Safely manage your Zendesk from the AI assistant you already use, via the Deltastring MCP. Beacon configuration platform
← Back to news

50% of inquiries handled by AI: Partner integrates artificial intelligence into customer service

Partner's deployment of AI agents handling 50% of customer inquiries represents the operational reality that agentic AI in contact centres has moved beyond pilot phase into mainstream implementation. The telecommunications provider's rollout of three distinct AI personas—Bar, Romi, and Rotem—signals a deliberate strategy to humanise automation whilst distributing inquiry volume across multiple agent profiles, a tactical choice that differs from monolithic AI implementations. Partner's VP framing this as a technology upgrade rather than a cost-reduction play suggests the industry narrative around AI in CX has shifted: vendors and operators are now positioning agentic systems as capability enhancers rather than headcount replacements, even as the 50% inquiry handling rate demonstrates clear labour displacement. For teams already operating within AI-first CX workflows, this benchmark raises an immediate question—is 50% the operational ceiling for AI-only handling, or does it reflect Partner's conservative deployment strategy relative to what their platform could theoretically achieve?

The implications for CX leaders are twofold. First, this deployment validates that agentic AI can sustain production-level performance at scale without catastrophic failure rates, which removes a significant implementation barrier for mid-market and enterprise teams evaluating similar rollouts. Second, and more critically, the 50% threshold exposes a hard constraint: half of all inquiries still require human intervention, meaning the hybrid model is not a transitional state but a permanent operational architecture. This demands immediate reassessment of team structure, training, and escalation protocols—your support staff are no longer handling the full inquiry spectrum but rather the 50% that AI cannot resolve, which typically skews toward complex, high-friction, or emotionally sensitive interactions. The question for support leaders becomes whether your current team composition and skill distribution are optimised for this inverted workload, or whether you're still structured as though AI were a peripheral tool rather than a primary handler.