LexBox scaled its litigation AI to production on Opper
- Industry
- Legal tech
- throughput headroom
- 30×
“Going direct to one provider, we kept hitting rate limits. With Opper we got the headroom we needed, and data sovereignty guarantees we could take to law firms.”
About LexBox
Litigation support for Nordic law firms, with lawyers in the loop.
LexBox is a Swedish legal-tech company whose AI platform supports litigation teams at Nordic law firms. It reads case documents, reasons across them, and produces structured, auditable output that lawyers review and act on. In the months after launch, it onboarded a growing roster of firms.
The Challenge
Long case files, heavy output, and a single-provider ceiling.
Litigation and dispute resolution work is compute-intensive. A single run reads long case files and generates detailed structured output, and with a growing roster of firms running workflows daily that volume compounds fast.
LexBox started with a direct integration to a single provider and outgrew it: one project, one region, and a default quota of 200,000 output tokens per minute that throttled the platform as demand grew. In legal work, rate-limit errors break trust. On top of throughput, law firms needed certainty about where their data goes.
The Solution
Thirty times the throughput, through one EU-hosted AI gateway.
LexBox moved its inference to Opper and routes frontier models through a single EU-hosted AI gateway, with no infrastructure to run themselves. Instead of one project's quota, the platform draws on pooled capacity across providers and EU regions: 200,000 tokens per minute became 6 million, removing the ceiling for a token-hungry workload.
- EU-hosted, with the gateway running in AWS Stockholm and EU model routes, under one DPA that covers every model.
- Automatic fallback across providers, so a degraded endpoint fails over silently instead of surfacing to lawyers mid-workflow.
Results
From first calls to production in days.
LexBox went from first test calls to production in days, with live traffic ramping steadily through the first week.
- Rate limits no longer a constraint, with throughput headroom to spare as more firms come on board.
- Room to grow, with data residency handled at the gateway, so the team can focus on the product.