Unlocking AI: The call to up-skill on AI fundamentals

Summary

We're witnessing a fundamental paradox in enterprise AI adoption. Executives believe AI will significantly change their businesses within the next five years. However, only a few organizations are achieving meaningful scale with AI initiatives. We believe this gap isn't technical—it's human.

In this article, we present four pillars for building organizational AI knowledge, discuss common pitfalls to avoid and outline steps for achieving AI fluency across your organization.

The Industry Reality

AI is here and discussed more than ever. One of the challenges is clear: driving meaningful value and ROI for AI use cases is very difficult. Organizations have recognized that AI literacy is paramount to evolve their workforce. However, most organizations are failing in effectively up-skilling their workforces. In creating new solutions, they are also treating it AI as a specialized technical skill rather than transforming the legacy technology approach into a democratized and foundational capability.

Why Traditional Approaches Are Failing

Organizations are repeating the mistakes of previous technology waves by:

• Centralizing AI expertise in centers of excellence. They democratizing general LLM models across the workforce, but are missing the key business component in more advanced capabilities. This approach creates bottlenecks and distance from the actual business processes.

• Focusing on today's AI solution and not building skills and capabilities for tomorrow's possibilities

• Treating AI as a tool to automate existing processes instead of reimagining workflows in an AI driven organization

• Underestimating the learning curve and hands on practice required to develop genuine AI fluency

Four Pillars of AI Capability Building

1. Balance realistic expectations with future AI promise

Organizations must acknowledge AI's current limitations in their approach to AI implementation. The limitations should be shared with the workforce openly. Disillusionment stifles AI initiatives and innovation. The current reality must be paired with a clear vision of how today’s learnings and skills build invaluable skills for an AI driven tomorrow. In 12-24 months, AI will be exponentially more powerful. Some of its applications are impossible to predict today. What we do know is that AI's capabilities and access will continue to grow; AI longer context windows, seamless multimodal capabilities, autonomous workflow orchestration, and near-human reasoning on specialized tasks. Executives must create this compelling vision of AI's transformative potential back to the workforce. They must do so outside of the typical productivity and headcount reduction metrics.

The critical balance:

• Be transparent about current limitations (hallucinations, context windows, reasoning gaps and current business value realization)

• Paint the trajectory. Show how capabilities are evolving at an unprecedented pace(quarterly vs. annually) and bring the workforce on the journey

• Frame "failures" as learnings. "This didn't work today, but here's what it taught us about our data, processes, prompting and current limitations."

Industry example: Pfizer’s R&D transparency1

Pfizer’s R&D teams are using data to predict relationships between biological mechanisms and symptoms as part of their drug discovery process. Leaders have showcased that AI can accelerate data analysis and pattern discovery, but human domain knowledge remains essential to interpret results and guide decisions due to the complexity of the relationships between biology and disease symptoms.

2. Democratize hands-on experience with continuous improvement loops

Provide every employee, not just technical specialists, direct access to AI tools and the opportunity to experiment. Don’t restrict fundamental AI tools to data scientists or IT departments. Deploy enterprise AI platforms (ChatGPT Enterprise, Claude, Microsoft Copilot, etc.) universally across the workforce, while giving employees protected time to explore. Recognize that the people closest to business challenges are often best positioned to identify valuable AI use cases. Help employees understand what's possible by providing and sharing tangible AI examples across the organization. Embed AI into the organizational thinking.

Establish organizational guidelines to ensure there is control over the general usage including security and data privacy and safeguard against a proliferation of AI. Create a feedback forum where employees can showcase their use cases, brainstorm additional ideas and determine opportunities to expand and productionize.

Implementation examples:

• Universal access: Every employee has access to AI tools (ChatGPT Enterprise, Claude, Copilot) with appropriate guardrails

• Experimentation budgets: Teams get "AI innovation hours"—protected time to explore without productivity pressure. These are included in their yearly performance goals

• Internal showcases: Regular forums where employees demonstrate what they've built, normalizing both successes and failures

Industry example: Citi’s AI champions2

Citi has implemented a human approach to scaling AI across its large global workforce. In their approach, they organized 4,000 volunteers across business units to receive training, best practices and early access to AI tools through an “AI Champions and Accelerators Program” as reported by Business Insider. These volunteers served as “accelerators” to champion AI usage across their peers. The volunteers receive ongoing training and share best practices to continue improvement on AI usage. Citi has reported that the scale achieved through democratization and peer to peer support has led to ~70% AI usage across their 182,000 employees.

