Email support: 70% automated
A purpose-built workflow processed around 200 incoming support emails each day — with clear rules, approvals and human oversight.
Discuss a similar projectAnonymised case study · Implemented in 2024
A deliberate combination of automation and human oversight.
Many recurring enquiries. High manual workload.
The retailer received about 200 support emails per day. Different issues required reliable categorisation, suitable responses and staff approval in certain cases.
~200 emails per day
A high volume of recurring customer enquiries.
Custom rules
Responses needed to follow internal guidelines and examples.
Data sovereignty
Data processing and AI infrastructure had to remain within the EU.
An intelligent workflow with defined decision points
The automation collected emails, stored them in a dedicated database, classified requests according to business-specific rules and drafted suitable replies.
The result: 70 / 20 / 10
Not every email was sent without review. Decisions depended on the request and the retailer's rules.
Automatic
Suitable messages were answered without additional approval.
With human approval
Staff made the decision in sensitive or unclear cases.
Manual
Unknown or unsupported requests remained with the team.
The infrastructure remained with the client
The solution was configured according to the client's requirements, with data processing and AI services within the EU. The database and LLM API ran in accounts owned by the client.
Qbumps designed and implemented the solution and handed it over in full. The retailer retained control over access, infrastructure and ongoing operating costs.
Clear data ownership
- EU-based processing
- Own accounts and access
- Dedicated customer database
- Complete handover
Technical implementation alone does not replace legal data protection review.
How many routine enquiries could your team hand over?
Let's explore where automation makes sense and where people should decide.