Salesforce recently released the company’s 11th annual Connectivity Benchmark Report, and it highlighted the fact that AI agents are no longer a niche experiment, but a primary driver of productivity. The report discusses the expected rise in AI agent adoption, the high expectations for them, and offers many other useful insights and stats.
The Adoption of AI Agents Expected to Climb Quickly
The report by Salesforce, which is based on a survey of over 1,000 enterprise leaders, found that the idea of businesses transitioning to an agentic enterprise is gaining momentum quickly. This is where human and AI agents work alongside one another, and the idea is already quite popular, as organizations currently use an average of 12 AI agents.
However, this number is projected to rise by 67% within the next two years. Also, 83% of organizations report that most or all of their teams have already adopted AI agents.
This makes sense, as AI agents can boost a company’s productivity, help with making better decisions, save money, and are available to work 24/7 without any breaks. They also have use cases throughout an organization, including customer service, HR, finance, IT, and others.
Also, platforms like the newly introduced OpenAI Frontier are making it easier for organizations to build, deploy, and manage these agents, and train them well to ensure they focus on the right things.
High Expectations for AI Agents
In addition to the growing use of AI agents, people have high expectations for them, as well. 96% of IT leaders say that AI agents have already improved, or are expected to improve, the employee experience. Also, 95% believe that using these AI agents will free up developers to focus on more pressing and higher-value tasks.
Many companies are also all-in on AI in general and excited about what it can do, as 76% of global ecommerce brands are actively investing in AI. Even consumers think relatively favorably of AI agents, as more than 75% of customers are open to certain agentic features.
Challenges That May Hold Agentic Transformation Back
However, the report also touches on a few challenges that may hold this agentic transformation back. First, around 50% of agents currently work in isolated silos as opposed to multi-agent systems, which leads to issues like disconnected workflows and redundant automations.
Also, the average number of apps in enterprises climbed from 897 to 957 year-over-year, and only 27% of them are integrated. Silos and integration issues already exist, and 86% of leaders are concerned that adding AI agents into the mix may introduce more complexity than value, without properly integrating them.
96% of organizations also experience barriers to using data for AI use cases, and 40% say that old or outdated architecture and infrastructure are a major blocker of AI agent adoption.
Finally, the report touched on a few of the primary challenges that hamper agentic transformation, which include:
- Risk management, compliance, security, and/or legal implications (42% of respondents)
- Lack of internal expertise when it comes to AI/agent design (41%)
- Legacy infrastructure or incompatible systems (37%)
- Integrating siloed apps and data (35%)
Bridging Integration Gaps With a Unified Foundation
Finally, in an effort to deal with integration gaps, organizations are moving toward unified foundations, where fragmented AI tools are turned into cohesive multi-agent systems.
The ability to share data between systems is crucial to the eventual success of AI agents, as 96% of leaders agree that this success depends on data integrating seamlessly across systems. 94% of these leaders also believe that AI agent success depends on IT infrastructure being more API-driven, as these APIs are crucial for connecting apps, data, and AI.
I think the information covered in this report is crucial for all businesses considering using AI agents to understand. It highlights the potential growth of AI agents and the high expectations and excitement surrounding them, while also being clear about the challenges and roadblocks.
If ecommerce brands want to build and deploy AI agents successfully, it’s important to start with specific use cases to test and then scale slowly. If you start with a full overhaul without a plan in place, you may realize that some systems, infrastructure, and business areas aren’t ready for AI agents or won’t benefit from them.
You should also define clear key performance indicators (KPIs) and measure them throughout implementation and use of the agents, to ensure they’re delivering results that justify the costs. Also, take time to ensure your data is clean before using AI agents, and have strong security measures in place to ensure you comply with any necessary regulations in your area.
