AI YOU CAN ACTUALLY SWITCH ON
Every Leapsome AI feature is built compliance-first, designed around European data rules from the start. Your team ships faster, and your DPO still sleeps at night.
drift
EU hosting for platform data
Hosted and processed in Germany
No model training on your data
Nothing you enter trains a foundation model
Admin on/off controls
Disable any AI feature, per role, at any time
THE HONEST ANSWERS
The six questions every HR lead asks us, answered directly.
Will our employees' data train AI models?
No foundation model is trained on your data. That covers everything: prompts, outputs, and any other customer content you put into Leapsome. Ours or a provider's model, it makes no difference. Nothing you or your team enters is absorbed into how a model works, and none of it can resurface in another customer's results.
Where does our data live?
Your Leapsome data is hosted and processed in the EU, in Germany. A few AI features rely on a specialised provider that processes data outside the EU. That's normal for enterprise-grade AI infrastructure, and each partner is carefully vetted: bound by standard contractual clauses, backed by a completed transfer risk assessment, and given only the exact input needed to produce your result, never a copy of your account. See exactly who, and where: leapsome.com/legal/subprocessors
What happens to a prompt after I send it?
It's sent to the AI provider that powers the feature, processed to generate your result, and sent straight back to you. The provider retains it for up to 30 days, purely for abuse monitoring and quality checks under strict access controls, then deletes it. It's never used to train a model, and never kept to profile your organisation.
Can we turn it off?
Yes, feature by feature and role by role. AI features arrive switched on. A super admin can disable any or all of them at any time, and decide which roles get access to what, in Settings > Employees > Feature access matrix. Switch everything off and Leapsome is still the full HR platform. Click through the actual settings screen before you decide anything: app.storylane.io/demo/e8vu3vujato4
Does the EU AI Act make this complicated?
Not for the way this is built. Leapsome AI drafts and summarises. It doesn't evaluate people or make decisions, so it sits outside the Act's high-risk categories. We still apply its transparency standards, so AI-assisted content is always labelled.
Your side is light too: as the organisation switching features on, you mainly need to tell employees when they're seeing AI-generated content, and keep a person in the loop. Both are built in. The full legal reasoning sits in the Trust Center, and we'll walk your DPO through it.
Your side is light too: as the organisation switching features on, you mainly need to tell employees when they're seeing AI-generated content, and keep a person in the loop. Both are built in. The full legal reasoning sits in the Trust Center, and we'll walk your DPO through it.
Can Leapsome connect to other AI tools, like agents using MCP or an API?
Yes. Leapsome exposes both an API and an MCP server, so you can connect it to the AI tools and agents your team already relies on. You decide exactly what gets connected and what it can see. Once you make that connection, how the other side handles the data is between you and that provider, so that part lives in your hands, not ours.
PLAIN AND SPECIFIC
Host and process your platform data in the EU
Encrypt everything, in transit and at rest
Let admins switch any AI feature off, for any role, at any time
Respect the permissions you already set, so AI sees only what that person can see
Use AI prompts only to deliver the feature you asked for
Hand your data to a provider that trains on it
Sell your data or hand it to advertisers
Make AI hard to find, or hard to switch off
Keep prompts to profile your organisation
Let a provider use your data for their own ends
A HUMAN ALWAYS HAS THE LAST WORD
Every AI output here is a draft with no standing until someone reads it, changes it, and makes it theirs. And when an agent acts on your behalf, it proposes and you confirm.
drafts the review from goals, feedback and your notes
→ you edit it before anyone else sees it
turns the call into a summary, agenda and action items
→ you confirm them before they go out
summarises thousands of comments into themes and sentiment
→ you decide what they mean and what to do
"We built Leapsome's AI the way we'd want it built for our own team's data: carefully, with a person always checking the work. Our mission is to make life more fulfilling for everyone, and that starts with people trusting the tools we hand them."
Jenny Podewils & Kajetan Armansperg
Co-founders & Co-CEOs, Leapsome
SOC 2 Type II
Independently audited controls
GDPR
Built for EU data protection
ISO 27001
Certified information security
EU hosting
Platform data hosted in Germany
FOR THE PEOPLE WHO NEED PROOF
The badges your security team screenshots for the vendor review.
TWO WAYS FORWARD
EVALUATING LEAPSOME
Walk through the product at your own pace, or get a personalised demo built around the features you actually care about.
Reviewers with tough questions: book a 30-minute call to ask them. No slides unless you ask for them.
ALREADY ON LEAPSOME
1
Click through the real settings screen, off switch included, in the interactive demo. No login needed.
2
In Settings › Employees › Feature access matrix, enable what you want, for the roles you choose.
3
Ask your customer success manager for an enablement session, including the compliance answers your own stakeholders will want.
FOR PROCUREMENT & IT SECURITY
Audit reports, DPA, subprocessor list, AI transparency documentation, and security policies, ready for your vendor review.