What is Zero Trust? The complete guide.
Security used to be a castle with a moat: inside the wall, trusted. The wall is gone. Applications live everywhere and so do the people. Zero Trust is the model that replaced it: no implicit trust for anyone, continuous verification, and access to exactly what was asked for.
Never trust, always verify.
the definition
A Zero Trust model is a security strategy that assigns no implicit trust to any user or device by default. Everyone, inside the office or outside it, is continuously validated and authorised before being granted the least access that does the job, and then only for as long as the job lasts. It is not a product. It is a set of principles that a platform can enforce, and the five below are the ones every framework agrees on.
- Never trust, always verify.No implicit trust for any user, anywhere, inside or outside the network. Location is a signal, not a credential.
- Continuous risk and trust assessment.Every access request is treated as unique. The context and the risk are validated before access, and re-validated during it.
- Least privilege.People reach only the resources they are entitled to, and nothing else. There is no network behind the application to explore.
- Centralised policy.Security teams decide who reaches what from one place, with role-based policy rather than per-system exceptions.
- Complete visibility.Every access decision is logged with the person, the device and the context, so a threat is identified from the record rather than reconstructed afterwards.
The components every Zero Trust model has.
Zero Trust is a combination of capabilities rather than one control. These four are the skeleton; a platform is how they are made to agree with each other.
- IDENTITY & ACCESSAuthentication that considers the person and the device, with multi-factor authentication as the second layer on top of a password, or in place of it.More
- MICRO-SEGMENTATIONThe estate divided into isolated segments, each with its own perimeter. A person reaches a segment by rule, so a breached device cannot move sideways.More
- POLICY ENGINEEvery request evaluated against identity, device identity and posture, location and behaviour, and answered on a need-to-know basis.More
- VISIBILITYComplete, real-time visibility of user and network activity. The posture of the whole estate is readable from one place.More
How the InstaSafe model works.
Four design decisions, each one answering a specific way the old model failed.
- SPLIT PLANEThe control plane, where trust is established, is separated from the data plane, where data moves. Enterprise data never transits InstaSafe's machines, which is what makes the privacy claim a property of the architecture rather than a promise.
- DROP-ALL FIREWALLThe gateway drops every packet it has not been told to expect. Nothing on the internet can learn that the address exists. The network is blackened to everyone outside it.
- SINGLE-PACKET AUTHORISATIONA single signed message tells the gateway to expect one authenticated device. The door opens for that device, for that session, and stays dark to everyone else.
- LEAST PRIVILEGEAccess is to the one application requested, never to the segment it lives on. Insider threat and lateral movement have no path to travel.
Five steps to implement Zero Trust.
Every organisation does this in phases and keeps refining. These are the phases.
- PROTECT SURFACEIdentify the assets that must be protected: the data, the applications, the cloud resources. This is smaller than the attack surface, which is the point.
- TRANSACTION FLOWSMap how traffic actually moves between those assets and the devices that reach them. The map decides where enforcement belongs.
- ARCHITECTUREDesign from the flows: how many gateways, where they sit, where the controller lives, what stays on-premise.
- POLICYWrite the rules in the controller: who reaches what, when, under which conditions, and how people are provisioned and de-provisioned.
- MONITORRead the logs. Every user and network event is evidence; corrective action comes from the record, not from a report of a breach.
What an enterprise gets from adopting it.
Secure connectivity, anywhere
Critical applications reachable by the people entitled to them from wherever they are, with the same check every time.
Policy per person and device
An identity-centric model lets access policy be written for the person and the device, not for the subnet they happen to be on.
Every access logged
All traffic decisions are recorded for audit. Vulnerabilities and threats are identified from the log rather than guessed at.
Remote work, outsourcing, cloud
The transitions that broke the perimeter model (remote teams, third parties, cloud migration) become policy changes instead of projects.
Every request decidedon its own evidence.
Three things that follow once trust is a decision rather than a default.
Breach contained
A compromised credential or device reaches one application, not a network. There is no lateral path to travel.
Nothing scannable
A blackened gateway cannot be found, fingerprinted or exploited ahead of the patch. The attack surface is the protect surface.
Evidence by default
The control writes the record. Audit questions are answered from the log the platform already keeps.
zero trust, answered.
Tap a question. If yours is not here, a specialist can answer it.
Talk to a specialist//Ready when you are//
See the model running against your own applications.
A 30-minute walkthrough: your identity provider, your application types, your auditor's questions.
Regulated, air-gapped, or on-premise? See deployment options