What’s slowing down AI workflow automation? Our respondents highlight 6 main challenges teams are likely to face in 2025. Practical tips on how to address them.
Table of Contents
According to McKinsey’s latest survey, 78% of organizations report using AI in at least one business function, a notable increase from 72% earlier this year. However, despite this growth, the integration of AI into workflow automation remains limited, with only a fraction of companies achieving full-scale implementation. And even those mostly focus on Generative AI, disregarding other types of AI models.
At the same time, according to another McKinsey survey, companies that use AI in their operations, outperform their competitors across all the sectors.
This prompts a critical question:
— If AI offers such clear advantages—why isn’t its adoption in workflow automation more widespread?
We’re exploring this topic in depth during our free webinar, From Hype to Impact. How Real Team Are Embedding AI Into Their Workflows. It also makes sense to examine the data to understand the key obstacles hindering AI-driven workflow automation and how digital transformation teams can address them in the most efficient way.
The #1 AI Workflow Automation Solution
Zenphi is the only AI business process automation solution that allows you to build scalable and efficient workflows without disrupting your existing operations — completely code-free. Book a call to learn more.
6 AI Workflow Implementation Challenges Teams Are Facing in 2025 — And How To Address Them
1. Employee Resistance and Fear of Job Loss
Change is hard—and the idea of AI-powered tools sometimes sparks fears about job security. For many employees, automation sounds like “replacement” instead of “relief.”
According to Claw, fears around job losses is one of the biggest concerns. 31% of teams cite labor displacement as a key concern when adopting automation and AI. Also, about 1/3 of employees believe that they might have to obtain new skills due to changes in labour demand and workflow automation.
How to Address Employee Resistance Challenge
Successful automation initiatives take a human-centered approach. That means designing workflows where AI handles repetitive, rules-based tasks—while keeping humans in the loop for decision-making, oversight, or creative input.
Even when deploying Autonomous AI Agents, human context and review still matter. The goal isn’t to eliminate roles—it’s to evolve them.
That’s exactly how Zenphi customer Josh Cohen, President of Tavezio, approached it:
- We used to bring in hundreds of interns to complete tasks that AI-powered automation from Zenphi now is completing. Now we still offer interns a chance to learn the business—but they have time and resources to handle more exciting tasks than just comparing numbers in invoices. Win-win
Josh Cohen, President @Tavezio
Instead of eliminating opportunities, AI workflows freed up bandwidth for more meaningful, growth-oriented work. That’s the real power of AI workflow automation done right.
2. Privacy, Security, and Compliance Concerns
Unlike general automation challenges, McKinsey’s latest AI report highlights inaccuracy, security, privacy, and compliance issues as the top concerns IT leaders are actively working to mitigate.
This particular research was focused solely on Generative AI—but the implications stretch far beyond it. Whether it’s AI summarizing documents or automating business approval workflows, data security and regulatory compliance are non-negotiables.
And industry experts agree: companies should think carefully about where their data is stored, how it’s processed, and who has access—before launching into any AI implementation project.
How to Address Privacy, Security, and Compliance Concerns
To mitigate risks and ensure responsible adoption, organizations should choose an automation platform that takes security seriously—from both a legal and infrastructure perspective.
Here’s what to look for:
- Certifications that matter: Look for vendors that are HIPAA, GDPR, and ISO certified, especially if your workflows touch sensitive, personal, or regulated data.
- Model transparency: If you’re leveraging AI models like DeepSeek or others, make sure the model is hosted securely—ideally on your vendor's or your own infrastructure. Hosting AI models from China, for example, may raise compliance red flags depending on your industry and geography.
- Regional data control: Ensure you can choose where your data is stored—whether that’s in the U.S., EU, or another compliant region. Zenphi, for example, allows customers to select from multiple hosting regions.
One of Zenphi’s customers, Robert Meyer from the NYC Department of Education, knows this process well:
- Our IT team spent 3 to 4 months thoroughly reviewing Zenphi’s infrastructure, security posture, and data policies. Only after that due diligence were we able to whitelist the platform. It passed every check.
