What SENTRA is

SENTRA is mission control for a cleaner internet — AI firewalls for websites and internet lines, governed from one command centre and run on your own infrastructure.

SENTRA — Orchestrated Threat Governance (OTG) is a family of security products that inspect and block attacks at two places:

• at the website — every web request is judged before it reaches the web server, and
• across the whole internet line — all traffic for an organisation's IP addresses is cleaned before it reaches its network.

Both are governed from one command centre, and everything runs on your own infrastructure: your servers, your virtual machines, your data centre. SENTRA is not a cloud proxy that your traffic has to pass through on someone else's network.

Two ways to use it#

Three layers, one chain of command#

SENTRA is organised in three layers. Each layer has one job:

1. Govern — SENTRA VIEW. One cockpit above every GUARD: all orchestrators, nodes, customers, tickets, fleet billing, the price book and access across the fleet.
2. Orchestrate — SENTRA GUARD. The multi-tenant control plane: onboards customers, provisions a firewall for each on a node, watches node health and bandwidth live, moves customers between nodes, and handles licences, billing and alerts.
3. Enforce — the shields. - Web App · ATS (SENTRA AI Threat Shield) — an AI web application firewall in front of websites and APIs. - Internet · NTS (SENTRA Network Threat Shield) — a network firewall that sits inline on an internet line and protects every service behind it.

Two supporting services complete the platform:

• SENTRA Licensing Authority (licensing.sentra-ai.com) — the root of trust. It holds the root signing key, designs the licence hierarchy and issues every licence.
• SENTRA release server (updates.sentra-ai.com) — publishes signed software releases that every SENTRA system checks before installing.

What makes SENTRA different#

• On your infrastructure. Shields run on your servers; the control plane runs where you choose. Traffic is never sent to SENTRA.
• Watch before you block. Every shield starts in Monitor mode, which shows exactly what would be stopped on your real traffic. You switch to Enforce when you are satisfied.
• Fails safe. ATS has a fail-open guard that sends traffic straight to the web server if the shield stops answering; NTS bridge segments keep forwarding even if its software stops. Licensing problems never stop protection.
• Works offline. Licences are checked offline against a public key built into every product. When a system is online it also checks in with the Licensing Authority for revocations and renewals.
• Signed releases. No SENTRA system installs software that has not been built, tested, approved with two-factor authentication and signed by SENTRA.
• Flexible. Roles and permissions are fully customisable on every product, dashboards can be arranged per user, and operators set their own plans and prices.
• Everything recorded. Every decision and every admin action is written to an audit trail; on ATS and the Licensing Authority the trail is hash-chained so later tampering is detected.

The SENTRA family

What each product does, where it runs and who uses it.

At a glance#

Web App · ATS — SENTRA AI Threat Shield#

An AI web application firewall that works as an inspecting reverse proxy. Point the website's DNS at it; it terminates HTTPS, inspects each request and forwards clean traffic to the web server. Nothing changes in the application.

• Five checks on every request — network, fingerprint, behaviour, AI intent and policy — scored together into one verdict: pass, challenge or block.
• AI agents and swarms — genuine AI agents are verified by signature and network, impostors are caught, and swarms that spread one job across many addresses are treated as one operator.
• Monitor, then Enforce — see what would be stopped on your own traffic first.
• Page Shield and API Shield, response hardening (security headers, cookie flags, HSTS, banner stripping), bans, incidents, reports, and a tamper-evident audit ledger.
• Optional network sensors and threat feeds (Suricata IDS, Zeek, signature updates) and an optional AI analyst & Copilot that can use a private, on-premises language model.

Internet · NTS — SENTRA Network Threat Shield#

An inline network firewall for a whole internet line. Every port gets a role (WAN, LAN, management); a WAN and a LAN port form a segment, which runs either as a bridge (bump in the wire, no addressing change) or routed (the box is the gateway, with NAT).

• Edge protection on every WAN port: allow and block lists, threat feeds, bans, flood limits, per-customer prefixes.
• Stateful policy with objects, services, schedules, expiry and connection limits; NAT; DNS and HTTPS-name filtering; application control; bandwidth shaping.
• Intrusion prevention (Suricata), an AI daemon that learns normal traffic and bans flood sources, VPN (WireGuard, IPsec), multi-WAN, BGP (including remotely triggered blackhole), and high availability pairs.
• Commit-confirm: every change rolls itself back after 60 seconds unless confirmed, so a mistake cannot lock you out of an inline box.

