Skip to content
Security & Trust

Security is not a feature. It's the foundation.

We build systems that hold customer, financial, and operational data. The security baseline isn't something we charge extra for — it's the floor every Akwazi build starts from.

Baseline

What ships by default. Across every engagement.

These aren't add-ons. They're the starting baseline for every Customer Portal, Integration Platform, and Enterprise Portal we build.

Encryption everywhere

AES-256 at rest, TLS 1.3 in transit, KMS-managed keys, no exceptions.

Identity & SSO

SAML 2.0, OIDC, OAuth 2.0. Integrate with Okta, Entra ID, Google Workspace, and more. MFA enforced.

Role-based access

Granular permissions, just-in-time elevation, per-tenant isolation.

Audit log

Every action — human or AI — logged with actor, context, payload. Searchable, exportable, immutable.

Data residency

EU, US, or your own VPC. Customer data never leaves the region you choose.

Tenant isolation

Logical and physical separation. Per-tenant keys. Strict data boundaries.

AI governance

Confidence thresholds, eval suites, human-in-the-loop, full decision transparency.

Backup & DR

Automated backups, point-in-time recovery, documented RTO/RPO commitments.

Compliance posture

Honest about where we are. Clear about where we're heading.

No over-promising on certifications or attestations. We tell you exactly what we have, what's in progress, and what we can support on engagement.

SOC 2-aligned operations

In practice

Our processes follow the SOC 2 Type II Trust Services Criteria. Independent attestation roadmap shared on request.

GDPR-ready

Supported

DPAs available, sub-processor list maintained, data subject rights tooling built in.

Sector frameworks

On request

ISO 27001, HIPAA, DORA, NIS2 alignment available depending on the engagement and the sector.

Penetration testing

Ongoing

Regular third-party pen tests and vulnerability assessments. Executive summaries available under NDA.

Secure SDLC

Standard

Code reviews, dependency scanning, secret detection, automated security testing — part of every release.

Incident response

Operational

Documented IR plan, runbooks, on-call rotation, communication SLAs. Customer notification commitments per engagement.

AI governance

Responsible AI by default — not by promise.

AI is powerful only when it's controlled. Every AI capability we ship has the same six properties — at minimum.

Confidence thresholds

Every AI decision carries a confidence score. Below your threshold, the action routes to a human reviewer.

Decision transparency

Every AI action includes reasoning, source documents, and the model output — visible in the audit log.

Human-in-the-loop

Any workflow step can require human approval. You decide what the AI can act on autonomously.

No training on your data

Your data is never used to train foundation models. Your business data stays your business data.

Eval-driven

Each AI capability ships with an eval suite. Regression on quality fails CI — like any other production component.

Boundaries enforced

AI tools have scoped permissions. They can't read what they're not allowed to read, or do what they're not allowed to do.

Start a conversation

Have something worth building?

30-minute intro call. We'll learn what you're trying to do, share whether we're the right fit, and — if we are — sketch what a first sprint would look like.

Most calls scheduled within 48h