Only one in six organizations achieves measurable value from their AI investments. That’s the conclusion of Harvard Business Review Analytics Services, based on a survey of 385 business decision makers conducted on behalf of Appian.
The picture emerging from the data is contradictory. However, AI is now widely accepted, with 59% of organizations already running the technology in production. On the other hand, the impact remains significantly narrow. While 64 percent of respondents report that AI has increased productivity and 58 percent have improved operational efficiency, only 30 percent say AI is contributing to new revenue streams. Less than 35% see an improvement in ROI. Although productivity has increased significantly, the incremental change in business value has not materialized.
The reason lies in the way AI is implemented. Only 18% of organizations are integrating AI primarily into their business processes. Additionally, many organizations do not have the infrastructure in place to run AI. Nearly 7 in 10 respondents say their legacy systems are actively limiting their ability to scale AI. Additionally, 34 percent cited fragmentation or poor data quality as a barrier, and 31 percent cited a lack of integration between systems. Process integration is therefore not only a strategic choice, but also a technical challenge. “Enterprises have reached a tipping point,” says Matt Calkins, CEO of Appian. “Rather than using AI for productivity, organizations must leapfrog to business growth. AI’s true potential will only be realized when it is no longer a standalone tool but an integrated force that generates revenue.”
Integration makes the difference
The numbers confirm that integration makes a big difference. Only 16% of all organizations surveyed achieve high measurable value from AI. For organizations that primarily incorporate AI into their business processes, this number rises to 71%. This study suggests that these two numbers are related. In other words, the deeper AI is entrenched in the process, the greater the benefits. Organizations that also invested in modernizing their legacy infrastructure and consolidating data sources performed even better, with three-quarters of those organizations reporting positive results.
One of the most impressive findings concerns AI agents. They are being deployed more and more frequently, but primarily in relatively safe areas: software development (35 percent), IT operations (31 percent), and marketing and sales (26 percent). Procurement, production, and supply chains, which are the core of operational business, have been significantly delayed in implementation. It is precisely an area where processes are complex, consistency is important, and mistakes can have serious consequences.
Rules and guardrails are essential
And that’s exactly where the real problem lies. 92% of organizations that already use or are considering using AI agents say they need rules and guardrails to help them function safely and effectively. However, less than half (48%) actually document those rules. Research shows that agents operating without a clear framework can behave unexpectedly, creating a risk of unintended consequences. Mark Talbot, vice president of CS AI incubation at Appian, points to additional risks. If employees rely too heavily on AI recommendations without building their own process knowledge, meaningful human oversight becomes an illusion. “Without process knowledge, meaningful human control becomes a sham,” says Talbot. Calkins said in a separate interview with Techzine that the organization already knows what’s missing. “They know what they’re missing. They know it’s too early to deploy AI in strategic applications.” That awareness is there. Execution is delayed.
Researchers are clear on what is needed to take the next step. Organizations need to not only deploy more AI, but also implement it differently, with more clearly defined rules, standardized processes, and better collaboration between departments. 50% of respondents say they are already actively addressing this issue by defining better guardrails. 49% are focused on standardizing workflows.
“Organizations are adopting AI, but many have yet to embed it into the core processes that drive business outcomes,” concluded Alex Clemente, managing director of Harvard Business Review Analytic Services. “Those who successfully incorporate AI into their workflows stand to realize real value.” The urgency is there. So is the intention. 86% of respondents say they want to derive more business value from AI. But intentions without structure do not produce results. As long as AI continues to run outside of the process, a promise will remain just that: a promise.
