The #1 Reason Why 3 Out of 4 Companies Fail at Building Their Own AI Agents

Agentic AIs have gained significant traction as companies seek to automate processes, streamline workflows, and unlock the value of generative AI. In fact, analyst firm Forrester named AI agents as one of its top 10 emerging technologies for this year. But despite this growing interest, companies are being warned: don’t try to build your own AI agents alone.

According to Forrester, three-quarters of organizations that attempt to build AI agents in-house will fail. This shouldn’t come as a surprise, given the complexity involved in building and deploying agentic AIs. The process requires significant expertise in AI, machine learning, and data engineering.

For companies that do fail, there are alternatives available. They can turn to outside AI consulting firms to build custom agents or use agents embedded in software from their current vendors. As Forrester analysts Jayesh Chaurasia and Sudha Maheshwari note, “Savvy firms will grasp current limitations and lean on their vendor and systems integrator partners to build agents at the cutting edge of this technology.”

But some companies are determined to try their hand at building their own AI agents. Goldcast, a software developer focused on video marketing, has experimented with a dozen open-source AI models to assist with various tasks. By linking these models together, the company aims to create agents that can perform tasks autonomously without human intervention.

Goldcast’s approach is more feasible than building AI agents from scratch, especially for companies without extensive AI expertise. By harnessing the power of open-source AIs, companies can avoid the complexities and costs associated with building their own discrete AI models. As Lauren Creedon, head of product at Goldcast, notes, “I don’t want people to think of [AI] as hard and a specialized thing that only people with PhDs can work with. The more people who are enabled on how to work with it, and the more teams that work with it, the better outcomes will get, not only for business operations, but for customers.”

However, even companies like Goldcast acknowledge the challenges involved in building AI agents. A fully formed MLOps plan is essential, and organizations will need advanced teams to set up and integrate the various components involved. Creedon notes that in many cases, organizations will need to turn to outside specialists to set up AI agents.

Slate Technologies, a data analytics provider for construction and related industries, has had success building its own AI agents. The company began rolling out its own AI agents three years ago, even before the AI boom kicked off with the release of ChatGPT. According to Senthil Kumar, CTO and head of AI at Slate Technologies, the key to building successful AI agents is human supervision.

As Kumar notes, “It’s a collaborative process of evolving between the whole AI ecosystem and the human counterparts. The focus would be on how those agents would learn, the knowledge acquisition of agents, and how the agents are going to be able to disseminate knowledge.”

But for many companies, the decision to build their own AI agents or work with a consultant is a difficult one. Chris Ackerson, head of AI at AlphaSense, notes that large companies may be tempted to roll their own highly customized agents but can get tripped up by fragmented internal data, underestimating the resources needed, and lacking in-house expertise.

AlphaSense has trained its own AI agents, but many companies lack internal expertise, Ackerson notes. In addition, organizations may project the development costs but ignore the cost of ongoing maintenance. As Ackerson notes, “This is the largest cost, as maintaining AI systems over time can be complex and resource-intensive, requiring constant updates, monitoring, and optimization to ensure long-term functionality.”

Partnering with an AI provider can give companies access to proven, ready-made agents that have been tested and refined by thousands of users. As Ackerson notes, “It’s faster to implement, less resource-intensive, and comes with the added benefit of ongoing updates and support — freeing companies to focus on other critical areas of their business.”

Adnan Masood, chief AI architect at UST, agrees that building AI agents from scratch is not always the best approach. “Reinventing the wheel is indeed a bad idea when it comes to complex systems like agentic AI architectures,” he notes. By turning to specialists or adopting pre-built solutions, companies can leverage the expertise and experience of those who have already navigated the challenges involved, ultimately increasing their chances of success.