Robert Meyer, NYC Department of Education
3. Poor Data Infrastructure
A December 2024 study by Ernst & Young in December 2024 echoes McKinsey’s findings, reporting that 83% of IT leaders cite poor data infrastructure as a key factor slowing down AI automation adoption.
Disconnected systems, inconsistent data formats, and outdated integrations can derail even the most advanced AI workflows. The issue isn’t just technical—it directly affects outcomes, introducing errors, duplications, and delays that shake trust in automation altogether.
And it’s having an impact on momentum.
Nearly 50% of senior leaders surveyed said they’ve seen a decline in company-wide enthusiasm for AI adoption due to disappointing results.
Employees are feeling it too—often overwhelmed by the pace of AI developments and unclear on where real value lies.
- Leaders are banking on AI as the future but our research uncovered challenges like data infrastructure, which are holding back adoption. Leaders must put emerging and evolving risks like data and change management at the top of their AI transformation agenda to maintain momentum and realise adoption.
Whitt Butler, EY Americas Consulting Vice Chair
How to Address Data Infrastructure Challenge
You don’t need perfect data to get started with AI-powered workflow automation. But you do need a smart entry point.
Here’s how to avoid overreach and build confidence early on:
- Start with small, manageable projects. Focus on workflows where your data is relatively clean and the logic is simple. Think approvals, structured forms, or internal requests.
- Avoid relying on Gen AI where inconsistency could hurt. Early-stage workflows shouldn’t delegate 100% of decisions to generative models. Instead, use models that are proven and predictable—like Document OCR to extract structured information from PDFs or scans; or Sentiment Analysis to quickly categorize incoming messages or feedback.
- Keep humans in the loop. Let AI handle the heavy lifting, but insert human checkpoints where needed—especially when AI outputs could trigger decisions, emails, or public-facing actions.
Zenphi’s customers successfully use this combination of structured automation and selective AI usage to build trust—not only in the workflows but in the broader automation strategy. When teams see real value from automation (without chaos), internal confidence starts to grow.
Want to See AI Workflow Automation in Action?
Interested in learning how Zenphi users embed AI in their workflows—without overcomplicating or overcommitting? Book a call with our automation experts. We’ll walk you through best practices tested by hundreds of companies worldwide.
4. Complex Integration with Legacy Systems
Integrating modern automation tools with existing legacy systems remains a significant hurdle for many organizations.
A survey by Opteamix conducted in March 2025 revealed that 58% of organizations cite legacy system integration as their biggest challenge in cloud transformation. Similarly, an RTD survey highlighted that 63% of insurance leaders identified financial pressures and legacy systems as core barriers to adopting automation.
These statistics underscore a common dilemma: while automation promises enhanced efficiency and innovation, outdated infrastructures often impede seamless integration, leading to increased costs and project delays.
How to Address Legacy Systems Challenge
Successfully navigating the complexities of integrating automation with legacy systems requires a strategic approach:
- Select Vendors with Proven Compatibility: Opt for automation platforms that are officially recognized and approved by your legacy system providers. For instance, Zenphi is a trusted partner of Google Workspace, ensuring deep integration and consistent performance.
- Prioritize Interoperability: Ensure the chosen automation tool can seamlessly communicate with your existing systems. Platforms designed with interoperability in mind can bridge the gap between old and new technologies, facilitating smoother transitions.
- Adopt a Phased Implementation Strategy: Instead of overhauling entire systems at once, implement automation in stages. This approach allows for testing and adjustments, minimizing disruptions and ensuring each component integrates effectively before proceeding.
By focusing on these strategies, organizations can mitigate the challenges posed by legacy systems and pave the way for successful automation adoption.
- Zenphi shows great compatibility with everything from Google, including Apps Script. While we still use Apps Script regularly, it’s now a part of the larger automation system, that includes Zenphi and 5 other solutions. They are all integrated and function to support each other, not to replace each other.