3. Reimagine AI driven processes. Don’t focus solely on solving current processes

Don’t only focus on ‘plugging’ AI into existing workflows. Leaders in organizations should also reimagine their workflows through an AI-native lens. Traditional automation replicates current steps faster, limiting the potential value. AI-driven transformation has the promise to redesign and streamline entire workflows. Leaders should ask, what workflow outcomes are truly needed? What if the current barriers are not present? This process means stepping back from "how do we make this faster" to "how would we design this from scratch if we could?" The goal isn't to digitize paperwork or speed up approval chains—it's to dissolve unnecessary steps entirely. The future of AI will enable new forms of decision-making, drive new workflows and create experiences that were previously impossible. True AI transformation requires the courage to abandon familiar processes rather than merely accelerate them.

The mindset shift:

• From: "How can AI speed up our current process?"

• To: "How would we design this from ‘scratch’ if we could”

Industry example: BMW’s design for humanoids3

BMW has partnered with Figure AI, a robotics startup, to test humanoid robots (such as the Figure 02 model) in real production environments at its Spartanburg, South Carolina plant. These general-purpose robots have been trialed on tasks that traditionally require manual dexterity. As part of the trials, BMW is reviewing how their current processes should be changed or augmented to incorporate these humanoid robots.

4. Develop skills for the horizon, not just today

Mastering AI today requires a blend of prompt engineering, critical evaluation, and strategic thinking rather than traditional technical skills. Learning how to communicate effectively with AI systems is a critical skill that is built through practice; crafting clear, structured prompts that elicit useful outputs while being prepared to iterate based on the results.

The ability to critically assess AI generated content for accuracy, bias and relevance is equally critical. These systems can produce confident-sounding but incorrect information. Understanding AI’s limitations helps know when to use it and when human analysis and judgement is needed.

The most impactful skill in AI usage is the mindset shift for enabling AI. Recognizing how to creatively apply AI to everyday tasks becomes second nature. AI becomes an everyday tool, not a forced exercise by executives. To a confident user, it’s a toolbox of tools; different interfaces and models selected based on the task at hand. In this scenario, AI begins to augment and accelerates the process of ideation, data analysis, design or thousands of other tasks. We are just scratching the surface of its value in ever day processes. Ultimately, the winners in this shift won't be those who understand the algorithms, but those who understand how to think alongside AI as a collaborative tool.

What to build now:

• Prompt engineering mastery: Understanding how to structure requests, provide context, and iterate. Understand when to ‘jump off’ of AI and finalize the steps through other means and solutions

• AI evaluation skills: Knowing when to trust AI output and when to verify against other sources

• Ethical AI judgment: Recognizing bias, privacy concerns, and appropriate use cases

• Data literacy: Recognizing what data AI needs and how data quality impacts outcomes

• Hands on AI usage: Get familiar with using different AI tools; what models and tools work better for different applications. Understand model pitfalls. Apply AI augmentation in both work and non-work settings.

Industry example: Moderna’s AI academy4

Moderna launched its own AI Academy in partnership with Carnegie Mellon University (CMU) to educate and empower employees at all levels. The academy teaches employees how to identify, integrate, and use AI and machine learning solutions across systems and processes. The curriculum includes both foundational concepts and hands-on learning tailored for working professionals. Hands on learning examples include data analysis, visualization, machine learning fundamentals, and AI ethics.

Common Pitfalls to Avoid

Mistake #1: Waiting for "AI readiness"

• Organizations delaying up-skilling until they have "perfect data" or "clear use cases" are falling behind. Upskilling and solution and infrastructure development should happen in parallel.

Mistake #2: Training only technical teams

• Every function needs AI literacy. This includes technical and non-technical groups within these functions. The highest-impact AI applications can come from those closest to the processes and are often non-technical users.

Mistake #3: One-time training events

• AI capabilities are evolving too rapidly for annual workshops. Effective programs include continuous learning, communities of practice, and regular capability refreshers.

Mistake #4: Focusing only on efficiency metrics

• Early AI wins often show up as quality improvements, faster innovation cycles, and employee satisfaction. Don't only focus on time savings and productivity

Call to Action: Your AI Upskilling Roadmap

1. Establish a baseline

• Develop the vision and reasoning as to why this initiative is starting. Determine the communication strategy for sharing across the stakeholders. How does it fit into the broader strategy? What are the goals and outcomes? What is being asked for?

  • Host a kickoff event painting the vision: where AI is going, why skills matter now, and what success looks like

• Launch an AI capability assessment—where are skills strong? Where are gaps? