Enterprises are no longer treating conversational AI as experiments. The focus keyphrase — enterprise AI agents in production — captures a shift that is practical and urgent. Companies now embed agents into core workflows, not just chat windows. Therefore, leaders face choices about security, system integration, and new commerce channels. This post walks through what the change means for operations, customer experience, and vendors. It also lays out clear next steps senior teams should consider.

## Why enterprise AI agents in production are happening now

A recent survey from CrewAI shows large companies are rapidly moving agentic AI from pilots into everyday workflows. Consequently, this is not a slow trend. Instead, it is a fast reallocation of budget and attention. Companies want agents that do work — not just answer questions. Therefore, they embed these systems to carry out tasks, update records, and trigger processes across tools.

However, adoption is not carefree. The survey highlights three immediate concerns: security, integration, and reliability. Security is central because agents may touch sensitive accounts and data. Integration is tough because agents must reliably connect to back-end systems. Reliability matters because failed actions erode trust quickly.

For businesses, the impact is clear. First, projects must include IT, security, and compliance from day one. Second, vendor selection now centers on connectors and governance features. Finally, project timelines must account for thorough testing in live systems. As a result, organizations that align technical, legal, and business stakeholders will move faster and safer. Looking forward, expect more tooling aimed at reducing integration friction and giving teams clearer controls over agent behavior.

Source: Digital Commerce 360

Security and identity: enterprise AI agents in production create a new access crisis

When agents can act on accounts, identity and access assumptions break down. CX Today warns that traditional automation only provided scripted responses. However, agentic systems change the game because they can perform actions — move funds, change records, or alter customer statuses. Therefore, companies face new identity, access, and audit challenges.

First, authorization rules must change. Previously, a human served as the gatekeeper for risky actions. Now, decisions can originate from an AI agent. Consequently, firms must define what an agent is allowed to do, and under what checks. Additionally, audit trails become essential. Organizations need clear logs that show which actions were initiated by an agent, which were approved by a person, and which were automated entirely.

Moreover, identity becomes layered. Agents may act using service identities, delegated tokens, or user impersonation. Each approach carries different risks. Therefore, security teams must evaluate how agent credentials are issued, rotated, and constrained. Also, compliance officers will demand explainability and traceability for regulated actions.

In practice, expect more emphasis on role-based access adapted for agents, stronger session recording, and policy-based approvals that trigger human review for risky requests. Ultimately, firms that design secure, auditable agent access will reduce incidents and speed adoption. Otherwise, adoption may stall as risk teams push back.

Source: CX Today

How vendors turn conversation into action with enterprise AI agents in production

Vendors are racing to give AI conversational systems the ability to act. Genesys recently introduced an agentic virtual agent into its cloud platform. Therefore, conversations can now move beyond words and into practical resolutions. The technology uses large action models that interpret intent, select next steps, and carry out actions across systems.

This shift matters for self-service. For example, when a customer requests a change, the agent can evaluate context, validate permissions, and then execute the change without human handoff. As a result, resolution rates rise and wait times fall. However, these gains depend on trustworthy integrations and robust decision logic.

Vendors emphasize adaptability. If context changes mid-interaction, the agent can pivot and choose a new path. Additionally, orchestration matters: agents must coordinate across CRM, billing, fulfillment, and other tools to complete a task. Therefore, platform vendors now highlight connectors, governance controls, and visibility features as selling points.

For enterprise buyers, the implication is practical. Look for systems that provide clear audit logs, easy-to-manage connectors, and policies that let business teams set safe boundaries. Also, demand proof of reliability in production-like scenarios. Going forward, expect vendors to expand action catalogs and to offer more pre-built integrations. In short, the vendor push is turning conversational AI from an information channel into a working part of operations.

Source: CX Today

Commerce shifts: Shopify’s growth and the new AI sales channels

Shopify’s recent report shows a 30% revenue rise in 2025, driven in part by a surge in B2B sales and new AI commerce channels. Therefore, the platform shift is a market signal: merchants are already experimenting with AI-driven buying experiences. AI agents that can recommend, configure, and even transact are changing how customers find and buy products.

For merchants, this is both opportunity and challenge. On one hand, AI commerce channels can mean higher conversion and simpler buyer journeys. On the other hand, integrating agents into payment, inventory, and fulfillment systems requires work. Additionally, B2B sales add complexity because orders, discounts, and approvals often involve multiple systems and people.

As a result, merchants and platforms must prioritize secure transaction flows and accurate order orchestration. They must also make sure the agent’s suggestions reflect real inventory and pricing. Moreover, customer trust is essential. If an agent provides a recommendation that cannot be fulfilled, the loss in trust can be costly.

Looking ahead, expect more platform-native agent tools and merchant-focused connectors that reduce integration work. Also, vendors will likely offer pre-built commerce flows that cover common B2B scenarios. In short, AI agents are not just a support channel; they are becoming a new sales channel that needs careful engineering and governance.

Source: Digital Commerce 360

Turning signals into action: Medallia’s approach to omnichannel CX and agent-driven insights

Medallia’s recent innovations aim to make customer signals more actionable through generative AI. Therefore, the platform is focused on reducing the effort required to analyze complex customer data. The goal is clear: convert insights into operational actions that improve experience.

The new tools emphasize accessibility. For example, generative AI can surface themes and suggest prioritized actions. As a result, teams spend less time sifting through data and more time executing changes. Additionally, when insights connect to agent workflows, organizations can close the loop faster. That means an insight from one channel can trigger action in another.

However, the challenge is integration. Signals live across surveys, chat, product telemetry, and transactional logs. Consequently, turning an insight into a reliable action requires consistent data, clear ownership, and an orchestration layer that routes tasks to the right systems or people. Medallia’s direction suggests vendors will invest in these orchestration capabilities.

For enterprises, the takeaway is practical. Invest in data hygiene and in tools that can translate insights into tasks. Also, require vendors to demonstrate not only analysis but also the pathway to action. Moving forward, expect a tighter coupling between experience analytics and agentic automation, which will let companies act faster on what customers say.

Source: CX Today

Final Reflection: Connecting the dots — from pilots to production, safely and strategically

Across these reports, a clear pattern emerges: enterprise AI agents in production are here, and they are practical. Companies want agents that act, not just chat. Therefore, vendors focus on action models, orchestration, and commerce channels. However, this progress raises urgent questions about identity, access, and auditability. As a result, security, integration, and governance are now top priorities for successful deployments.

Looking ahead, leaders should treat agent projects like any mission-critical system. That means involving security, IT, legal, and business owners early. Additionally, demand transparency from vendors about connectors, audit trails, and policy controls. Finally, focus on small, high-value workflows that demonstrate ROI and can be scaled safely. If organizations follow this path, they will capture the productivity and customer experience gains these tools promise. Moreover, they will do so without sacrificing control or trust.