Case studies
Examples of projects we've worked on. Client identities are kept confidential unless explicitly agreed otherwise.
Inbound document processing for a logistics SME
A logistics company with 25 staff was processing supplier invoices and delivery confirmations manually. Documents arrived in different formats across email, PDF, and a web portal. The team spent several hours daily extracting, validating, and entering data into their ERP.
We built an agent that monitors inbound channels, classifies documents, extracts relevant fields, validates against expected data, and routes results to the ERP automatically. Documents that fall outside expected parameters get flagged for manual review rather than processed blindly.
Tier-1 support automation for a SaaS company
A Berlin-based SaaS company with a growing customer base found their support team spending the majority of their time on repetitive tier-1 questions — password resets, billing queries, basic feature explanations. These didn't require expertise, but they blocked time for more complex issues.
We built a support agent connected to their knowledge base, their billing system, and their ticketing platform. The agent handles common requests, processes straightforward account actions, and escalates to the human team when queries require judgement or account-level decisions. GDPR handling of customer data was designed into the agent architecture from the start.
Employee onboarding workflow for a mid-size consultancy
A professional services firm hiring 15–20 new employees per quarter found that onboarding coordination was consuming significant HR and operations team time. Account setup, equipment requests, access provisioning, and induction scheduling were spread across five different tools and required constant manual follow-up.
We built an internal agent that orchestrates the full onboarding sequence from offer letter confirmation to first-day readiness — triggering actions in each relevant system, chasing outstanding items, and keeping the relevant team members informed. The agent handles the coordination; humans still make the decisions that require judgement.
Projects that didn't proceed — and why
Not every discovery call leads to a project. Some workflows aren't suitable for automation at a given stage. Others require more foundational work (like documenting the process properly) before an agent makes sense.
We turned down or delayed projects because: the underlying process was too inconsistent, the expected data quality was too low, the integration environment was too constrained, or the expected ROI didn't justify the build cost at that team size.
This is part of how we work — telling you when something isn't the right fit is more useful than building something that won't perform.
The examples above ranged from 3 to 8 weeks from kick-off to deployment. Timeline depends on integration complexity, the number of edge cases to handle, and how quickly we can access test data from the client's systems.
Only with explicit written agreement from the client. Most clients prefer anonymity. If you'd like to speak with a previous client about their experience, contact us — we can try to arrange an introduction if appropriate.
Our primary market is Germany. We've done projects with Germany-based clients who have operations elsewhere in the EU. Work is conducted under German law and contracts are in German or English depending on client preference.
Your workflow might look similar — or completely different.
The discovery call is where we figure out what's actually automatable in your specific context. Start there.
Book a discovery call →