Deepfaic

Platform

Not a model. A detection platform.

Deepfaic is a configurable detection platform: six specialised models, a decision engine tuned per use case, and a toolkit that deploys wherever your data already lives. New generators appear every few weeks, so a single model is never enough. The platform is built to swap parts without changing your integration.

Book a demo

Platform — How it works

Models. Decision engine. Toolkit.

01

Six specialised models.

Face, voice, image, document and temporal detectors trained on more than 15 million real and synthetic samples, refreshed as new generators appear, with research pipelines at A*STAR and NTU.

02

A decision engine you configure.

Each use case gets its own profile: which models run, how their scores combine and where the thresholds sit, tuned on a sample of your own data before go-live.

03

A toolkit that fits your stack.

API-first. Runs on-premises, in your cloud, in ours, or as a sealed appliance. Existing processes call it and get a verdict back. Nothing to redesign.

Deploys as

  • On-premises
  • Your cloud
  • Deepfaic cloud
  • Appliance

Deploys where your data is

On-premises

Containers in your own data centre. No internet egress, no data leaving the network.

Your cloud

Inside your own AWS, Azure or Google Cloud tenancy. Deepfaic is never the data processor.

Deepfaic cloud

Hosted by Deepfaic for teams that prefer SaaS, with the same API and profiles.

Appliance

A sealed hardware unit for air-gapped sites, delivered pre-configured and sized to the workload.

Configured per use case

01

Profiles

Each use case gets a profile: which models run, how their scores combine and where the thresholds sit. Profiles are tuned on a sample of your own data before go-live.

02

Explainability

Every verdict carries the region, the timestamp and the likely technique, so an analyst or an agent can act on it and an investigator can rely on it.

03

Integration

Called from your systems through an API, joined to meetings as a participant, or pointed at an archive for batch scans. Alerts go to your SIEM or workflow.

Kept ahead of new generators

Detectors are retrained as new generators ship, with real and synthetic training data across faces, voices, images and documents, and research pipelines at A*STAR and NTU. A model is not a moat. A pipeline is.

Platform — Working with us

From first call to live detection.

  1. Step 01

    Discovery and exposure audit

    A working session on where synthetic content can enter your processes and what it would cost.

  2. Step 02

    Sample data and profile configuration

    You share a small set of real and suspect samples. We configure a profile for your use case and size the deployment.

  3. Step 03

    Proof of concept in your environment

    A scoped engagement with fixed milestones, running where your data lives.

  4. Step 04

    Production licence and support

    A platform licence that covers every use case you add, with our team on call.

Pricing is scoped per engagement and depends on deployment and volume. Tell us about your use case and we'll come back with a proposal.

Request a proposal

Next step

Talk to sales.

Tell us what you need to protect. A member of the team, not a bot, replies within one business day.

Book a demo