Zürich · Licensing company archives
Your archive has value.
E-mail, project folders, tickets and accounting records from many years: everyday material for you, and for AI labs exactly the data the open internet lacks. Datalpin finds a lab that will pay for it, removes names and other identifiers and sub-licenses it to the lab for a limited time. You receive a payout – and the rights to your data stay with you.Years of e-mail, project folders and tickets are valuable to AI labs. If a lab is interested, we license them on, de-identified and for a limited time – you receive a payout and keep the rights to your data.
- No file leaves your company, until a lab confirms interest and you have signed.No file leaves your company before you sign
- Every deal approved individually, in writing, by you.Every deal approved by you
- Payout – Datalpin pays you your amount as soon as the lab has paid. Our share comes out of the lab's payment and is shown openly in the offer – you never receive an invoice from us.Paid out once the lab pays – never an invoice
What Datalpin does
The large AI labs have largely exhausted the public internet. What they lack is real working data from companies – often sitting in your systems for decades.
How Datalpin works: we describe your holdings without company or personal names, gauge lab interest and negotiate a time-limited licence. Datalpin started in Zürich in 2026, so what we say about prices and timelines comes from comparable providers, not from our own deals.
For companies with an archive
Companies from roughly 50 staff with a long digital history and short decision-making paths. An archive is an asset – in a wind-down or a succession, too, or when a start-up needs cash.
Our role
Licensor to the lab, not buyer of your data. For each deal you approve, you grant Datalpin a time-limited right of use; Datalpin sub-licenses the de-identified dataset to the lab, collects the lab's payment and pays you the agreed amount. Our share is shown openly in the offer.
How it works
From your archive to the payout into your account.From your archive to the payout. Tap a step for details.
Your archive
-
Enquiry
Company, size, years of digital records, systems in use – two minutes. We reply within two business days.
-
Gauge interest
We describe your holdings without company or personal names and ask the labs. Under an NDA if you wish.
-
Offer and agreement
You see which data qualifies and what amount you will receive. Not a single file moves before you sign.
-
Review and de-identification
You export the data yourself; we remove names and identifiers and show you a sample to approve.
-
Licence and payout
The lab pays Datalpin, and we pay the agreed amount into your account. We delete our working copies once the dataset is delivered and confirm the deletion to you in writing.
Payout
Our estimate, not a commitment, based on comparable providers: roughly two to three weeks from form to payout, provided a lab accepts.Usually two to three weeks to the payout if a lab says yes – an estimate, not a promise.
How we handle your data
No certification badges – six principles that apply before you show us anything, and that are written into the agreement.Six principles, binding in the contract. Tap a point for details.
Your rights stay with you
The rights to your data stay with you; Datalpin only receives the usage rights that a deal you approve requires. Only what you approve in writing, and what you are entitled to license, is licensed. We aim for one-year licences; the licence sets how long the lab may keep the dataset and what it may use it for. What a model has learned in that time remains. Until you sign, you can walk away at any time.
We remove personal details
Before a lab sees a single line, we strip names, e-mail addresses, phone numbers, customer and account identifiers, check what remains for re-identification risk and document every step. No method is perfect – which is why you see the result before you approve, and you decide what stays excluded. What one line looks like before and after: "Dear Ms Meier, I am sending invoice no. 48213 to m.meier@…" → "Dear [PERSON], I am sending invoice no. [IDENTIFIER] to [E-MAIL]".
Swiss company, Swiss and EU data protection
Datalpin is operated by Zero Knowledge GmbH in Zürich and is subject to the Swiss Federal Act on Data Protection (FADP). Where your archive holds EU data or your company is based in Germany or Austria, the DPA also covers the GDPR; transfers to Switzerland are covered by the EU adequacy decision. Where the dataset goes to a lab outside Switzerland and the EEA, for example in the US, the licence agreement covers that transfer with recognised safeguards such as standard contractual clauses. Where review and de-identification take place, and who operates the infrastructure, is written into the agreement. Before their licence agreement is signed, labs see only a redacted preview that you have approved, and afterwards receive nothing but the de-identified, licensed dataset.
Encrypted, logged, deleted – by contract
The agreement sets out encrypted transfer and storage and logged access. Once the dataset is delivered, we delete our working copies and confirm the deletion to you in writing; the retention period is in the agreement, as is the deadline for notifying you of a security incident.
You decide what stays out
Any system, folder, record type or time period can be excluded – HR files, client data, contracts with confidentiality clauses; we check with you up front what is protected. The exclusions go into the agreement before we get access; what is excluded is not analysed. Every licence prohibits publication, resale and onward transfer of the dataset. You learn which lab is bidding before you sign, and you can decline.
A payout, never an invoice
The lab pays Datalpin; we pay you the amount set out in your agreement as soon as the payment has arrived. Our share comes out of the lab's payment and is shown in the offer. If no licence comes about, you pay nothing – not for the enquiry, review, offer or agreement. Your internal export effort is yours; we help with it.
Which data is in demand
Labs want data that shows real work: long histories, real decisions, closed cases with a known outcome. The longer and more complete, the more labs pay for it – and the higher your amount. A typical fit: 50 or more staff, ten or more years of digital history and short decision-making paths. Focus on Switzerland, Germany and Austria, data in German or English; other countries and languages on request.Labs want long, complete records of real work – typical for companies with 50+ staff and ten or more years of digital records.
