Can a biotech or life sciences lab sell its data to AI companies?
In some cases. A working lab records what it tried, what it measured and why it changed course, and most of that is unpublished. Intellectual property decides first: unfiled ideas, sponsor-owned results and patient-derived data all limit what a lab can offer. Method notes and instrument records are often the part that is clear.
What a life sciences lab has that is licensable
| Lab notebooks | Dated entries on setup, observations and next steps, including the failed attempts that rarely reach a paper. |
| Protocols and method revisions | Step lists with the changes made over the years, and the reason for each change. |
| Assay records | Measurements for each run with controls, plate layouts and the notes on why a run was repeated or thrown out. |
| Instrument logs | Run histories, calibration checks and error messages from sequencers, mass spectrometers and imaging systems. |
| Quality and deviation reports | What went off plan, the investigation and the corrective step. Plain, structured writing. |
Where the records live
Labs keep entries in electronic notebooks, protocols in a document system, and raw output on instrument computers or a shared server. Older work may be on paper or on drives that nobody has opened in years.
We ask which systems hold the notebooks and measurements, which years are digital and who is the system owner. Unreadable formats add effort, so we note them early.
Who owns the result
A result generated under a sponsored research agreement, a grant or a collaboration may belong in whole or in part to the sponsor or partner. University labs often have technology transfer rules, and employees may have signed invention assignments. A company lab owns more, but a licensing deal can still trigger obligations to partners.
Patents matter in a different way. Sharing details of an invention before a patent application is filed can harm the chance of protecting it, and the lab's patent counsel should clear anything that touches a pending idea. Take this as general background, since only counsel can advise on your contracts.
Human samples and clinical data
Records that come from human subjects, such as patient-derived samples or trial data, carry consent terms, ethics board conditions and privacy law. They are a separate track from a lab's own method and instrument records, and a lab should assume they are out of scope until its research compliance office says otherwise.
What stays out for a life sciences lab
| Unfiled inventions and pending patent material | Anything that could harm a patent position is held back until counsel signs off. |
| Clinical and patient-derived data | Trial results, subject identifiers and consent-limited samples are not part of a first package. |
| Sponsor and partner results | Data owned or restricted under a collaboration or grant agreement stays with the agreement. |
| Proprietary sequences and compound structures | Where the value of the company lies in a specific molecule or strain, those details are removed or left out. |
More on this: how personal details are removed, who carries the risk and how you get paid.
A worked case
A mid-sized biotech with ninety staff has twelve years of electronic notebook entries and instrument run logs. Its patent counsel marks which projects are cleared. The first package covers method notes, calibration records and deviation reports from those projects only, with sequences and compound identifiers removed. Clinical work is excluded. The company's leadership reviews the sample before deciding.
Questions owners ask about this
Can we license data that is already in a published paper?
The paper's results are public, but supplementary raw files may carry journal or sponsor terms. The unpublished notes around the paper are usually more useful and need their own check.
Does a university lab need approval first?
Usually, yes. The technology transfer office and the sponsor agreements decide who can license, so they are contacted before anything is prepared.
What about failed experiments?
Negative and failed runs are among the most useful records because they are rarely published. The same ownership and patent checks apply to them.
Are older paper notebooks worth scanning?
Sometimes. Handwriting and sketches are costly to digitise, so we usually start with what is already in a digital system.
Didn't find your question? Ask us directly.
Keep reading
Records to look at first
- SOPs and knowledge bases: Procedures with their edit history and comments; published help articles add nothing.
- Lab notebooks and experiment records: Settings tied to results, failures included, with sponsor-owned work and patient data kept out.
- Google Drive export: How to keep the comments and the version history when documents leave Google Drive.
- Labeled tables and datasets: Rows, a result column and a data dictionary: when a table can be licensed.
- Word, Excel and PowerPoint files: Spreadsheets, reports and decks in their original format, and the details hidden inside them.
Other industries
- Pharma companies: Compounds and patients stay out; procedures, deviations and training records can go in.
- Universities and research labs: Theses, protocols and grant drafts, and the ownership questions to answer before listing any.