Skip to content
Fluency
←Back to Blog
Fluency Team

Fluency Team

SOC 2 Type II at Fluency: What the Audit Covers and How Work Data Is Protected

Independent auditors tested the security controls on Fluency's hosted service for a full year, and Fluency now holds a SOC 2 Type II attestation.

Security reviewers will assess Fluency differently from other SaaS tools. For example, a CRM stores only the records your team chooses to type into it. Fluency, a work intelligence platform, runs a desktop agent on employee computers to find which tasks are worth automating first. The agent observes work as it happens, so it can come across things employees never meant to share with a vendor, like a personal email or a banking tab.

A security review has to cover what the agent collects, along with how Fluency stores it. Below, we cover what the attestation tested and where its scope ends, what the agent keeps and discards, how Fluency stores, protects, and deletes your data, and how to request the full report.

What the SOC 2 Type II attestation covers

Sensiba LLP, a certified public accounting firm, issued Fluency's SOC 2 Type II attestation for March 1, 2025, to February 28, 2026, covering the Security, Availability, and Confidentiality criteria. Its opinion is dated March 27, 2026, and its SOC 2 attestation letter is public on the Trust Center.

The letter names the scope as Fluency's software-as-a-service system, which includes the desktop agent. The full report is available on request through the Trust Center.

The current attestation follows a Type I from AssuranceLab CPAs LLC, as of December 18, 2024. Sensiba is examining the next period now, from March 1, 2026, to February 28, 2027, with a report expected in March 2027.

What Fluency keeps and what it discards on the device

Fluency keeps only work activity, meaning the record of how a process runs across systems and apps. Here’s an example: An accounts payable clerk approves supplier invoices in your finance system, with a personal email tab open in the background.

The Fluency agent on the clerk's computer classifies each event before anything is sent. It discards the personal email tab as non-work activity. You can also block any app from the agent, so nothing in it leaves the computer.

What leaves the computer is a set of screenshots of the clerk's work, with sensitive fields redacted, blocked apps excluded and non-work activity removed. Fluency's AI system then works out the invoice approval step from those screenshots. It identifies which systems and apps the step passes through, such as the finance system and work email, and calculates how often the step runs, how long it takes, how many people it involves, and how often an invoice needs correction.

Redaction removes personal data, passwords, and other sensitive fields automatically. Your organization can also decide what else counts as sensitive, such as supplier bank details, and have those redacted too.

Where the data is stored and how it is protected

Once the screenshots leave the clerk’s computer, Fluency stores them in the region your organization chooses: either Australia, the US, or Singapore. All processing of the record happens in that region, too, including AI processing. Fluency plans to add more regions as customers need them.

Data is encrypted on its way from the clerk's computer to Fluency and while it's stored. Only Fluency staff whose role requires it can open customer data, and each access is logged, showing who opened the approval record and when.

Between annual audits, Fluency monitors these protections with Vanta, a compliance automation platform, and publishes the current status of each control. An independent security firm also runs penetration tests on the platform at least once a year, most recently in June 2026.

These are public on Fluency's Trust Center:

  • The control list, with the current status of each control, including encryption key access, production database access, and the removal of access when a Fluency employee leaves
  • The list of subprocessors, showing every third party Fluency uses, including its AI providers
  • Fluency's data processing agreements with Anthropic and OpenAI, on the resources page

What Fluency does not do with the data

Fluency observes work, not workers. It looks at each process across everyone who does it, rebuilds it digitally, and shows where it breaks down and how to improve it. It doesn't track how long any one employee takes or compare employees with each other.

In the invoicing example, Fluency estimates how much time automating the approval step would save, based on how long approvals take and how often invoices are sent back to be corrected. Once the automation is live, it measures the step again to show what changed. Fluency aggregates the data for each automation candidate across the whole team, so it doesn't attribute activity to individuals or rank employees.

Fluency builds a separate model of each customer's work, a map of its processes, systems, and handoffs, from that customer's data alone. It never pools data from several customers into a shared model, so details of how your organization works, such as your suppliers or approval steps, can't show up in another customer's results. Customer data is also never used to train third-party AI models, including those of the AI providers on Fluency's subprocessor list.

How Fluency deletes customer data

When your contract ends, Fluency deletes your data, including the screenshots and the analysis built from them. You can ask for deletion at any point before then and get confirmation in writing if you need a record of it.

Your organization can also set a shorter retention period, for example, to match an internal policy that limits how long vendors keep employee activity data.

Deletion at contract end is one of the security controls Fluency monitors in Vanta, so you can check its current status on the control list.

GDPR and HIPAA compliance status

Fluency is working toward compliance with the General Data Protection Regulation (GDPR) and the Health Insurance Portability and Accountability Act (HIPAA). If your review depends on one of them, ask the team for the current status when you request the report.

Request the report or talk to the team

Request the full SOC 2 Type II report through the Fluency Trust Center, along with any of these documents:

  • Penetration test report
  • Redaction method statement
  • Data classification, handling, and retention policy
  • Information security policy
  • Business continuity and disaster recovery statement
  • Network endpoints IT teams allow for the desktop agent
  • High-level network and connectivity diagram
  • Certificate of cyber and professional liability insurance

If your legal team needs its own terms, custom master service agreements (MSAs) are available on Fluency's Enterprise tier. Contact the team to set one up.

SOC 2 and data security FAQs

Is Fluency SOC 2 compliant?

Yes. Fluency holds a SOC 2 Type II attestation from Sensiba LLP for the Security, Availability, and Confidentiality criteria, covering March 1, 2025, to February 28, 2026. The attestation is issued to Fluency's legal entity, Farmor Software Pty Ltd. The next Type II examination covers March 1, 2026, to February 28, 2027.

What is the difference between SOC 2 Type 1 and Type 2?

A Type 1 examination checks whether controls are designed correctly on one date. A Type 2 examination also tests whether they kept working across an audit period. Fluency's Type 1 attestation is as of December 18, 2024. Its first Type 2 audit period ran from March 1, 2025, to February 28, 2026.

Is SOC 2 a certification?

No. Vendors often say "SOC 2 certification" or "SOC 2 Type 2 certified," but a SOC 2 report is an attestation issued by a licensed CPA firm. The firm examines the controls under the attestation standards of the American Institute of Certified Public Accountants (AICPA) and issues an opinion. Sensiba LLP issued the opinion on Fluency's controls.

Where does Fluency store customer data?

In Australia, the US, or Singapore, whichever region the customer chooses. Storage and processing happen in the chosen region, AI processing included. Data is encrypted at rest and in transit.

Does Fluency train AI models on customer data?

Not across customers. Each customer gets a separate model, and customer data is never used to train third-party AI models.

Is Fluency employee monitoring software?

No. Fluency observes work, not workers. It doesn't log keystrokes, record video of the screen, attribute activity to an individual, or rank employees. Activity in apps you block from Fluency never leaves the device, and the agent discards non-work activity before anything is sent. Fluency analyzes work at the workflow level, across a team.

Useful AI starts with understanding the work.

Fluency shows you where AI will return value before you deploy it.

Useful AI starts with understanding the work.