This risk and regulatory consultancy in the United States has 210 people. Many of its client contracts limit where client data can go, and several rule out sending it to third-party AI providers.
The firm now runs Hebno on top of open-weight models it hosts in its own data centre. Client data stays inside the firm, and most of the firm uses Hebno every week.
“Client data stays on our own servers. That’s what let me say yes to AI, and it’s what our clients’ security teams want to hear.”
The challenge
The firm had blocked public AI tools, but some consultants used them on personal devices anyway. There was no record of what data went where, and no clear answer when client security reviews asked about AI. Meanwhile, competitors were adopting AI faster.
The firm considered building its own tool. That project was scoped at close to a year and several engineers.
Why Hebno
Hebno works as the interface on top of the models the firm hosts itself. Client-confidential work runs only on those local models, so client data does not leave the firm.
Single sign-on, existing permissions and usage logs were ready for IT and compliance from the start.
How the team uses Hebno
Consultants use Hebno for regulatory research, document review and client deliverables. Teams also use it to search internal knowledge and to answer client security questionnaires.
The result
The firm went live in under two months, compared with the year an internal build would have taken. Most of the firm was using Hebno weekly within the first quarter, and each person saves around 7 hours a week.
