Local AI processing · ECOLOR Technologies
3CX AI Server — call intelligence that runs inside your walls
The standalone 3CX AI Server separates AI services — transcription, summaries and sentiment analysis — from the phone system and lets them run locally on your own hardware instead of the public cloud. As of August 2026 it is 3CX Update 10 Alpha: for testing and evaluation only, not for production. ECOLOR Technologies builds evaluation and pilot environments today, and production rollouts once it reaches a stable release.
For healthcare, finance and government enterprises that prefer to keep audio and transcripts inside their own environment — we track the roadmap and get you ready for it.
- Update 10 Alpha · Preview
- Local / on-prem processing
- Data control
- MENA 3CX Solution Provider
- Saudi-based technical team
01 · What it is
What is the 3CX AI Server?
The 3CX AI Server is a dedicated, standalone AI server that separates AI services — call transcription, summaries and sentiment analysis — from the 3CX phone system and lets them run locally, in a private cloud, or on-premise inside your environment instead of being sent to the public cloud. Its purpose is greater data control, a more scalable architecture, and independent updates of the AI layer.
The core idea is simple: instead of sending call recordings to external cloud services to be transcribed and summarized, that processing happens on hardware your organization owns and manages. As a result, audio and transcripts do not leave your environment during processing — an important consideration for enterprises in regulated sectors that prefer to keep sensitive call data under their own control.
This page describes the feature honestly as a preview capability on the roadmap, not a production-ready product. What runs today, stably, is cloud transcription (via OpenAI Whisper and xAI Grok) within the AI Edition — the standalone local server is in Alpha.
02 · Release status
Release status — honestly, Alpha
As of August 2026, the standalone 3CX AI Server (local / on-prem AI processing) is released by 3CX within V20 Update 10 as Alpha — for testing and evaluation only, not for production. The current stable 3CX release is V20 Update 9 (Final); Update 10 has not been made generally available. We therefore present this server as a preview capability, not a solution ready for production deployment.
An important migration note from the research: Update 10 changes how onboard AI works. Existing onboard-AI installations are not compatible with Update 10 and later, and must be reinstalled or migrated to the standalone 3CX AI Server. If you run onboard AI processing today, this is an architectural change to plan for when Update 10 reaches stability.
What we offer now: evaluation and pilot environments on the Alpha release so you can test the architecture, performance and local data flow, with production rollout later once 3CX announces a stable release. We make no guarantee about a general-availability date, and we track the roadmap continuously.
- Current status: V20 Update 10 — Alpha (evaluation only, not production).
- Current stable 3CX release: V20 Update 9 (Final).
- Migration: onboard AI is not compatible with U10+ and moves to the standalone server.
- Cloud transcription (Whisper / Grok) in the AI Edition is available today and stable — the available-today option vs the local-server preview.
03 · Why it matters
What a local server means for your enterprise
Architectural value aimed at enterprises in regulated sectors (Alpha preview capability):
Processing stays local
Transcription, summaries and sentiment run on hardware you own, so audio and transcripts do not leave your environment during processing — supporting your data-governance efforts.
Greater data control
Instead of sending recordings to external cloud services, sensitive call data stays under your entity’s control and policies — a core consideration for regulated sectors.
Decoupled, scalable architecture
Separating AI services from the phone system lets them be updated and scaled independently, giving a cleaner architecture for higher-volume enterprises.
Local vs cloud choice
Your entity chooses between cloud transcription (available today) and local processing on the standalone server (Alpha preview) according to your data requirements.
Independent AI-layer updates
Because AI services are separate from the phone system, the processing layer can be updated without directly touching the core communications architecture.
Only documented hardware
Deployed on NVIDIA GPU-based systems (highest processing capacity) or Apple Mac / macOS (lower-cost local option) — per 3CX documentation, with no invented specs.
04 · Capabilities
What the server processes locally
Capabilities tied to the standalone local server (Update 10 Alpha preview):
- Local call transcription
- Local call summaries
- Local caller sentiment analysis
- Local AI analytics and reporting
- Separation of AI services from the phone system
- Independent updates of the processing layer
- More scalable architecture for larger enterprises
- Local / private-cloud / on-premise option
- Deploy on NVIDIA GPU (highest capacity)
- Deploy on Apple Mac / macOS (lower cost)
- Alternative to cloud transcription for local-first orgs
- Migration path from onboard AI to the standalone server
05 · Who for
Enterprises that care about local processing
- ◆
A Saudi healthcare entity handling calls that involve patient data and preferring to keep recordings and transcripts inside its own environment rather than the public cloud — supporting its data-governance efforts (with no compliance claim).
- ◆
A financial or banking institution handling sensitive customer calls that wants transcription and analysis to run on hardware it owns and manages under its internal policies.
- ◆
A government or public-sector body that prefers to keep audio and transcripts local within its data-residency preferences, with stable communications architecture behind it.
- ◆
A large enterprise with high call volumes that wants to separate AI processing load from the phone system so it can be scaled independently.
- ◆
An organization running onboard AI today that needs to plan its migration to the standalone AI Server before adopting Update 10.
- ◆
A regulated sector that wants to evaluate local processing in a pilot environment before any production commitment, to measure real performance and data flow.
- ◆
An enterprise that prefers to control where call data is processed as part of data-residency requirements, balancing the cloud and local options.
- ◆
A company that wants a long-term AI architecture allowing the processing layer to be updated separately from communications.
06 · Architecture
Cloud vs local — and data control
In the 3CX pipeline, transcription, summaries and sentiment first require call recording to be enabled. From there, two processing paths exist. The cloud path — available today and stable within the AI Edition — sends recordings to 3CX cloud servers via transcription providers (OpenAI Whisper or xAI Grok), processes them, then auto-deletes them after processing. This is a real option today and suits many enterprises.
The local path — via the standalone 3CX AI Server, currently an Update 10 Alpha preview — keeps transcription, summaries and sentiment on hardware you own, so data does not leave your environment during processing. This is where 3CX’s local AI technology applies, and it is distinct from the conversational agents (such as the AI Receptionist) that rely on the customer’s own OpenAI key.
On attribution: the local AI Server and cloud transcription are both native 3CX capabilities — the former Alpha, the latter GA. What ECOLOR provides is the professional service around them: architecture design, sizing guidance, pilot environment setup, integration, monitoring and support. None of these is presented as ECOLOR-developed or as production-ready before its time.
- Shared requirement: call recording enabled before transcription.
- Cloud path (GA today): sent to cloud via Whisper / Grok, then auto-deleted after processing.
- Local path (Alpha): processed on your hardware, data does not leave your environment during processing.
- Conversational agents (different): rely on the customer’s own OpenAI key — not part of the local server.
07 · Security & data residency
Security, data residency and governance — honestly, no guarantees
For enterprises in healthcare, finance and government, the value of a local server is that call data — audio and the resulting transcripts — can stay on hardware inside the entity’s environment during processing, rather than being sent to the public cloud. This supports your data-governance and compliance efforts and helps keep data residency under your control — but it does not make your entity automatically compliant with any regulation or authority.
We make no compliance guarantees (not PDPL, SAMA, HIPAA or any other), and no absolute guarantees of security or availability. We provide a real architectural capability — keeping processing local — and design sound operating practices around it: access controls, monitoring, and cybersecurity for the underlying infrastructure (networks, firewalls, backup, business continuity). Accountability for compliance and policy alignment rests with your entity.
Because the feature is currently Alpha, any security evaluation should be run on a non-production evaluation environment. Data-residency options, processing locations, responsibilities and service levels are confirmed during solution design and in the official solution documents.
08 · Why ECOLOR
Why enterprises evaluate the local server with ECOLOR
Practical reasons that matter to an enterprise decision-maker:
- Architecture and sizing guidanceWe design the local-server architecture and guide you on sizing and appropriate hardware (NVIDIA GPU or Apple Mac per 3CX documentation) against your call volumes.
- Evaluation and pilot todayWe set up an evaluation environment on the Alpha release so you can test performance and local data flow before any production commitment — honest about the limits of a preview.
- Integration, monitoring and supportWe connect the server to 3CX, recording and reporting, monitor and support the environment in Arabic and English from a Saudi team — and deploy production once stable.
- Honesty on release statusWe do not sell you Alpha as production. We are clear on what is GA today (cloud transcription) and what is preview (the local server), and plan your migration with no surprises.
Want to evaluate local AI processing in a pilot before production? Talk to the team that will design the architecture, prepare the evaluation and support it.
Book a consultation09 · FAQ
Frequently asked questions about the 3CX AI Server
What is the 3CX AI Server?
Is it available now for production?
What is the difference between cloud and local processing?
Does it help with data residency and privacy?
What hardware is required?
Which edition is required?
Does it do transcription and analysis fully locally?
How do we start a pilot?
Evaluate local AI processing on 3CX — honestly, starting today
From the cloud transcription available today to the preview of the local 3CX AI Server, ECOLOR Technologies designs the architecture, guides sizing, prepares the evaluation environment, integrates, monitors and supports — on 3CX from Riyadh across Saudi Arabia and the Gulf, honest about the Update 10 Alpha status and without claiming any compliance or premature production readiness.
ECOLOR Technologies — ECOLOR Technologies — Riyadh, Saudi Arabia
3CX Solution Provider · Fanvil Platinum Partner | Yealink Gold Partner
Unified number: 920033987 · WhatsApp: +966920033987 · [email protected]
Book a consultation and tell us about your call volumes and data-residency requirements.
Related links
