The R&D Leader's Playbook for Agentic AI Success
This article references topics presented by Michael Connell and Michael Heiber during our recent webinar, Making Sense of Agentic AI: A Strategic...
Software & AI
Scientific Software Development, Legacy Software Modernization, UI/UX,
Predictive Modeling, Custom Simulations, Web Applications,
Multimodal Knowledge Systems, API Development
Data Systems
Data Engineering, Process Engineering, Data Pipelining and Augmentation,
Workflow Automation and Redesign, Scientific Data Management Systems,
Data Capture Systems, High Volume Data Management, Database Design
Strategy & Design
R&D AI Transformation, R&D Digital Transformation, Strategic Roadmap Development,
Data System Design, Process Analysis
Infrastructure
Technical Upskilling for Scientists & Engineers, R&D Systems Integration,
R&D IT and Data Ops
Core Technologies
Machine Learning, Deep Learning, Baysian Optimization, Generative
Adversarial Networks, Graph Neural Networks
Advanced Modeling & Systems
Reasoning Models, Multi-Scale Modeling, Surrogate Modeling,
Simulation, Image Processing, Agentic AI Systems
Language & Generative AI
Natural Language Processing, Foundation Models, Generative AI,
Large Language Models
Discovery & Development
Property Prediction, Formulation Optimization, Structure Generation,
Materials Discovery, Materials Compatibility
Data Insights
Text Data Mining, Automated Data Analysis, Time Series Analysis,
Multimodal Search, Literature and Patent Search, Dashboards, Data Visualizations
Decision Support
Chatbots, Predictive Maintenance, Preventative Maintenance, AI
Recommendation Systems
Making Sense of Agentic AI | You can now watch this timely webinar on agentic AI in materials & chemistry R&D on-demand.
Table of Contents
This article references topics presented by Michael Connell and Michael Heiber during our recent webinar, Making Sense of Agentic AI: A Strategic Briefing for Materials & Chemistry R&D Leaders. Link to recording below.
The majority of AI initiatives fail to deliver their promised value and it’s usually not because of the technology itself. Often, initiatives only address part of the problem, focusing on technical implementation while overlooking the business strategy and change management challenges that, in the end, determine success or failure.
Leaders tasked with leading agentic AI initiatives must recognize that they’re not just rolling out a new technology, they’re transforming how work gets done. To achieve success, it requires solving the complete problem across three pillars; doing the right things, doing things right, and driving adoption.
Too often, AI projects fail due to unclear or intangible ROI. As a leader, your first priority is to align initiatives with clear business value. As Enthought COO @Michael Connell states: When "everything is a priority," nothing truly is. Without a clear strategic filter, R&D organizations face real costs such as wasted time, budget and talent on low-value projects, alongside opportunity costs of missing crucial market opportunities while competitors advance. The solution is defining a North Star: a single, clearly articulated strategic outcome that every AI initiative must support.
Consider how AI might help you capture market share, accelerate time-to-market, or create entirely new revenue streams through breakthrough capabilities. To make the right choices, involve all key stakeholders in a discovery process to make sure the chosen project is not only valuable but also attainable. This collaborative approach uncovers hidden blockers like data-privacy constraints or technical execution risks early, and makes sure you're solving real pain points rather than problems that may just sound impressive in presentations.
An agentic AI solution isn't just a technology rollout; it's a new business process that creates entirely new workflows. These new methods require careful orchestration of technology, people, data, and organizational structures.
Start by scoping your initiatives around the business process they will serve, not just the technology itself. Map out the complete workflow from start to finish and identify all the touchpoints, decision makers, and handoffs involved.
Consider using a framework that organizes projects by strategic impact and time horizon. Quick wins (6-12 months) prove value and build momentum. Capability-building initiatives strengthen your core infrastructure and upskill teams to “think digitally.”
A common mistake is underestimating these non-technical costs, leading to budget overruns. It’s important that the team has the necessary expertise, both technical and operational, to deliver the complete solution successfully.
The best technology provides no value if no one uses it. Adoption must be a core part of your strategy from the beginning yet it’s often treated as a post-launch activity. Managing users of the technology, which includes those who may be resistant to change, is critical for success.
Involve them early and consistently in the development process using an active approach. This guarantees the solution meets their needs and avoids surprises at launch. Regular feedback sessions, prototype testing, and iterative design will make sure your solution meets real user needs rather than what you think they need.
You must also budget for adoption and change management, which McKinsey suggests can be as much as half to two-thirds of your total budget. This includes promoting the tool, providing training, and confirming the new process is easy for new employees to adopt, making it sustainable in the long run. By proactively managing adoption, you make sure your investment delivers its intended value.
And as a final note, be sure to build in long-term sustainability. Design your new processes so they're easy for new employees to learn and adopt. If your AI-enhanced workflows are complicated, they'll gradually degrade over time as people find workarounds or revert to old methods.
Beyond the technical implementation and adoption, successful agentic AI projects hinge on creating an organizational culture that embraces continuous learning, experimentation, and adaptation.
An important part of building an AI-ready culture is fostering digital literacy across all levels of the organization. Try to understand not just what AI can do, but also its limitations, ethical implications, and the data dependencies it creates. This doesn’t mean everyone needs to be a data scientist, but instead, there should be an understanding of AI concepts so that there is enhanced decision-making and more effective collaboration between technical and business teams. To do this, you can invest in targeted training programs, workshops, and internal communication that is centered around demystifying AI and highlighting its potential impact on different roles and departments.
Additionally, since an agentic AI system is designed to operate with a degree of autonomy, you should be comfortable with delegating tasks to AI agents and trusting their outputs. However, there should always be clear oversight and human interventions established for when it's necessary. The balance between autonomy and control is an important factor for leaders to take into account.
As you evaluate your current or planned AI initiatives, ask yourself:
The organizations that answer these questions thoughtfully, and execute against all three pillars at the same time, will capture the real value that agentic AI promises. Those that focus only on the technology will join the long list of AI initiatives that worked in the demo, but failed to deliver in the real world.
Want to dive deeper? Our recent webinar on Making Sense of Agentic AI: A Strategic Briefing for Materials & Chemistry R&D Leaders explores real-world implementation examples, and our Guide to Agentic AI in Scientific R&D provides curated resources specifically for R&D leaders navigating these challenges.
This article references topics presented by Michael Connell and Michael Heiber during our recent webinar, Making Sense of Agentic AI: A Strategic...
This article references topics presented by Dr. Michael Heiber at Enthought’s 2025 R&D Innovation Summit in Tokyo. Link to video below.Over the last...
This article was originally published on Forbes and can be foundhere. By Michael Connell, EdD | Chief Operating Officer, Enthought Inc. AI is...