A study tracking companies that continuously utilized Agentforce between February 2025 and April 2026 highlights industry-specific AI deployment strategies, as well as how these organizations establish trust and measure financial returns.
MANILA (10 August 2026) — Salesforce has launched its 2026Agentic Enterprise Index, leveraging anonymized data from its Agentforce platform to illustrate how companies deploy AI agents, integrate them into daily operations, and capture tangible business value.
This year’s findings reveal a clear division in the global marketplace, where a sector’s unique operational structure and distinct business requirements directly shape how it implements a digital workforce.
Retail and consumer industries typically deploy targeted, high-capacity solutions to resolve immediate customer issues with maximum speed.
Operationally intricate sectors, including manufacturing and the public sector, typically deploy adaptable agents designed to handle multifaceted workflows and strict regulatory compliance.
While both rollout models build trust and deliver measurable returns, the future of agentic AI depends on their coexistence—combining rapid, large-scale results with the ability to execute highly complex workflows across every industry:
- 1. Financial services uniquely bridges high volume and heavy regulation; the sector deploys highly sophisticated agents at a consumer-facing scale, demonstrating that advanced task complexity and mass automation can seamlessly integrate during peak demand periods.
- 2. Retail customer service agents manage seasonal surges by shifting focus to straightforward tasks while significantly increasing their total output and workflow speed.Â
Dive deeper:
Consumer-facing companies are rapidly rolling out agents to meet escalating business needs.
Organizations are scaling up their AI agent fleets at a staggering pace, with the average number of activated agents nearly tripling over the last year. Deployment is becoming friction-free; companies now launch their first agents within just two days of setup—a setup time that plummeted by 53% over the study period.
To measure if this rapid adoption translates into real utility, Salesforce introduced the Agentic Work Unit (AWU), a metric that tracks discrete tasks completed by an AI to ensure intelligence translates into tangible work.
As of April 2026, Agentforce agents’ AWU output is increasing by a 15% CMGR (compound monthly growth rate).
While high-interaction consumer sectors dominate AI output metrics, the actual scope of retail agents remains surprisingly narrow. Most of these digital assistants execute only one or two specific actions throughout the majority of the year, proving that current deployments are optimized for high-volume, repetitive functions.
Global jewelry giant Pandora provides a prime example of this trend during high-stakes shopping windows like Valentine’s Day and the winter holidays. Facing massive spikes in shopper inquiries, the brand uses Gemma—an AI concierge built on Agentforce—to protect its signature personalized customer experience. Gemma fields high volumes of routine questions regarding tracking, order statuses, and jewelry care, while simultaneously analyzing customer preferences to deliver tailored gift suggestions.
By integrating seamlessly with Pandora’s product catalogs and back-end inventory systems, Gemma scales fast, customized support during intense shopping rushes. The AI concierge successfully absorbs 60% of routine service requests during peak periods. This automation not only drives a 10% lift in Net Promoter Score (NPS), but it also frees up human customer service representatives to dedicate their time to more nuanced and complex customer needs.
Agents scale their capabilities, unlocking greater flexibility to handle intense spikes in traffic.
The functional capabilities of AI agents are expanding rapidly beyond simple, single-task execution. While traditional deployments relied on an agent mastering only one or two basic workflows—such as retrieving database records or answering FAQs—today’s agents are taking on far more complex responsibilities. Reflecting this evolution, the average number of specialized skills an individual agent can perform has tripled, rising from just two at the beginning of 2025 to six today.
This versatility becomes even more pronounced under pressure. When peak holiday shopping seasons arrive, the average retail agent successfully handles nine distinct skills—a staggering 350% increase in functional capability. This dramatic jump indicates that businesses are
intentionally leveraging AI to orchestrate highly complex, multi-step customer journeys exactly when seasonal demand threatens to overwhelm human teams.
The latest enterprise data highlights a major shift toward cross-cloud execution. AI agents are no longer confined to single-system workflows; for instance, a front-line service agent can autonomously pivot to cross-functional systems to pull sales history and trigger personalized recommendation engine workflows. This structural flexibility turns basic informational bots into versatile, full-service operational assets.
Operating across disparate systems proves the real-world value of a headless setup. Decoupling an agent’s logic from rigid user interfaces grants it total operational flexibility. This architectural freedom allows the agent to execute actions, run workflows, and process data natively behind the scenes from anywhere.
Complexity by Design: How Regulated Industries Deploy Advanced Agents
Industrial giant Siemens provides a powerful example of how multi-agent coordination can conquer operational complexity. Managing thousands of products across seven isolated business units, the company’s 18,000 sellers were overwhelmed by 2,800 unqualified inbound leads every week, often wasting time on dead ends while prime opportunities were missed. To untangle this long, technical sales cycle, Siemens deployed Agentforce to divide lead qualification into a tag-team workflow. One AI agent actively engages and nurtures the prospect, while a second agent autonomously aggregates missing data, applies qualification logic, and routes the lead with full organizational context—providing seamless, 24/7 end-to-end management.
To measure AI maturity across industries, a “Sophistication Index” was created to categorize agent workflows into five levels of complexity:
- 1. Levels 1–3 (Read, Coordinate, Synthesize): Low-risk, baseline operations including data retrieval, email drafting, and document summarization. Â
- 2. Levels 4–5 (Write, Analyze, Parse): High-complexity operations including updating live databases and programmatically extracting nested parameters from raw inputs. Â
The Sophistication Index serves as a benchmark for true AI utility. By mapping out active complexity tiers, it highlights the pioneers deploying agents for multi-step problem solving over basic task-trigger bots.
