References

Harput Holding is a customer across our whole stack: cloud, security and AI. We moved 100+ servers to AWS and have managed the cloud estate since, run security operations for the group, and most recently built a family of AI agents on AI Central and wired them into the group's internal portal. Employees ask in plain Turkish and get answers from live business systems — behind corporate sign-in, with read-only access and personal-data boundaries enforced by design.

• Cloud migration and management: 100+ servers migrated to AWS and operated since; their IT Director reports a 40% reduction in infrastructure expenses (testimonial)
• Security operations: a SOC for the group — proactive security as part of day-to-day operations
• Harput Assistant: a persistent assistant on every portal page; answers from live production and ERP data, heavy reports cached so nobody waits
• KYS Assistant: the quality-management hub — Q&A over documentation and architecture, a pending-work board fed by the team's issue tracker, ticket automation and a coding agent attached to the repository
• Ops agent: a daily system-health summary to the IT team's chat, plus inventory and routine infrastructure chores
• Teams agents: product and stock lookups for business teams, each agent with its own least-privilege credentials
• Guardrails: enterprise SSO and role-based authorization, read-only least-privilege credentials, sensitive data excluded by design, no secrets in code
• A native home: assistants embedded in the group's internal portal, where reports, IT ticketing and file services already live

Thrace Alfa Capital runs trading systems that cannot stop. We operate its multi-cloud estate as a fixed-fee managed service and own the security side with it: a 24/7 SOC on our SIEM, a security gate on every change, and independent security testing with every finding remediated. Our AI agents keep the watch and explain what they see.

• Cloud management: unified operations across cloud and dedicated servers — patching and upgrades on every host, right-sizing and cost cuts, daily backups, full-stack observability
• Zero Trust access: no internal service exposed to the internet; every privileged session authenticated, short-lived and audited
• SOC: 24/7 security operations on Graf SIEM with cloud audit trails and threat findings correlated, plain-language alerts to the team's chat, documented access governance, and an independent penetration test with all findings closed
• Code security: a mandatory security gate on every change in every repository, centralized secrets management with short-lived credentials, hardened and continuously scanned container images
• AI integrations: AiMon watches the whole estate and attaches an AI diagnosis to every alert; security events arrive as plain-language explanations with what to do; our agents run scheduled audits, cost analyses and patch rounds, engineers decide
• DevOps and integrations: keyless, gated release pipelines; runbooks and infrastructure inventory kept as code next to the team's documentation

Wettarena runs a regulated betting platform on a large cloud estate. We took over day-to-day operations, measured the estate, and hardened it in phases — each closed with a written report that shows before and after with live numbers. The same team handled the company's move to Google Workspace and supports its brand-protection work.

• Cloud management: managed operation of a multi-region estate — change control, patching, incident response and vendor-maintenance handling; unused capacity retired and storage waste cut to near zero
• Edge defence: protection extended to every public entry point; bot, scraper and application-layer DDoS traffic absorbed at the edge before it reaches the platform
• Identity and network hardening: least-privilege administration, short-lived credentials, reduced internet exposure, stale resources removed, and a documented incident-response process
• Backup and DR: immutable, ransomware-resilient backups, written backup and disaster-recovery plans, and a successful restore drill
• Reporting: phase reports for management with measured before/after figures and a prioritised open-items list, so every decision has a date and an owner
• E-mail and brand protection: company-wide move to Google Workspace in groups with hybrid routing and mail authentication fixed; technical support for brand-protection actions against clone sites

Oblira helps EU companies comply with NIS2, DORA, CSDDD, EUDR, CSRD and more — supply-chain and cyber-security due diligence in one product. Graf Clouds designs and operates the Azure platform it runs on, from the first development environment to production, with every customer-facing architecture decision written down as an ADR.

• Greenfield Kubernetes platform: AKS with Argo CD GitOps, environments promoted by merge, Terraform for every environment
• Zero Trust in front: non-production environments reachable only through Zero Trust access; the product signs users in with enterprise identity
• Secrets done right: centralized secrets with least-privilege runner identities, nothing secret in git
• Quality gates everywhere: SonarQube on every repository and pull request, conventional commits, decisions as ADRs
• Two product modules: Supply Chain (EUDR / CSDDD / CSRD) and Cyber Security (gap analysis, compliance dashboard) on a shared service platform