SENTRA GUARD#

The orchestrator an operator uses every day. A GUARD:

• onboards customers, each with their own firewall (ATS or NTS) provisioned on a node;
• registers nodes and watches them live on a radar cockpit (health, bandwidth, latency, threats);
• moves customers between nodes, pushes NTS policy templates, and offers software updates to its nodes;
• issues licences for its boxes (within the limits the Licensing Authority gives it), runs billing (plans, invoices, payments) and sends alerts;
• manages users, roles and access — including who may sign in to each ATS and NTS box — with maker-checker approvals for sensitive actions.

A GUARD works fully on its own. When it is linked to a VIEW, VIEW becomes the boss and the GUARD keeps working (and keeps its own copy of everything) if VIEW is unavailable.

SENTRA VIEW#

Fleet command for operators with more than one GUARD:

• Fleet cockpit — every GUARD, node and customer at once, with an interactive hierarchy map.
• Customers, services and nodes across all GUARDs, with the ability to create and change them on any GUARD.
• Tickets with SLA (ticketing lives only in VIEW).
• Central billing and the fleet price book, pushed to every GUARD.
• Access across the fleet — users and roles on every GUARD and box.
• Licensing — VIEW is certified as a signing authority and licenses its whole fleet automatically from the plan designed in the Licensing Authority, even while disconnected.
• Software updates for itself, its GUARDs and their nodes, from the signed release server.

SENTRA Licensing Authority#

SENTRA's own portal at licensing.sentra-ai.com. It holds the SENTRA root signing key in an AES-256-GCM encrypted vault that is sealed after every restart, forces two-factor sign-in, and:

• designs the hierarchy for each customer deployment (stand-alone ATS or NTS, GUARD with boxes, VIEW fleet) and what each level is entitled to;
• issues licences and certifies VIEWs and GUARDs as signing authorities with limits;
• keeps a register of every licence ever issued and sees which systems check in;
• approves and signs software releases for the release server.

It is operated by SENTRA staff; customers never need access to it.

How the pieces fit together

The hierarchy VIEW → GUARD → node → ATS/NTS, who connects to whom, and what keeps working when a link is down.

The hierarchy#

SENTRA VIEW                     governs (optional)
 └── SENTRA GUARD (one or more) orchestrates
      └── node (server or VM, runs the node agent)
           └── Web App · ATS  or  Internet · NTS   enforce on customer traffic

• A node is a server or VM registered on a GUARD. It runs the node agent, which receives work from the GUARD and reports health, traffic and threats back.
• On each node runs one shield: ATS (for websites) or NTS (for internet lines). Each customer gets their own firewall instance.
• VIEW is optional. A GUARD on its own is a complete system with customers, billing, licensing and users. VIEW adds fleet-wide command above several GUARDs.
• ATS and NTS can also run stand-alone, with no GUARD at all: each has its own console, licence and updates.

Who connects to whom#

SENTRA is designed so that lower tiers connect out to the tier above. Nothing needs to reach into a node or a GUARD from outside.

When a link is down#

Two directions of change#

When VIEW is linked, both VIEW and each GUARD can change customers, nodes, instances and billing:

• A change made in VIEW is sent to the GUARD that owns the record, which runs it through its own rules and records it as VIEW <user>.
• A change made in a GUARD reaches VIEW with the next snapshot.
• If the same record changed on the GUARD after the VIEW user last saw it, the GUARD refuses VIEW's change and VIEW shows the current values — nothing is silently overwritten.

Some things are owned by VIEW when it is present: the fleet price book (once VIEW has pushed it, GUARDs show prices read-only) and tickets. A GUARD can also be set to Billing from VIEW, in which case only VIEW bills that GUARD's customers.

Who SENTRA is for, and how it is deployed

Enterprises, SMEs and homes protect themselves; telcos, ISPs, MSPs and data centres sell protection as a service. Four deployment patterns cover both.

Protect yourself#

Enterprises, government agencies, banks, universities, SMEs and home offices use the shields directly:

• Web App · ATS for websites, portals and APIs — online banking, e-commerce, booking systems, customer portals, public sector services.
• Internet · NTS for the whole internet line — mail servers, VPNs, remote access and every other service behind the line, not just the website.
• Both together when an organisation has public websites and wants its whole line cleaned.

Each shield is managed from its own console, licensed on its own and updated from the signed release server.

Sell protection as a service#

Telcos, ISPs, MSPs and data centres run SENTRA multi-tenant:

• SENTRA GUARD onboards customers, provisions a firewall for each on the operator's nodes, bills them, and shows the whole operation on one radar.
• Web App · ATS nodes protect customers' websites; Internet · NTS nodes clean customers' Dedicated Internet Access (DIA) lines.
• SENTRA VIEW sits above several GUARDs (regions, brands, subsidiaries) for fleet-wide tickets, billing, prices, access and updates.