What qualifies
- E-mail archives
- Documents & drives
- Tickets & support
- CRM & sales
- Code & repositories
- CAD & drawings
- Project files
- Accounting & ERP
- Telemetry & logs
E-mail and communication
Outlook, Exchange, Gmail, Teams, Slack: customer correspondence over years. Long histories are often worth more to labs than sheer volume.
Documents and projects
SharePoint, Drive, file servers: quotes, contracts, reports, minutes, plans, drawings, whole project folders.
Tickets, support and CRM
Zendesk, Jira, ServiceNow, Salesforce, HubSpot: resolved requests, complaints, customer histories – with their outcome.
Orders and accounting
SAP, Abacus, Bexio: orders, invoices, dunning runs – records where you can see how things ended.
Code and specialist data
Your own code, CAD drawings, bills of materials, measurement series, sensor and machine data from your own plant, owned datasets, reinforcement-learning environments.
Sectors where archives are most in demand
- Legal
- Finance & accounting
- Software
- E-commerce
- Manufacturing
- Construction & engineering
- Logistics
- Consulting
- Healthcare
- Semiconductors
- Robotics
Legal, finance and healthcare only where professional, banking or patient secrecy does not stand in the way – we check that with you first.
Data the internet doesn't have
Real business processes from Switzerland, Germany and Austria, in several languages. Datalpin licenses de-identified datasets with a written licence, a provenance record and a de-identification log – useful input for your own training-data documentation.De-identified business data from Switzerland, Germany and Austria – with a written licence and a provenance record.
- What you get A redacted preview before signing, provenance details as far as the licence allows, a de-identification log. Delivered as structured exports (e.g. JSONL, Parquet, converted mailboxes) with a datasheet covering period, volume, languages and systems. Licence with a target term of one year and a renewal option; permitted use: training and evaluation.
- How to ask Send a short e-mail to hello@datalpin.ch with your team and the domains, languages and formats you need. Teams with a clear research or training need go on our request list and hear from us as soon as matching datasets are in review – without company names until an agreement is in place.
Who is behind it
There are three of us, and we handle everything ourselves – from the first e-mail to the payout. You talk to us directly, not to a sales desk.There are three of us, and we look after you personally – from the first e-mail to the payout.
Co-Founder · Product and technology
Completing an MSc in Computer Science at ETH Zurich, with a background in medicine. Previously an associate at Boston Consulting Group, a working student in software engineering (DevOps team) at KPMG Switzerland and co-founder of the AI and compliance start-up Green-Biz. Builds Datalpin’s review and de-identification tooling.
Co-Founder · Strategy and deals
Studied business administration, management and sociology at the University of St. Gallen (HSG). Spent five years as a general partner at Dialectic and at the venture fund Vine Ventures, and co-founded the Zurich blockchain hub Trust Square, among other ventures. Owns lab relationships and the structure of every licence.
Co-Founder · Legal and partnerships
Attorney-at-law (MLaw, University of Zurich) and entrepreneur. Co-founder of Obligate, a platform for digital securities, and a member of the Lisk Foundation Council; currently on the LSE Executive Master in Behavioural Science. Leads first contact with companies and owns the contracts.
What you probably want to ask
Is this legal?
Yes. You can license data you are entitled to make available. We comply with the Swiss Federal Act on Data Protection (FADP) and, where it applies, the GDPR. Before a lab receives any data, we remove names and other identifiers, and you approve the result. We check with you up front what is covered by professional secrecy, confidentiality obligations or third-party rights; whatever you exclude, such as HR files or client data, goes into the agreement. Every licence sets out purpose and term and prohibits publication and onward transfer. Your duties as data controller towards staff and customers stay with you; we document every step and confirm the deletion of our working copies in writing.
Is the data anonymous?
We deliberately say "de-identified", not "anonymous": no method offers a hundred-percent guarantee. That is why you see a sample before release and check with us whether the dataset could point to your company. The lab learns your company name and the data’s origin only with the agreement.
What is my archive worth?
That depends on the years covered, completeness, sector and whether your data shows outcomes: completed projects, resolved tickets, paid invoices. Comparable brokers abroad advertise five-figure sums and more in US dollars; where your archive sits, only the review will show. We promise no number – you get an honest assessment and see the amount we will pay you in the offer, before you decide.
How and when do I get paid?
By bank transfer, as soon as the lab has paid Datalpin. You see the amount beforehand in the offer, and it is binding in the agreement. Based on comparable providers, we estimate the payout could come roughly two to three weeks after your enquiry, provided a lab accepts. If the same holdings are licensed again later with your consent, you get paid again.
Is the licence exclusive?
We negotiate non-exclusive licences by default: with your consent the same holdings can be licensed again later, and you get paid for each deal. If a lab wants exclusivity, it is agreed separately and paid at a higher rate – only with your consent.
How much work is this for my team?
Your IT does the export; instructions, scripts and a checklist come from us. For common systems this usually takes a few hours; legacy systems can take longer, and we help with those.
Who are the buyers?
AI labs and research teams that train or evaluate models – no data resellers, no publication. Every licence fixes permitted use, term and a ban on onward transfer. Whether a bidding lab counts as your competitor is your call: you learn who is bidding before you sign, and you can decline.
What is your archive worth?
Two minutes, no obligation, no files from your archive – just the basics about your company and a contact address. We reply within two business days with a first, non-binding view on whether your archive is of interest to labs – or tell you honestly if it is not a fit. The amount we will pay you is set out in the offer. Know a company with a large archive? Put us in touch, as long as that person is happy to hear from us – one line by e-mail is enough.Two minutes, no obligation, no files from your archive. We reply within two business days with a first assessment.
hello@datalpin.ch