Industries like manufacturing, finance, and healthcare (HLS) are building more sophisticated agent networks than traditional digital front-runners like tech and retail. Instead of sticking to basic task automation, these sectors are deploying AI across the complete operational spectrum—spanning everything from data retrieval and document summarization to autonomous communication drafting and direct database updates.
Their total task volume is more modest, but growing exponentially: public sector output skyrocketed 227x while healthcare expanded 19x.
The financial services sector perfectly mirrors this trend, accounting for a solid 10% of total monthly agent output. Just like retail, this highly regulated industry sees its AI activity driven heavily by seasonal consumer rushes—such as the massive spike in support demand surrounding Tax Day.
PenFed exemplifies how AI agents can thrive within strict regulatory guardrails. Serving military members as a federal credit union means navigating intense verification and compliance standards. To move past basic, rule-based bots, the institution implemented stringent risk controls and required comprehensive legal and compliance oversight to ensure safe, secure agent deployments.
To achieve secure, automated service at scale, PenFed introduced “Ace,” an intelligent assistant operating safely behind authenticated online banking logins. Ace handles multi-action workflows like checking loan application updates, evaluating balances, transferring funds, and pulling verified answers from a secure knowledge base. PenFed extended this advanced functionality to phone support with “Echo,” a voice-based agent engineered to completely replace legacy interactive voice response (IVR) and automated teller systems.
PQ: “Our vision has always been to use AI in a trusted, practical, and meaningful way to serve our members seamlessly across every channel. Bringing that vision to life meant going beyond simple chatbots to build sophisticated agent experiences with trust baked in. By pairing robust governance with our unified platform, we’ve safely deployed multi-action agents like Ace and Echo that perform real, complex banking tasks—turning our goal of a truly connected, AI-enabled credit union into a reality.” – Shree Reddy, CIO, PenFed
Boosting the Bottom Line: The Dual Impact of AI Agents on Efficiency and Sales
The modern enterprise AI landscape is rapidly shifting from basic conversation to direct execution. Agents are increasingly triggering background workflows and executing core business logic rather than just generating text response drafts. In fact, this action-to-text output ratio is expanding at a 15% Compound Monthly Growth Rate (CMGR). Today, a sophisticated service agent doesn’t merely compose a reply regarding a customer’s order; it independently retrieves the system record, applies company business logic, and autonomously processes the refund or rebooks the appointment.
Companies leveraging AI agents are outpacing their competitors in sales growth. This commercial advantage is most visible in consumer-centric markets and retail environments, where digital agents directly fuel higher transaction volumes and better performance throughout peak holiday shopping seasons.
“Whether you’re spinning up agents to operate at massive scale or orchestrating them through deep, multistep pipelines, the bottom line is they’re shipping real value. That ROI isn’t just showing up on the top line in sales numbers but in execution efficiency. We are moving from passive chatbots and predictive models to execution-driven agents that actually roll up their sleeves and drive real value,” stated Joe Inzerillo, Salesforce President of Enterprise AI and Technology.
Organizational and consumer trust in AI is deepening. Employees are increasingly integrating these tools into their daily workflows, while service agents are successfully scaling customer conversations—all without experiencing any drop in service quality.
Employee interaction with AI agents has surged by 300% on a weekly basis, serving as a powerful proxy for growing workplace trust. This workplace adoption is highly visible on collaboration networks, where Slack agents managed an average of 67 sessions per week by April—a rapid 3X increase from just two months prior in February.
Customer Zero: How Salesforce Reengineered Work Culture via Slackbot
Salesforce’s internal Slackbot has smashed company records as its fastest-adopted software tool ever, capturing an active user base of 83% of all employees. This widespread implementation yields massive time savings, freeing up an average of five hours per week for workers. Employees utilize the embedded agent to drastically streamline daily workflows—using it to instantly distill missed conversations, write text
drafts, and build executive briefs by seamlessly aggregating relevant data and files from across the entire workspace ecosystem.
Consumers are also embracing autonomous support. Over five quarters, total customer service chats handled by AI agents skyrocketed 170-fold, with agents independently resolving 70% of all cases without human intervention.
This automated support directly impacts the bottom line by eliminating buyer hesitation. In fact, recent research shows that 77% of shoppers who interact with an on-site, branded digital assistant express greater confidence in their buying decisions than those who browse alone.
Customer satisfaction leads all other performance gains from AI agent deployment. CSAT improvements outrank metrics like average handle time, first-response time, and internal representative productivity.
Looking Ahead
As the enterprise digital workforce matures, businesses will no longer face a trade-off between rapid deployment and complex task execution. This traditional compromise will disappear entirely, enabling companies to instantly scale their output to survive intense traffic spikes while seamlessly executing the deep, cross-functional logic required for intricate corporate workflows.
Methodology:
The insights in this article stem from the Salesforce Agentic Enterprise Index, a comprehensive study tracking active AI agent usage from February 2025 to April 2026. To ensure data integrity, the dataset strictly evaluates businesses that kept agents deployed in live production environments every month of the tracking window. This aggregated telemetry isolates how modern enterprises utilize tools like Agentforce to achieve ROI, though the findings do not serve as an indicator of Salesforce’s internal business performance.
