Follow ZDNET:Add us as a preferred
ZDNET’s key takeaways
- The number of active AI agents in organizations has tripled in the last year.
- AI agents have improved their capabilities by 350% and can now handle complex tasks.
- Employee use of AI agents has increased 3X as trust deepens.
The average number of AI agents activated per organization increased nearly threefold since 2025, according to the 2026 Agentic Enterprise Index, new research from Salesforce that analyzes aggregate AI usage data from the company’s Agentforce platform.
The index analyzes the activity of AI engagements in production for five consecutive quarters of 400 real businesses leverage AI agents to drive ROI. In addition, the index report includes a survey of nearly 5,000 respondents across nine key markets.
Also: ‘Specialists aren’t required’ anymore: How to stay valuable in an AI agent workplace today
The biggest takeaways from the Agentic Enterprise Index report are:
- The scale of agentic AI adoption is accelerating, fast. The average number of agents per organization nearly tripled (going from 5 to 13), while creation time dropped by 53% to an average of 1.9 days to create an agent.
- People are starting to trust agents — a lot. Since Feb. 2025, there’s been a threefold surge in employees’ weekly number of initiated agent sessions.
- Industries are finding the strategies that work best for their business needs: Consumer-facing industries like retail and travel lead in the speed of deployment and the use of agents overall, especially during peak-demand seasons like the holidays. Regulated industries have been slower to adopt and implement agents, and their percent of Agentic Work Units (AWUs) — a term defined by Salesforce that represents one discrete unit of work completed by an AI agent — in the last year is lower, but the agents themselves are often much more sophisticated.
Scaling AI agent adoption
Enterprise AI agents have moved from experimental tests to rapid, production-scale deployment. The average number of AI agents in production grew from five in February 2025 to 13 agents in April 2026. In addition to expanding the use AI agents, businesses have also increased efficiency in terms of how long it takes to add a new agent to production. The time to create a new agent has decreased by 53%, going from 4 days in early 2025 to 1.9 days today.
AI agents are becoming smarter and more capable. Businesses saw a 31% increase in the compound monthly growth rate of agents across five quarters. The average share of agentic action per month grew from zero in early 2025 to 15% by April of 2026. The steady-state growth of agentic actions reveals the continued improvement in agentic skills and in their ability to execute and manage complex tasks.
Also: The 3 types of people who will excel in the AI agent era, according to tech leaders
AI agents are becoming smarter and faster
The index report revealed that AI agents are evolving beyond simple text generation and can now execute complex, multi-step business logic across various systems and boundaries. In 2025, the average number of unique actions per agent was 2; today, that number is 4. In the retail industry, the peak average unique action per agent is 9, which translates to consumer expectations rising to agentic scale intelligence, speed, and scale across all touchpoints.
The expanded skill set of AI agents also means they can now take on secondary roles. Agents are handling a greater scope than before and becoming more capable of handling cross-functional activity over time. The share of secondary functions grew for 1% to 6% of functions in the past year.
Also: Token-maxing is an AI cost sink – how to use agents without busting your budget
The index report found that AI agents are also becoming more proactive. The compound monthly growth rate (CMGR) of the action-to-output-token ratio grew by 15% month-over-month. Agents are doing more of the work and the work is becoming more complex and cross-functional.
So are the agents becoming more efficient as they take on more work? One way to answer this question is to study AWUs. (When an agent reasons through, makes a decision, and takes action, then one AWU is consumed.) Salesforce saw 734 million AWUs consumed in April 2026, representing a 15% month-over-month increase in the action-calls-to-output-token ratio.
AI agents are delivering measurable ROI
The index report found that autonomous scaling is delivering measurable ROI while improving employee adoption and driving customer satisfaction. Employees are engaging with AI agents more than ever before. Weekly user sessions with AI agents have tripled year-over-year. The average weekly sessions between AI agents and employees is approaching 8 sessions per week.
Also: Moving from AI pilots to business-wide value requires a superhighway – how to ramp up
Even though AI agent interactions are tripling in number YoY, the number of customer escalations remains steady. In early 2024, zero customer service sessions were handled by autonomous agents. Today, 7 in 10 customer service sessions are handled autonomously by AI agents. As of August 2026, over 5 million conversations at help.salesforce.com were handled by AI agents, compared to 2.4 million by human agents. The expanded use of autonomous AI agents go beyond customer service. The index found four-fold increase in sales when shopper agents are present. In addition, 77% of shoppers report greater confidence after interacting with agents. Employees and customers are deepening their trust in AI agents.
Industry adoption of AI agents is accelerating
Enterprise adoption of autonomous AI agents splits into two distinct operational strategies based on regularity and workflow requirements. The index found that retail and travel industries are scaling for end-of-year demand. The two industries had 60% surge in agent output from November 2025 to January 2026. The retail consumption of AWUs grew by 18X (22% of total monthly output) in early 2026, while the travel industry saw a 7X increase (10% of total monthly output). The public sector industry saw a 227X growth while financial service saw a 13X increase.
The index report revealed that regulated industries require greater AI agent sophistication. Level 1-3 AI agents are expected to read, coordinate, and synthesize, while Level 4-5 AI agents are able to write, analyze, and parse.
Also: 70% of companies deploying customer service AI agents see ROI in 60 days
The Agentic Enterprise Index report from Salesforce highlighted three key learnings:
- Agents can do more than you think.Businesses can build a more versatile, interconnected digital workforce by trusting agents to take on adjacent responsibilities outside their original scope.
- Leverage seasonal surges with agents.Particularly in consumer industries, agent support grows during busy times like holidays or tax season to drive top-line business revenue.
- Integrate agents into daily channels. The data shows employees are engaging in more sessions with agents than ever before, particularly in communication channels they use throughout the day.
To learn more about the Agentic Enterprise Index report, you can visit here.
Artificial Intelligence
US workers are the world’s biggest AI skeptics – and it’s not just about job loss
I used ChatGPT to build a free PDF editor because I didn’t trust it to change my files – it’s glorious
How to audit what ChatGPT knows about you – and reclaim your data privacy
I tried Google Drive’s new AI cleanup tool to fix 14 years of storage clutter – here’s the result
- US workers are the world’s biggest AI skeptics – and it’s not just about job loss
- I used ChatGPT to build a free PDF editor because I didn’t trust it to change my files – it’s glorious
- How to audit what ChatGPT knows about you – and reclaim your data privacy
- I tried Google Drive’s new AI cleanup tool to fix 14 years of storage clutter – here’s the result
