AI in healthcare: Top 5 pitfalls in clinical communication workflows

A smarter clinical communication AI strategy

At HIMSS26, unsurprisingly, one topic consistently surfaced across sessions and conversations: AI is moving from experimentation to operational deployment within healthcare. Organizations are exploring AI across ambient documentation, virtual agents, predictive analytics, and patient outreach. At the same time, many health systems are still working through fundamental challenges in clinical communication workflows.

What is a clinical communication workflow?

Clinical communication workflows include any communication channel that supports or is triggered by a patient interaction. If it impacts how patients connect with providers, access care, or receive information, it is part of the workflow. This may include:

• Inbound and outbound voice communications
• Appointment reminders and confirmations
• Nurse call systems
• SMS and MMS messaging
• Email communications
• Secure messaging platforms
• AI-powered virtual agents
• Legacy telephony systems

When these channels operate independently across departments or platforms, organizations often experience reporting gaps, duplicated spend, and inconsistent patient experiences.

What is driving AI adoption in clinical communication?

Three primary factors are influencing adoption:

• Ongoing pressure to improve efficiency while managing labor costs
• Increased reliability and scalability of cloud platforms 
• Maturation of AI capabilities across the omnichannel experience 

AI solutions are being evaluated for their ability to:

• Reduce average handle time 
• Improve first-call resolution 
• Increase appointment adherence 
• Automate high-volume inquiries 
• Improve staff productivity

However, outcomes vary significantly depending on governance, integration planning, and cost alignment. At GDT, we understand the benefits AI can deliver; working with our team for strategic implementation, our customers have achieved results including: 

• 40% reduction in telecom costs
• 95% reduction in call center costs 
• 2 minutes saved per call
• $1M saved on 3-year telecom costs 

Is agentic AI the right solution for your organization?

While agentic AI-powered solutions offer a seemingly simple solution for streamlining and automating workflows and alleviating overburdened teams, it’s important to assess how these AI-enabled tools align to your operational realities, cost structures, and patient communication strategies before expanding AI initiatives. 

GDT has worked with leading healthcare organizations to evaluate and deploy AI solutions within clinical communication environments. When aligned with clear business objectives and a structured implementation strategy, AI can streamline operations and deliver measurable outcomes across patient care delivery. However, organizations that rush into AI initiatives without defined priorities or governance often encounter avoidable setbacks. Below are five areas healthcare leaders should evaluate before scaling AI further. 

The top 5 healthcare AI pitfalls

5. Disconnected AI point solutions

Healthcare organizations already manage multiple communication platforms across departments. Introducing AI without consolidating AI point solutions can increase complexity. Common overlapping communication tools include: 

• SMS for patient engagement 
• Third-party outbound notification platforms that duplicate native platform capabilities 
• Shadow AI tools purchased without enterprise review 

The operational risks include: 

• Fragmented reporting 
• Increased security exposure 
• Redundant usage costs 
• Limited visibility into performance 

Before adding new AI tools, organizations should inventory existing capabilities and evaluate overlap. 

4. Biting off more than you can chew

AI can support a wide range of use cases, but initial deployments are most successful when focused on repeatable, high-volume tasks. Examples of practical starting points include: 

• Hours and location inquiries 
• Appointment confirmations 
• Intelligent call steering and deflection
• Prescription refills 

For example, if a system receives 100,000 monthly calls and 10% to 20% are related to hours or directions, that may represent a clear automation opportunity. 

Expanding AI use cases without first validating performance in controlled environments often leads to stalled adoption.

3. Failure to define ROI

AI deployments should be mapped to operational metrics before purchase decisions are finalized. Each use case should be evaluated against: 

• Current call volumes
• Labor allocation
• Cost per interaction
• Patient experience metrics
• Avoided transfers
• Reduced handle time

Organizations sometimes invest in AI platforms without defining the baseline metrics required to validate ROI. Cost alignment should be established prior to deployment.

2. Not doing a real-life pilot based on the real solution you plan to use

Proofs of concept must be set up to prove real solutions, not just demonstrate manufacturers’ capabilities, to ensure you will achieve the outcome that you are looking for. Common gaps include: 

• Lack of integration planning with Epic or other electronic health record (EHR) systems 
• Undefined cloud migration strategy 
• Limited change management planning 
• No defined reporting framework 

Without a roadmap, successful pilots may not scale effectively. AI deployment requires structured milestones, governance, and operational ownership.

1. Operating without a defined strategy

The most consistent differentiator between successful and stalled AI initiatives is the presence of a defined strategy. Effective programs typically: 

• Analyze communication reporting data before selecting tools.
• Identify repeatable workflows with measurable impact.
• Align AI initiatives to operational KPIs.
• Coordinate across departments. 

Integrate AI strategy with cloud and communication modernization plans. Organizations that purchase tools without defined outcomes often revisit platform decisions within 12 to 18 months. 

What works and what doesn’t

Effective approaches: 

• Consolidating communication platforms into a unified management view
• Aligning AI use cases to documented reporting data 
• Establishing governance across departments 
• Creating a clear, defined plan based on tangible outcomes 

Higher-risk patterns: 

• Department-level AI purchases without enterprise oversight 
• Investing in custom development when native functionality exists 
• Adding tools before assessing infrastructure impact 
• Expanding AI use cases before validating initial results 

AI adoption requires operational discipline and executive alignment. 

GDT for clinically aware healthcare AI 

GDT is a technology solutions integrator with healthcare-specific expertise and strategic experience as an Epic app development partner. Our focus is on improving efficiency and performance across patient-driven communication workflows through a structured advisory-led model. GDT has deep experience helping healthcare organizations increase clinical and operational productivity and reduce risk and cost volatility to deliver an efficient patient experience. 

Get started with a healthcare technology strategy workshop

If you’re looking to improve clinical outcomes with AI (and avoid common pitfalls), register for our healthcare technology strategy workshop. This workshop is tailored to help align your healthcare infrastructure, AI, and technology solutions to measurable business outcomes. Our team can walk you through strategic options around contact center and collaboration, ambient and EHR AI, RCM automation, cyber resilience, and more.

Reach out to schedule your no-cost healthcare technology strategy workshop. We can help you devise a plan that delivers tangible outcomes. We look forward to learning more about your goals and helping you realize the outcomes that matter most.