Qlik has published the results of its study on Agentic AI, in which it analyzes how large companies are planning, financing and deploying this new generation of artificial intelligence. The study, carried out by Enterprise Technology Research (ETR), shows that there is a strong commitment to Agentic AI, although there are still important obstacles in its actual implementation.
Almost all of the companies surveyed have already allocated a specific budget to this technology, but most recognize that it will still take several years to be able to operate it on a large scale. Data quality and integration with existing systems are the main brakes.
“Companies do not lack ambition or resources. What is missing are adequate databases and analytics that allow agents to work throughout the organization with reliability and control,” says James Fisher, Chief Strategy Officer at Qlik. “If we want Agentic AI to really make a difference in 2026, we must first invest in having trusted data, interoperability to obtain a clear and realistic ROI.”
Main conclusions
- Companies are betting and investing in Agentic AI. 97% of large companies have allocated funds to Agentic AI. 39% plan to invest more than $1 million, and 34% reserve between 10% and 25% of their total AI budget. Agentic AI is already part of the official budget items, which increases expectations of visible results in 2026.
- Strategies are clearer, but value measurement fails. 69% of companies say they have a formal AI strategy, up from 37% in 2024. However, Only 19% have a defined framework to measure return on investment (ROI). The conversation about governance has moved from “should we do it?” to “what results are we getting?”
- Deployment will take time. Only 18% have fully implemented Agentic AI, and 46% estimate that full adoption will take three to five years. Additionally, less than half (42%) fully trust their in-house expertise. For the majority, 2026 will be a year of construction and preparation, rather than massive deployment.
- Data remains a critical point. Data quality, availability and access top the list of obstacles, followed by integration, technical competencies and governance. In practice, the problem is more in the business infrastructure than in the power of the AI models.
- The risk increases in the deployment phase. The main concerns are cybersecurity, trust in results, and legal issues, closely followed by the ability to explain and audit models’ decisions. Those responsible for mitigating risks will set the pace and purchasing decisions in the coming months.
- Where AI agents are being deployed. The IT and software development areas are those that lead the first deployments, with cost reduction as the main objective and productivity as a key metric. The first success stories are concentrated in environments where telemetry systems and clear baselines already exist.
As a main conclusion, the study reveals that Agentic AI has passed the experimental phase and has entered 2026 operational plans. Organizations are opting for pragmatic strategies, focused on measurable use cases within IT operations and software engineering, where the results are more tangible.
The study reveals that Agentic AI has passed the experimental phase and has entered the 2026 operational plans
The challenge now is not so much the capacity of the models, but how to integrate reliable and governed data into existing workflows, connecting systems without adding operational risks. Until that happens, many initiatives will remain pilots or proofs of concept, rather than fully productive operations.
““As AI spending moves from experimentation to fixed budget lines, the challenges are the classic ones of any large company: data quality and integration, governance and specialized talent,” explains Erik Bradley, Chief Strategist at Enterprise Technology Research (ETR). “Our study shows widespread intent to move forward, but only a minority are ready to scale. The next year will be key to turning narrow IT and development projects into stable, measurable implementations.”
