Security and policies

Security and privacy

Research offices and faculty use Atom to find funding and prepare proposals, so it holds unpublished research and draft applications. This page describes how we protect that material: where it's stored, who can access it, how AI is used with it, and how we respond if something goes wrong.

Data isolation

Each institution on Atom is its own group, and access is enforced in the database as well as in the application. When you use Atom, Postgres row-level security policies limit your requests to the records your institution and role allow. Background jobs that run without a signed-in user, such as email digests, limit each query to the person the email is for. A user at one institution can't read another institution's data. Within your institution, other researchers can't open your proposals unless you add them to the team; owners and admins can, as described in roles and permissions.

  • Proposals are visible to their team and to your institution's administrators.
  • Department scoping. Your institution can limit administrators to their own departments, and keep proposals read-only for administrators who aren't on the team. See roles and permissions.
  • Files. Proposal documents and files you upload are stored privately with no public link, and every download is checked against the same permissions.

Encryption and infrastructure

  • All connections are encrypted with TLS 1.2 or higher, and browsers are instructed to connect over HTTPS only.
  • Your data is stored with Supabase, and the application runs on Vercel. Both are established cloud providers with SOC 2 Type 2 reports, and they host Atom for us as service providers.
  • The database, its backups, and file storage are encrypted at rest.

Access controls

  • Single sign-on. Institutions can connect Atom to their identity provider over SAML, so campus password, multi-factor, and offboarding policies apply to Atom. Your IT team sets this up during onboarding.
  • Roles. Member, admin, and owner permissions are enforced by database policies and server-side checks. See roles and permissions.
  • Incognito Mode. Researchers can hide their profile, search activity, and analytics from their administrators in profile settings.

How AI uses your data

  • We don't use your proposals, documents, or chats to train AI models, and neither do our AI providers.
  • We use providers' paid business APIs, not consumer apps or free tiers, and each request includes only the information that task needs.
  • We have zero data retention agreements with our AI providers, so they don't store what we send them.
  • Our AI Policy lists the providers we work with and how they handle data.

Retention and deletion

We keep your information while your account is active. If you ask us to delete your account, we delete or de-identify your personal information within 30 days, unless we're required by law to keep it. To make a request, email ham@atomgrants.com. The Privacy Policy has the full details.

Incident response

We maintain a written incident response process. If we confirm a security incident that affects your institution's data, we'll notify your institution without undue delay, with what happened, what we're doing about it, and any steps you should take. To report a suspected issue, email support@atomgrants.com.

Our approach to AI

We don't think AI should write grant proposals. The scientific argument, and the judgment behind it, belongs to the researcher. Atom is built to help with the work around the writing: finding the right opportunities, pulling requirements out of long solicitations, and checking budgets and drafts against the funder's rules. Everything Atom produces is a draft or a suggestion for you to review, edit, or discard.

Questions

If your faculty, department chairs, or IT team have questions this page doesn't answer, we're happy to talk them through. Book a call with our co-founder, Hamilton Evans, or email support@atomgrants.com.

Last updated: November 18, 2026.