Head of IT at a Large Distribution Company, US
5. Lack of Scalability or Flexibility
Scaling automation initiatives from pilot projects to enterprise-wide solutions remains a significant hurdle for many organizations. A survey by Boston Consulting Group revealed that 74% of companies struggle to achieve and scale the value of their AI initiatives.
This challenge often stems from automation solutions that are either too rigid to adapt to evolving business needs or incapable of handling increased operational demands. Consequently, organizations may find themselves constrained by systems that cannot grow alongside them, leading to stalled automation efforts and unrealized potential.
How to Address Scalability & Flexibility Challenge
To overcome scalability and flexibility challenges in automation, consider the following strategies:
- Select a Scalable Automation Platform: Choose a platform designed to grow with your organization—without forcing you to automate everything all at once. Especially in the face of economic unpredictability and volatility in 2025, it makes perfect sense to start small and scale at your own pace when you feel ready. Zenphi is one of the few platforms that truly supports this flexible approach. You get access to enterprise-grade features while automating just a single business process—at an affordable price—without dragging your entire company into a full-blown transformation initiative.
- Ensure Flexibility in Workflow Design: Opt for tools that allow for easy modification and customization of workflows. This flexibility enables your organization to adapt processes in response to changing market conditions or internal requirements without extensive reconfiguratio
- Prioritize User-Friendly Interfaces: Adopting platforms with intuitive, no-code or low-code interfaces empowers non-technical staff to create and manage workflows. This democratization of automation fosters broader adoption and reduces reliance on specialized IT resources.
As numerous Zenphi’s customer confirm, it’s not that hard to build a robust automation framework that not only meets current operational demands but is also poised to adapt to future challenges and opportunities — merely by following these simple strategies.
6. Limited Executive Support or Organizational Readiness
Securing executive buy-in and ensuring organizational readiness are critical for the successful adoption of AI and automation initiatives. A study by ScottMadden highlights that 29% of organizations identify a lack of executive support as a primary challenge in adopting predictive or generative AI technologies.
Without strong leadership endorsement, automation projects may struggle to secure necessary resources, face resistance from employees, and ultimately fail to achieve their intended outcomes.
How to Address Lack Of Executive Support Challenge
Winning executive support (and getting the whole organization ready) starts with reframing automation from a tech project into a business transformation initiative. Here’s how to lay the groundwork:
- Tie automation to measurable business outcomes. Instead of pitching automation as a “cool tool,” show how it supports specific goals: reducing costs, improving compliance, speeding up processes, or boosting employee satisfaction.
- Start with high-visibility, low-risk wins. Small, successful projects—like automating invoice approvals or IT access requests—can demonstrate value quickly and help build trust across the org.
- Make adoption easy and non-disruptive. Choose platforms that don’t require big IT overhauls or months of training. Zenphi, for example, allows business users to build powerful workflows while still meeting enterprise IT standards.
- Create internal advocates. Identify stakeholders from different departments who can champion automation and share early wins with leadership.
When executives see automation not as a disruption but as an enabler of smarter, faster work, support follows naturally—and organizational readiness grows alongside it.
Real Barriers — And Real Opportunities
AI workflow automation holds incredible promise—but as we’ve seen, the path to adoption isn’t without its bumps. From security concerns to legacy system integration and internal resistance, it’s no wonder many organizations are still testing the waters.
But the good news? You don’t need to overhaul everything to get started.
By focusing on practical use cases, starting small, choosing the right platform, and keeping humans in the loop, it’s absolutely possible to build smart, scalable automations that add real value—without disrupting your team or your systems.
The barriers are real, but so is the opportunity. With the right mindset and tools, you can move beyond the hype—and start unlocking meaningful impact, one workflow at a time.
FREE EBOOK
Not sure where to begin? Download our free eBook
Real world success stories of AI-powered business process automations
