Coordinating that network in real time, from replenishment runs, to fleet routing, to customer support, requires operational tooling that can adapt to match.

Before Retool, the business units at flaschenpost had to rely on IT to build or approve almost any new internal tool. As Aron Spohr, CTO at flaschenpost, put it: “I think it’s important to make sure that IT doesn’t stay your bottleneck forever. And the only way to do that is to broaden the number of developers and actually find them in other departments, like in the business units themselves.”

The bottleneck manifested differently across the organization. Niklas Hoffmann, Head of Operations Solutions, described what it looked like on the operations side: “Before Retool, we had too many different tools and process steps that led to a lot of orchestration work.” The Logistics Resource Orchestration team had the same frustrations, but from the other perspective: requests for anything with a user interface had nowhere to go, since Team Lead Stefan Kirschner’s engineers had the backend skills but not the bandwidth or interest in building UIs themselves.

Technical maturity varied widely across flaschenpost’s business units, and a single off-the-shelf vendor tool couldn’t meet the needs of both a logistics specialist in one warehouse and a software engineer in another.

The problem on the surface was too few tools. The real problem was deciding who could build what, and how to keep those custom tools secure. Centralizing every build through a small platform team would have been too slow for a business that runs in real time. Letting each warehouse or business unit solve its own problems independently would have meant losing consistency, security standards, and a shared source of truth across the organization.

Flaschenpost needed a way to let many teams build, without losing control over how they built.

Hundreds of internal users, one governed platform

Rather than maintaining a patchwork of vendor licenses or letting vibe-coded tools spread unsupervised across business units, flaschenpost chose Retool to apply its governance model. “We have some guidelines in place, and we handle the basics in Retool, like creating the resources and the database and API connections,” Aron says. “But apart from that, the business units can feel free to create the apps however they need them.”

For builders, that means IT departments can focus on building the foundation and bottom layers, and operational teams can then build on top of those layers within the guidance of the IT department. Those builds connect directly to flaschenpost’s core data infrastructure (Databricks, PostgreSQL, T-SQL, and a range of internal and external APIs) rather than sitting on top of it as disconnected tools.

Today, 26 developers across flaschenpost’s business units build tools on that foundation for hundreds more internal end users. Six production applications show what the platform looks like in practice:

HubView is the source of truth for real-time KPIs

Deployed across all 30+ warehouse hubs, this app displays real-time operational KPIs on screens throughout each facility. Warehouse managers and floor staff can see live performance data without pulling a report or asking an analyst.

The Fleet Operations Suite manages vehicle maintenance and logistics

Niklas’ team built four interconnected apps to manage flaschenpost’s delivery fleet of 1,500+ vehicles across Germany: vehicle and tour management, a scheduling interface, a real-time status dashboard, and a driver support tool, all connected to both internal databases and external APIs. He says: “We’ve connected multiple databases, APIs internal and also external, and are now able to manage our fleet totally on our own.” For Niklas, this is the real distinction Retool offers over a typical UI builder: “Retool is not only a UI tool, but more a tool with an execution layer. We are using Retool apps as the user-interfacing tool, and also Retool Workflows for the backend orchestration and logic.”

The Customer Care Tool replaced siloed data across legacy systems

Built largely by a single developer, this tool replaced more than 20 pages of legacy systems that customer support agents had previously navigated across multiple platforms. Everything from customer details and order history to voucher creation and a full changelog now runs through a single application.

Beyond these flagship builds, flaschenpost has extended the platform into warehousing operations, (including storage location management and forklift coordination, connected to Databricks and PostgreSQL) and into marketing and HQ operations, where campaign management and task workflows that once required manual coordination now run on the same platform as the warehouse floor. Stefan’s analytics team built a self-service inspection tool that lets operations colleagues look up the information they need directly, rather than routing requests through his engineers.

Scaling AI automation without ungoverned sprawl

flaschenpost felt the automation impact quickly. The Customer Care Tool alone replaced multiple legacy systems that agents had previously needed to context-switch between throughout their working day. Stefan’s inspection tool had a measurable impact within its first month: support requests that had been coming directly to his engineering team dropped sharply, freeing the team to focus on the work only they could do. “That felt so great,” he says. “We were so liberated.”

The 10,000+ workflow executions running monthly on Retool are the backend logic powering flaschenpost’s day-to-day operations, from fleet scheduling to customer care ticket routing. These runs replace manual handoffs across 30+ warehouse locations.

For Aron, these wins are evidence that building on a platform like Retool is “the path to AI and automation,” because the governance and integration work is already in place.

Looking forward: AI-generated tools with a secure foundation

With the governance model, data integrations, and security guardrails already in place, flaschenpost can embrace AI without rebuilding the foundation.

Aron expects more non-IT people building apps and writing code in using AI. He emphasizes, “You have to make sure that you have guardrails and things in place to make sure that security and governance is given, but at the same time allow these people to do these kinds of things.”