Orchestration Is Key To Crossing The AI Trust Chasm: Report

At this point, the consensus is clear: While enthusiasm for AI agents remains high, adoption struggles are keeping businesses from realizing the transformational benefits of the technology.

Only 9% of organizations have fully deployed agentic AI, according to the recent “From the Edge to the Core: Bringing Agentic AI to the Heart of the Enterprise” report from Harvard Business Review, Workato, and AWS. The next two years are critical: 86% of respondents are preparing to increase their investments in AI agents, per the Harvard Business Review Analytic Services survey of over 600 IT decision-makers. And as this technology edges closer to the core of business operations, the biggest barrier many companies face is trust.

“Organizations are realizing that AI models alone cannot deliver predictable outcomes. They are investing elsewhere to improve their agentic strategies,” Workato CIO Carter Busse wrote in the report.

Just 6% of respondents said they fully trust agentic AI to securely and accurately handle core, end-to-end processes. Results are better for less mission-critical use cases, and those where employees and AI agents work closely alongside one another. When supervised, trust in agentic AI rises to 30%. And when deployed in a very limited scope, enterprise trust goes up to 43%, per the survey.

To close the trust gap and prepare for increased use of agentic AI, 74% of organizations are either currently working on, or planning to implement enterprise orchestration. Delivered through a unified platform, orchestration helps to solve the three biggest barriers to trust: Context, connectivity, and control.

“If your data is not amplified or orchestrated across your different systems, you get a garbage-in, garbage- out solution. And then when it doesn’t work, people blame the AI,” said Ramanujam Theekshidar, chief digital office, U.S. Electrical Services.

For example, siloed systems make it difficult for AI agents to tackle end-to-end workflows without running into barriers that require human involvement. It’s why 82% of survey respondents are aiming to use orchestration to connect applications. Meanwhile, 80% are deploying it to provide the relevant data and context so AI agents can take action. And 56% believe orchestration will help navigate the emerging problem of AI agent sprawl.

“If you can’t automate a process, you can’t easily‘agentify’ it,” explained Workiva CIO Kim Huffman. “If a workflow is automated and the data is clean, it is probably more ripe for agentic AI.”

‍The main advice: Start small

Beyond building a unified, governed, and orchestrated foundation to support the AI deployments, early AI leaders all tend to take another common tactic: starting small.

Software provider Amplitude, for example, identified "every very simple low-impact use cases that an IT agent could perform,” according to Vikram Singhvi, vice president of corporate engineering and information technology. These repetitive, low-risk workflows, like adding new users to online teams, made-up as much as 40% of the company’s support tickets. By automating them, Amplitude freed its IT team to focus on building new AI use cases, per Singhvi.

Overall, 56% of respondents expect AI to impact the IT department, with operations, marketing, and sales also expected to see AI-driven disruption. Regardless of the department, a “start small” approach helps to make sure AI adoption is moving forward in a secure, cost-effective way that actually produces results for users.

“There’s a lot of hype, so we have taken a very thoughtful and pragmatic approach; one that is grounded in our business outcomes,” Huffman said.

Read the full “From the Edge to the Core: Bringing Agentic AI to the Heart of the Enterprise” report here.