Epsilon AI Analytics
Ask AI
العربية
Book a Demo

SafeACS Series

SafeACS Access Control

Facial-recognition access control with multi-factor authentication and attendance.

The problem

Card and PIN access controls the credential, not the person. Cards get shared, lost and cloned; PINs get written down; and the audit trail records that a badge went through a door, which is a different fact from the one an investigation needs. Biometric access removes the transferable credential — but it introduces a data category that carries obligations most buyers have not thought about until the system is already installed.

Who this is for

  • Security managersAn access record that identifies a person rather than a credential
  • Facilities and site operationsFewer lost-card replacements and less door-by-door administration
  • HR and complianceBiometric data handled in a way that survives being asked about

What people use it for

Main entrance and turnstiles

High-throughput entry where a queue at shift change is the operational constraint.

Restricted and high-value areas

Server rooms, cash handling, laboratories and stores where a shared card is not acceptable.

Multi-site access

One enrolment and one permission model across sites, rather than a card per location.

Contractor and visitor control

Time-bounded access that expires on its own, which is the control most often missed manually.

What it needs to work

  • Door and turnstile locations, with expected throughput at peak
  • Your identity source — HR system or directory — as the authority for who exists
  • Existing door hardware, controllers and network, which usually can be reused
  • A lawful basis and consent position for biometric processing, agreed before enrolment

How it works

  1. Enrol

    The person is registered once, with their permissions inherited from your identity system rather than typed in again.

  2. Present at the door

    Recognition at the reader, with a second factor where the area's risk warrants it.

  3. Decide locally

    The controller decides at the door, so an entrance keeps working when the network does not.

  4. Log the event

    Person, door, time and outcome — including the refusals, which are usually what an investigation needs.

  5. Revoke centrally

    A leaver loses access everywhere the moment HR records it, which is the control cards fail at most often.

How we deliver it

  1. Site survey

    Doors, throughput, lighting, mounting positions and existing hardware. Lighting and mounting decide recognition performance more than the device does.

  2. Design and sizing

    Device selection, controller layout and capacity, quoted against the datasheet for the model proposed.

  3. Privacy and enrolment plan

    Lawful basis, notices, consent and the alternative for anyone who declines — settled before enrolment starts.

  4. Install, integrate, hand over

    Installation, integration to HR and any existing system, then training and documented support terms.

Where it runs

  • On-premises server, with the biometric database inside your network
  • Distributed controllers so doors keep working during a network outage
  • Integration with HR or directory as the source of truth for identity and leavers

Security and governance

  • Biometric templates stored, not photographs — a template cannot be reconstructed into a face
  • Templates encrypted at rest and held inside your infrastructure
  • Consent, notice and a non-biometric alternative for anyone who declines
  • Deletion on leaving, and retention of access logs set separately from templates

Timeline

Driven by door count and installation access rather than by software. The survey establishes it. Enrolment is the phase most often underestimated: it needs people to attend, and where a workforce is shift-based that has to be planned around production rather than around the project.

What you receive

  • Installed and configured readers, controllers and server
  • Integration to your HR or directory system for joiners and leavers
  • A documented privacy position: basis, notice, retention and deletion
  • As-built documentation, admin training and written support terms
  • The datasheet for every device supplied, with its actual specifications

What this does not do

Recognition performance depends on lighting, mounting height and angle, and on throughput at peak — which is why the specification comes from the datasheet for the model quoted after a site survey, not from a web page. We publish no accuracy or capacity figure here, because one that is true for a controlled indoor reader is false for an outdoor gate in direct sun. Biometric access also is not appropriate everywhere: where the workforce or the regulator will not accept it, the right answer may be card plus PIN, and we will say so.

Questions we are asked

Where is the biometric data stored?

On your infrastructure, as encrypted templates rather than images. A template is a mathematical representation and cannot be turned back into a photograph of a face.

What if an employee refuses to enrol?

There has to be an alternative, and designing it is part of the deployment rather than an exception handled later. In several jurisdictions consent is not valid if the only alternative is not working.

Can we keep our existing doors and cards?

Usually yes for the door hardware, and running card and biometric together is a common and sensible transition. The survey confirms what can be reused.

About the figures on this page

This page describes capability and method. It does not publish accuracy figures, throughput numbers or delivery dates, because those depend on your data, your systems and your scope — and a number published here would be wrong for most readers. You get them, in writing and against your own data, at scoping.

A first call is a technical conversation, not a pitch: what you have, what you need, and whether this is the right approach at all.

Key capabilities

  • Facial recognition · MFA · door control · blacklist · attendance
  • 30K-face database · ~1s recognition
  • Models: SafeACS · PRO · TH
  • Win-CMS manages up to 50 devices

Built on the Unified Intelligence Layer

Every InsAI product runs on the same four-stage backbone.

  1. 1

    Data Integration

    ERP · IoT · BIM · CRM

  2. 2

    AI Models & Predictive Engines

    Forecasting, detection, optimization

  3. 3

    Automation & AI Agents

    Acting on predictions, end to end

  4. 4

    Real-time Dashboards & Decision Systems

    From the floor to the boardroom

SafeACS Access Control

Facial-recognition access control with multi-factor authentication and attendance.

Type to search across Epsilon.

navigate open esc close Open full search →

Get this download

Enter your details and we'll email you the download link right away.

We'll email the link to you — no spam.
WhatsApp Call Book a Demo