3 Pillars to AI Success: Leaders Playbook
Generative AI has moved beyond the “if” conversation, the question is “how”.
The MIT study, published earlier this year, stated 95% of AI Pilots fail to achieve measurable ROI from their AI pilots. Here is how to set your organization up for success!
Drawing on four successful launches and dozens of conversations with peers, consultants, and vendors, I’ve observed clear patterns of what drives success, and what derails it. Generative AI’s impact hinges not only on technology, but on three pillars
This playbook offers practical takeaways: align leadership on a North Star goal, adapt delivery for AI’s unique traits, and prepare your data for transformation.
Shift the Mindset
Generative AI compels us to think differently and stretch the boundaries of creativity. Yet in the pursuit of innovation, many organizations overlook two critical factors that often determine success or failure: Organizational Change Management and Delivery Methodology.
Organizational Change Management
Change Management is an important aspect for any implementation. However, with Generative AI, to achieve transformational success, Change Management is imperative at the organizational level.
Assessing your organization’s AI readiness and maturity is essential to implementing solutions at speed and scale. When the foundation is set correctly, organizations will unlock:
Transformational use cases demand not just adoption, but deep process alignment across functions and regions. This requires a fundamental rethinking of how cross-functional and cross-regional teams collaborate across the customer lifecycle. Ignoring this step risks undermining the very transformation leaders hope to achieve.
The organizations that realize the greatest success will be those that view Generative AI as a force to augment and support teams, not replace them. Separating the hype from reality, allows you to focus on use cases available today, while preparing for the possibilities of tomorrow. When implementing a Sales Development Rep (SDR) Agent, one business unit saw exponential success, the other stagnation, leveraging the same agent. The difference? Mindset. The business unit that incorporated the agent in their daily workflow achieved much greater success in meaningful connections and revenue attainment.
Delivery Methodology
One of the defining traits of being human is our individuality - in speech, mannerisms, and expression. These differences, while natural for us, create unique challenges for AI systems that must interpret and respond effectively. This human aspect must be accounted for in our delivery planning. Have you sent or received an email that was interpreted differently than intended?
Our Service Agent experienced a 4x TRUE Resolution Increase after evaluating customers interactions and adjusting the agent to merge two topics into one. While User Acceptance Testing (UAT) helps teams adjust before launch, it cannot ensure that all end users will phrase their questions in the same manner. This is exacerbated with external agents, as external customers bring their own diverse language and expectations.
This variability aligns naturally with the Agile methodology: Generative AI solutions demand rapid iterations and continuous adjustments. Embedding structured monitoring cycles and a feedback loop mitigates risk while building stakeholder confidence.
When planning Generative AI delivery, leaders must account for two critical phases beyond deployment: Monitoring and Adjustment. Post deployment, the team monitors the agent interactions to ensure the agent:
Insights from these interactions must then inform structured adjustment cycles, enabling teams to refine performance and correct deviations. Embedding these phases into the roadmap not only mitigates risk but also sets clear delivery expectations for leadership and stakeholders - building confidence in both the technology and its business impact.
Navigate to Your North Star
Over the past year, media and social platforms have propagated the narrative on how organizations have quickly achieved ROI from their Generative AI implementations. This visibility has been a double-edged sword: it inspires new thinking, yet also set unrealistic timeline expectations for complex use cases. This may be achievable if your organization is AI ready:
Most organizations will not have all those stars aligned. The question becomes, “How do I achieve ROI from our investment, while working towards those Attractive, Extensive, and typically longer delivery timelines?”
Begin with realigning with leadership priorities, consider the core asks across functions:
Leaders are asking their teams to achieve more within the same time. This defines our North Star: Time Savings. By reducing admin and time intensive tasks, teams can redirect their energy toward
Recommended by LinkedIn
Keeping the Time Savings goal in mind while you develop your roadmaps will be key in demonstrating ROI and progress,
Discovery
Regardless of delivery methodology, sufficient time in the Discovery Phase will lead to greater success. In addition to ideating and defining the use cases, capture the
Prioritization
When prioritizing Use Cases, evaluate them against clear criteria:
Roadmapping
The roadmap is ready to be compiled once the above activities have been completed. At the outset, focus on use cases that can be delivered within the Agent team. Create a runway for supporting teams to conduct any prep work. Your roadmap should balance complex and low lift use cases to ensure continuous value delivery. In our implementation, a 4 month runway was required to establish Salesforce Data Cloud and a Unified Profile.
