The AI contact center isn't a software problem. For most enterprise organizations, it's an infrastructure problem.
Your team has evaluated AI voice solutions. The demos look impressive. The KPI projections are compelling. And then someone from IT asks: "How does this connect to our phone system?"
The answer, for a significant portion of enterprise contact centers still running on-premise PBX infrastructure, is: with difficulty, if at all.
This article is about why that is, what it costs you, and what the path forward actually looks like.
The Legacy PBX Landscape in 2025
A substantial portion of enterprise contact centers in the US are still operating on infrastructure purchased between 2005 and 2015. Cisco Unified Communications Manager (CUCM), Avaya Aura, and Genesys on-premise deployments were state-of-the-art when they were installed. They represented significant capital investment — hardware, licensing, professional services, custom integrations.
They were built to last. And they did. Which is precisely the problem.
These systems were designed for a world where: calls were handled by human agents, call volume was relatively predictable, audio stayed within the PBX, caller data lived in the PBX or a connected CRM retrieved manually by agents, and "intelligence" meant a well-configured IVR tree.
None of those assumptions hold in an AI-augmented contact center.
The Five Ways Legacy PBX Blocks AI Value Creation
1. Caller ID Gaps — The AI Has No Context to Work With
AI voice agents deliver value through context. Knowing who is calling before the conversation starts — pulling their order history, account status, previous interactions — is what separates an AI agent that feels knowledgeable from one that feels like an automated attendant. This requires reliable Caller ID (ANI — Automatic Number Identification) delivered with every inbound call.
Older Cisco CUCM and Avaya Aura deployments frequently have ANI delivery issues. Configuration problems, trunk settings, ISDN PRI vs SIP mismatches, carrier handoffs — any of these can result in calls arriving at your contact center without a phone number attached. For a human agent, this is a minor inconvenience. For an AI agent, it's a structural failure.
What this looks like in practice: AI agent answers, says "Welcome, I'm pulling up your account" — and pulls up nothing, because there's no number to look up. The customer experience degrades immediately.
2. Fixed Concurrent Call Capacity — AI Scalability Hits a Hardware Ceiling
One of the core value propositions of AI contact center solutions is elastic capacity: the ability to handle 50 calls or 5,000 calls simultaneously, scaling automatically with demand. Legacy PBX systems have fixed concurrent call capacity determined by hardware. A Cisco CUCM cluster licensed for 500 concurrent calls handles 500 concurrent calls — and not 501.
Expansion requires hardware procurement, installation, configuration, and additional licensing. In a world where call spikes are unpredictable, the time required to expand capacity (weeks to months) is incompatible with the business need (now).
What this means for AI: An AI system capable of handling unlimited concurrent calls is limited, in practice, to whatever ceiling your PBX hardware defines.
3. Audio Routing Constraints — Legacy Systems Can't Stream to an LLM
AI voice agents work by streaming audio in real time to a language model, processing the response, and streaming synthesized speech back — in under a second. This requires low-latency SIP trunk connections, support for modern audio codecs (Opus, G.722), and flexible media routing.
Older CUCM and Avaya Aura deployments typically route all media through centralized hardware gateways configured for G.711 or G.729 — codecs designed for compressed phone calls, not low-latency AI audio streaming.
The result: audio that reaches the AI with added latency, degraded quality, or via workarounds that introduce their own failure points. Conversations that should feel natural develop perceptible delays.
4. Dynamic Routing Limitations — No Intelligence Before the Call Reaches the Agent
Modern AI contact centers anticipate the call before it's answered. Based on who is calling, what time it is, what their recent interaction history shows, the platform can pre-load context, select the appropriate AI agent configuration, and begin preparing a response.
This requires dynamic, programmable call routing with real-time data access. Genesys on-premise, Avaya, and Cisco routing configurations are typically static — defined in configuration files, managed through admin consoles, deployed through change management processes. Routing changes that would take seconds in a cloud-native environment take days or weeks in legacy infrastructure.
5. Integration Complexity — The Data Stays Siloed
AI agents derive value from integrations: CRM, order management, ticketing, payment systems. Legacy PBX environments often have complex, custom integrations built over years — CTI layers connecting Cisco or Avaya to Salesforce or SAP through proprietary middleware.
These integrations weren't designed for AI. They were designed for screen-pop.
The integration work that takes days in a cloud-native environment takes months in a legacy one.
Two Paths Forward — And How to Choose
Path 1 — Overlay via SIP Trunk (No PBX Replacement Required)
For organizations where the legacy PBX is deeply embedded and full replacement isn't realistic in the near term, a SIP trunk overlay approach allows AI capabilities to be deployed without touching the core system.
How it works: A modern SIP trunk layer sits between your carrier and your PBX. Inbound calls are intercepted at the SIP layer before reaching the PBX. AI handling occurs in the cloud. Calls requiring human agents are transferred back through the SIP trunk to your existing agent desktops.
What this solves: ANI delivery issues, concurrent capacity limits, audio routing constraints.
What it doesn't solve: Deep integration with PBX-native routing logic, existing CTI integrations that need significant rework.
Best for: Organizations with Cisco CUCM or Avaya Aura systems with 3+ years of remaining useful life.
Path 2 — Full Migration to Cloud-Native Voice Infrastructure
For organizations where the legacy system is approaching end-of-life, migration to cloud-native voice infrastructure is the cleaner long-term answer.
What this means: Carrier SIP trunks connect directly to a cloud voice platform. No on-premise hardware in the call path. AI and human agent routing handled by a unified software layer. Integrations built API-first. Capacity is elastic by design.
Best for: Organizations with PBX systems approaching end-of-life (Cisco CUCM pre-12.x, Avaya Aura pre-8.x), greenfield contact center builds, or organizations where integration debt makes overlay approaches as complex as migration.
What Unicall's Infrastructure Approach Enables
Unicall is built on cloud-native voice infrastructure — modern SIP trunking, elastic concurrent call capacity, low-latency audio streaming optimized for AI, and API-first integrations.
For organizations on legacy infrastructure, both paths are supported:
Overlay deployment connects via SIP trunk to your existing environment, resolving the ANI, capacity, and audio routing constraints without requiring PBX replacement.
Full cloud deployment replaces the legacy voice layer entirely, with a migration approach that phases the transition to manage risk.
The Strategic Framing
Legacy PBX infrastructure was a competitive asset when it was installed. It standardized communications, reduced costs relative to what came before, and served the contact center model of its era.
In 2025, it has become a liability — not because it failed, but because the value creation opportunity has moved. The organizations that will lead on customer experience over the next five years will be the ones running AI on infrastructure designed for AI.
The question isn't whether to modernize. The question is whether to do it incrementally, through overlay, or completely, through migration — and that depends on your specific system, your timeline, and your business priorities.
Both paths get you to the same destination: a contact center where Service Level is 100%, abandonment is zero, and the customer experience improves with every call.