The Agentic CX Era: From Support Tickets to Supply Chains

What happened
Customer‑experience leader NICE released its Agentic AI CX Frontline Report, which it claims is the first data‑driven look at large enterprises using agentic AI for customer service. The report says early adopters are moving from scripted automation to outcome‑driven AI that adapts to customer journeys and human intent. Companies deploying agentic assistants reported 3× faster deployment cycles, cost‑per‑contact reductions in the double digits, containment rates above 80 % for tier‑one inquiries, and up to 20 % improvement in customer‑satisfaction scores. The report also notes that workers’ roles are shifting toward judgment and oversight as AI takes over repetitive tasks.

Why it matters
The shift from chatbots and decision trees to fully agentic AI in customer support signals that conversational agents are moving from experiments into critical infrastructure. High containment rates and improved customer satisfaction show these systems can deliver real operational gains. Such data could accelerate adoption across industries.

What happened
Sabre, PayPal and travel startup Mindtrip announced they will launch what they call the first end‑to‑end agentic AI travel experience. The platform lets travelers describe their trip to Mindtrip’s AI assistant, get personalized flight and hotel options, pay via PayPal, and manage the itinerary in a single conversation. The companies say the service, expected to debut in Q2 2026, unifies travel search, booking, payment and support through a conversation instead of multiple websites or apps.

Why it matters
Travel booking has long been fragmented across search engines, airline sites and payment pages. Integrating sabre’s travel inventory, PayPal’s payment rails and Mindtrip’s conversational agent could show how agentic platforms can streamline complex consumer journeys. Success here would pressure other travel brands to plug into AI‑powered interfaces.

What happened
Nvidia touted a cost breakthrough: open‑source large language models running on its forthcoming Blackwell GPUs can reduce inference costs by up to ten times compared with the previous generation. Healthcare startup Sully.ai cut the cost of AI‑generated medical notes by 90 % using Blackwell’s low‑precision NVFP4 format and optimized software stacks. Gaming company Latitude and infrastructure firm DeepInfra reported a 4× reduction in cost per token while maintaining accuracy. Sentient Labs, which builds multi‑agent chat systems, said it cut inference costs by 25–50 % and that lower costs enabled more complex multi‑agent workflows.

Why it matters
High inference costs have kept many developers on proprietary cloud platforms. Cheaper open‑source models could democratize access to advanced generative and agentic AI by making it economical to run large models on‑premises or at the edge. The improvements also hint that multi‑agent workflows may become viable for smaller companies as hardware catches up.

What happened
Didero, a startup building AI agents for procurement, raised $30 million in Series A funding from investors including Microsoft’s M12 and Chemistry Capital. Didero’s agents communicate with suppliers, track orders and handle exceptions within existing ERP and email systems, learning a company’s products, pricing rules and order histories. Investors say automating routine procurement will become core infrastructure for manufacturers and distributors; Didero plans to extend its agents to sourcing and payments.

Why it matters
Procurement remains a manual, time‑consuming process in many supply chains. Embedding AI agents into emails and ERP systems promises to shorten order cycles and free staff for strategic sourcing. The backing from major venture funds signals confidence that agentic AI can move beyond customer service into operational workflows.

What happened
Cyber‑security firm Proofpoint acquired startup Acuvity to deliver what it calls the first platform for AI security and governance across the agentic workspace. Acuvity’s technology provides runtime inspection and enforcement for AI agents, Model Context Protocol (MCP) servers and large models, protecting against prompt injection, data exposure and model manipulation. Proofpoint says the combined platform will let enterprises continuously authorize and monitor agents across applications and APIs, addressing the risks of shadow AI and hallucinations.

Why it matters
As organizations deploy autonomous agents that can act without human approval, security and compliance become paramount. The acquisition underscores a growing market for tools that provide visibility and control over agents’ actions and data flows, complementing traditional endpoint and cloud security.

Investors rethink AI boom: Reuters reported that enthusiasm for AI’s profitability is giving way to worries about volatility and ballooning capital costs; software stocks have fallen as investors pick winners and losers, particularly after Anthropic’s plug‑ins sparked a 15 % selloff.

Expedia integrates agentic commerce: Expedia told investors it is plugging into generative‑AI search platforms and agentic assistants to ensure its brands appear in AI‑driven travel queries, touting verified reviews and customer support as differentiators.

CIOs warn about friction and flow: Opinion pieces argue that agentic AI fails without an “architecture of flow” that unites back‑end systems and open protocols, and that many firms won’t achieve agentic operations without modernizing infrastructure and data governance.