Security & Compliance

Built so your customers' data stays your customers' data.

CLOSER reads inbound messages, drafts replies in your voice, and sends them. Every byte is encrypted, every reply is logged, and nothing — not your inbox, not your customers' messages — is ever shared with another client, ever sold, or used to train public models.

🇬🇧 UK-hosted (London) 🔒 GDPR-compliant 🛡️ Per-client isolation 📋 Full audit log
99.987%
Uptime (30d)
Last incident 47d ago
12,847
Replies handled
Last 24h, all clients
0
Data breaches
Since launch
£10m
Cyber liability
Beazley policy

Three pillars, no surprises

Security isn't a checkbox. It's how the product is built — from the database schema up.

🗝️

Encrypted end-to-end

TLS 1.3 in transit. AES-256 at rest. Per-client KMS keys — even Agent Networks engineers can't read your customer messages without your written approval and a recorded audit trail.

🧱

Isolated per client

Your data lives in its own logical tenant. Voice models are trained on your replies only. CLOSER cannot "see" any other client's inbox — not even other clients in the same vertical.

📋

Auditable end-to-end

Every inbound, every draft, every sent reply, every escalation — logged with timestamp, channel, sender hash and reviewer. Export the whole audit trail any time. ICO will love you.

Where your data lives, end to end

A message arrives. CLOSER replies. Here's every stop in between.

1. Inbound

WhatsApp / Email / IG arrives via channel API (TLS 1.3)

2. Your tenant

Stored encrypted in your isolated UK database

3. Reply drafted

Your voice model (yours only) generates a draft

4. Reviewed

Suggest mode: you tap send · Autonomous: rules check

5. Sent

Out through the same channel, logged with hash

6. Audit

30-day rolling log + on-demand export to your S3

What never happens: your customer's data never touches another client's tenant, never trains a shared model, never leaves the UK without your written instruction.

CLOSER vs the alternatives

What "GDPR-compliant" actually looks like once you read the small print.

CapabilityCLOSERGeneric AI chatbotVA on Upwork
Hosted in UK✓ London (AWS eu-west-2)✗ Usually US✗ Anywhere
Per-client encryption keys✓ Yes✗ Sharedn/a
Voice model trained on shared data✓ Never✗ Oftenn/a
Full audit log exportable✓ One click✗ Rare✗ Manual
Data deletion on cancellation✓ 30 days, certifiedVariesTrust-based
UK ICO registration✓ ZB123456Variesn/a
Cyber liability insurance✓ £10mTheir problemNone

The questions that matter

Most legal & IT teams ask the same eight things. Here they are.

Where is the data stored?

AWS London (eu-west-2). Encrypted at rest with per-client KMS keys. Daily backups stay in the same region — nothing replicates outside the UK without your written instruction.

Who can read my customer messages?

You and the users you invite. CLOSER's automation reads them to draft replies. No human at Agent Networks reads them without a written support ticket from you — and that access is logged and surfaced in your audit trail.

Does CLOSER train on my data?

Only your own voice model is trained on your data — and only your data. We don't pool clients' replies. We don't sell, license, or share your message contents with anyone, ever, including our model vendors.

What's the GDPR lawful basis?

You are the data controller; we are the processor. There's a DPA in your contract (template available on request before signature) that covers retention, sub-processors, breach notification (72h), and your right to audit.

What happens if I cancel?

Day 0: replies stop. Day 1–29: your data is in cold storage, restorable if you change your mind. Day 30: certified deletion. We issue a signed deletion certificate within 7 days of that.

What about abusive or spam messages?

CLOSER flags abusive content, refuses to reply to obvious spam/phishing, and escalates anything that mentions self-harm or safeguarding directly to you with the channel and timestamp.

Have you been audited?

Independent UK pen-test by NCC Group (Nov 2025). Report available under NDA. SOC 2 Type II in progress — target completion Q3 2026. We don't gate the report behind a sales call.

What happens during an incident?

Detection in <5 min via our SIEM. Containment SLA: 30 min. Customer notification SLA: 4 hours. ICO notification (if applicable): 72 hours. Post-mortem published publicly within 14 days on /status.

Want our security pack?

SOC 2 progress letter, pen-test summary, sub-processor list, sample DPA, and the data-flow diagram in full. We send it before you ask for it.

📅Book a Demo