Security and data governance

Where your data goes,
in plain language.

Adept sits between two groups of people who both have something to lose: scientists running unpublished work, and manufacturers whose product performance data is commercially sensitive. This page explains exactly what we capture, what we never see, and what each side can and cannot access.

If you are a scientist

Your science is not the product.

  • Your run records belong to you and export to your ELN at any time.
  • Your results, sample identities, and project context are not shared with manufacturers.
  • Your data is not retained by third-party model providers and is not used to train third-party models.
  • You can request deletion of your account and your records.
If you are a manufacturer

Your performance data stays yours.

  • Strong multi-tenant isolation, with a per-client siloed knowledge base and customer analytics.
  • Customer data is logically isolated by deployment and is not shared across customers.
  • Transcripts and usage stats are used only to deliver and improve your own deployment.
  • At contract end your data is returned or deleted, subject to the agreed retention window.

What we capture, and what we do not.

Adept records how a protocol was executed, along with the results you choose to enter for your own recordkeeping. It does not record what the experiment was for or what you concluded from it.

We capture

  • Which protocol was run, and which steps were completed
  • Timing and dwell time per step
  • Where a run was paused, stalled, or abandoned
  • Deviations from the documented protocol
  • Questions asked of the assistant, as text transcripts
  • Which kit or SKU the protocol belongs to
  • Your raw experimental results as you input them into the app for recordkeeping

We do not capture

  • Your experimental conclusions or insights
  • Sample identities, patient data, or subject information
  • Sequence data, genomic data, or analysis output
  • Your hypothesis, grant, or project context
  • Continuous audio or video of your bench
  • Anything from other applications on your device

How it is protected.

Encryption

Data is encrypted in transit and at rest.

Segregation

Each manufacturer's bench-level data is processed in a logically segregated environment, not pooled by default.

Access control

Role-based access with multi-factor authentication for privileged accounts, and least-privilege access for staff.

Zero 3rd-party retention

Customer and scientist data is not retained by third-party model providers and is not used to train third-party foundation models.

The questions we get asked.

Can a manufacturer see which lab or which person ran their kit?

No. The data on the manufacturer dashboards is aggregated and anonymized to protect individual user identity.

Can a competitor see our product's failure rates?

No. Your data is logically isolated by deployment and is not shared across customers in any form, aggregated or otherwise. We do not operate a cross-manufacturer benchmarking product. The analytics you see are built from your own deployment only.

Does the app listen continuously?

When you have paused the mic, the app is listening only for the wake word "Adept." When the mic is active, it is listening for the commands used to navigate the app, such as next step, go back, question, comment, and repeat.

Still have questions?

Ask us something specific. A person who understands the architecture will answer, not a form letter.

Contact us