On 15 July 2026 PwC US announced, together with OpenAI, a set of agentic contact-center and customer-service solutions built on OpenAI's multimodal APIs, spanning voice and digital agents, and a dedicated Center of Excellence to deliver them. PwC also markets a separate agentic contact-center offering built on Salesforce technologies. The announcement was reported by PwC's own newsroom, PR Newswire and CX Today. It signals that the Big Four have entered the packaged agentic customer service market alongside the large systems integrators.

Per a Gartner survey referenced across the trade press including CX Today, roughly 91 percent of customer service and support leaders report being under pressure to deploy AI in 2026. The pressure figure is significant because it makes the deployment inevitable for most mid-market operators; the operating decision is not whether to deploy agentic customer service but how to verify a deployment and how to price the run so the programme survives past year one.

Because a Big Four or tier-1 consultancy deployment is priced against the deployment phase — assessment, design, build, go-live — and against an enterprise buyer with matching integration budget, data maturity and internal team. A mid-market operation typically has a smaller knowledge base, a less clean contact-reason taxonomy, tighter integration budgets and no dedicated internal AI ops team. Importing the enterprise pattern without adaptation is how mid-market contact-center AI programmes stall between go-live and the second quarterly business review.

Data readiness — knowledge base, taxonomy, identity resolution and historical interactions assessed before the agentic layer is quoted. Integration reality — CRM, telephony, ticketing, knowledge and identity systems verified in a week-2 spike, not on slides. Human-in-the-loop design — escalation triggers, warm handovers, synchronous human approval on regulated actions, feedback loop measured. Lock-in — model, stack, data and operating-model lock-in seen explicitly and priced against switching cost. Operating model — the split between deployment economics and run economics visible and matched to the buyer's actual cost structure. Cost-to-serve and TCO — baseline, containment, human time saved, model compute cost and blended cost-to-serve tracked from week 1, not demoed on day 1.

By pairing a credible deployment partner — which can be a Big Four integrator, a specialised systems integrator or an experienced BPO with the required certifications — with a run partner whose cost structure matches the run economics. Run-phase work is 24/7 monitoring, model drift detection, knowledge-base curation, prompt and policy iteration, human-agent supervision and continuous CX measurement, delivered at a run-rate the buyer can afford for the next three years. Where the deployment and the run are delivered by the same partner, the buyer verifies that the run economics are actually costed into the SOW, not just the deployment.

A nearshore hybrid model — roughly 80 percent automation and 20 percent human, with the human line delivered from a Morocco or comparable nearshore operation in French, Arabic, English and Spanish across EU time zones — is a defensible run partner for mid-market buyers who have deployed agentic customer service on any of the credible enterprise stacks. The 80/20 mix is not doctrinal; the specific ratio is a function of contact-reason distribution, regulatory constraint and brand tolerance. What is doctrinal is that the run partner is contractually accountable for the six checks, month after month, with evidence.

Call IT Dev operates BPO, customer support and AI-automation engagements from Morocco with nearshore EU-time-zone delivery, multilingual coverage in French, Arabic, English and Spanish, and a posture aligned with CNDP Law 09-08 and GDPR obligations. The engagement can pressure-test an existing deployment against the six-check framework, or scope a hybrid deployment-plus-run programme with data readiness, integration verification, human-in-the-loop design, lock-in transparency, a run operating model matched to mid-market economics, and a monthly cost-to-serve trajectory reported in the same pack as classical CX metrics.