Anchor back to the North Star: Time Savings when communicating ROI and roadmapping. Research shows that saving an employee just five hours per week equates to one full month annually - a powerful reframing that refines the North Star to ‘5 Hours Saved per Week . Select measurable KPIs, and ensure a baseline is captured to demonstrate value delivered.
The below depicts two examples of light effort use cases that can provide ROI in a short timeframe, while making significant progress towards your 5 hour goal.
Some may ask why these outcomes are not expressed in direct revenue terms. Efficiencies are quantifiable; revenue impact follows when time saved is strategically reinvested.
The Corporate Gold: Data
As Shawn “Jay-Z” Carter once noted, “Men lie, women lie, numbers don’t.” The statement holds power, but only when numbers are placed in the right context. In business, numbers only tell the truth when the data is well governed and quality controls implemented. For Generative AI to deliver accurate and trustworthy outcomes, disciplined data management is not optional; it is the foundation.
Data Readiness & Investment
Process standardization, data quality, and technical debt remediation have traditionally ranked low on organizational priority lists. A mindset shift is required to succeed, as an AI solution is only as strong as the data that powers it. Leaders must prioritize technical debt cleansing, data, and process initiatives, alongside the AI initiatives themselves.
When our Service Agent initially showed a low deflection rate - despite 70% of inquiries being FAQs - we discovered the root cause was outdated and contradictory knowledge articles, not the platform. We started small and worked with the Global Enablement team to review and update the highest accessed articles.
The goal is to think big, but start small, building momentum through achievable data initiatives. When composing the roadmap, Data Readiness and Common Data Elements were attributes that were accounted for.
Data Analysis & Alignment
During Discovery, an initial data analysis informs the roadmap and helps estimate delivery effort. Once the data team identifies source systems, tables, and fields for ingestion, a deeper analysis should follow. Before data ingestion begins, close collaboration with the AI (or agent) team is essential. Together, the teams can validate data sources, identify usable fields, define transformation rules, and refine requirements - preventing costly rework and ensuring smoother delivery.
Data Team Resourcing
Each organization must determine the pace of its Generative AI investments. Some will begin cautiously, while others will pursue an aggressive, enterprise-wide approach. Whichever approach is chosen, organizations must ensure their data teams are sized to support their corresponding AI initiatives. Without adequate resourcing, data teams can quickly become a bottleneck, stalling progress on high-impact use cases.
Matching Agents Role & Data
It is crucial for Agents to have a clearly defined role. Every agent has a distinct purpose and requires specific, tailored data to perform effectively. Access to incorrect types of data may actually produce the opposite impact you are wishing to achieve. The wrong data not only limits performance, it actively erodes trust and undermines adoption. Imagine a service agent serving up marketing content to an irate customer, or an SDR agent delivering implementation instructions when asked about value. Misalignment confuses customers and undermines adoption.
Generative AI success is not determined by technology alone. It depends on three pillars: a mindset shift to embrace organizational change, a delivery methodology built for iteration, and a strong scalable data foundation. When guided by a clear North Star - such as measurable time savings - organizations can move from experimentation to transformation. The path forward is clear: start small, iterate quickly, and invest in the foundations that unlock enterprise-wide impact.
What has been your experience with Generative AI implementations? What have been your largest hurdles, and how have you leaped over them?
“One of the defining traits of being human is our individuality” I absolutely love this line. Very well written article Al Kassam and thank you for sharing it with me. We need more such stories from the trenches that distinguish hype and reality. Bravo 👏 👏👏
Great guide Al Kassam! You laid out practical yet critical points to consider when transitioning to an Agentic enterprise. That you managed to tie in a Jay-Z quote makes this even better 😆
This is very well thought out and insightful, deff worth the read! Thanks for sharing Al Kassam