Four deployment patterns#

Patterns can grow: a stand-alone shield can later be adopted by a GUARD, and a GUARD can later be linked under a VIEW without starting again.

ATS placement#

NTS placement#

On an operator network NTS can also be a routed hop for customer prefixes or sit inline at layer 2 in front of the line, as the network design requires.

The licensing model

One root of trust, keys that work offline, authorities that license their own fleets, and protection that never stops because of a licence.

Every SENTRA licence traces back to one root of trust: the SENTRA root signing key held in the SENTRA Licensing Authority. Every product ships with the matching public key, so it can check any licence offline.

Three kinds of deployment#

Keys issued by a GUARD or VIEW authority are OTG2 keys: they carry the authority's certificate, so a box can check the whole chain back to the SENTRA root. Authorities report every key they issue back to the Licensing Authority (online, or later as a ledger file), so the register stays complete.

What a licence covers#

A licence also has a start date, an expiry (or perpetual), and the Installation ID of the box it is for. ATS licences have no user or site count: the shield protects everything behind it.

Protection never stops#

• An unlicensed module has 30 days of grace.
• After that it is restricted: inspection, rules in force, bans and the traffic path stay exactly as they are, but changes, updates and AI calls for that module are paused until a key is added.

Online check-in#

When a system can reach the internet, it checks in with the Licensing Authority — stand-alone ATS and NTS about 2 minutes after starting and then every 6 hours (or on demand with Check in now). The Authority can:

• revoke a key (the box shows it as Revoked by the SENTRA licensing authority until a later check-in lifts it), and
• hand back renewed keys, which are added automatically after the normal checks.

A failed check-in never restricts anything; offline keys keep working exactly as before.

Working disconnected#

A VIEW or GUARD certified as an authority keeps licensing its fleet while disconnected, within its limits. It syncs its ledger with the Licensing Authority when it is back online.

How updates work

Signed releases are built, tested, approved with 2FA, published, and checked again at every step before anything installs — and nothing installs until an admin says so.

SENTRA ships software as signed releases. No SENTRA system installs code that SENTRA has not built, tested, approved and signed.

The release path#

1. Build. The release builder keeps a clean copy of each product. When a new version is ready, it runs the product's full test suite and prepares a candidate. The builder never signs anything.
2. Approve. A SENTRA approver opens Releases in the Licensing Authority, reviews the version, what's new and the test results, and clicks Approve & publish with a fresh two-factor code. With four eyes on, a second approver is needed.
3. Publish. The Licensing Authority signs the product's release list with the release key and publishes it on updates.sentra-ai.com. A bad release can be withdrawn: systems then never offer or install it.
4. Check and offer. VIEW and stand-alone ATS and NTS look for new releases periodically (and on Check now). Each release is checked end to end — SENTRA root key → release key certificate → signed release list → package → every file — and then offered in the console with its release notes and expected downtime.
5. Install. Nothing installs until an admin clicks Update now or schedules a time — except where an admin has deliberately turned on automatic updates. VIEW hands releases to its GUARDs with the proof; each GUARD checks every file again, and each node checks what its GUARD hands it.

What the admin decides#

Rolling back#

GUARD and VIEW keep the previous release as a roll-back target, and a failed install is recorded with its reason; the running version stays in place.

Your data, and what stays on-premises

Traffic is inspected where it lands. What leaves a SENTRA system, when, and how to switch it off.

SENTRA inspects traffic on the box it arrives at. Customer traffic, request contents and logs stay on your infrastructure. The shields only contact outside services that you switch on.

What stays on your systems#

• All traffic and its inspection: requests, responses, packets, flows.
• Events, incidents, bans, audit ledgers, reports and backups (unless you add an off-box backup destination).
• User accounts, password hashes and two-factor secrets.
• Licence keys and the box's Installation ID.

What leaves, and why#

SENTRA's own services#

The Licensing Authority stores everything — even customer names — encrypted in its vault (AES-256-GCM), and the release server only serves public, signed release files. Neither ever receives customer traffic.

Choosing a deployment

A short decision guide — which products, which placement and which licences for your situation.

Answer these questions in order.

Quick recommendations#

Next steps#

• Read the Product Specifications for requirements, ports and capabilities.
• See the Hardware Appliance Guide for SENTRA NTS appliances for SME and home.
• Contact SENTRA (demo@sentra-ai.com) for sizing, a demo on your own traffic, and licensing.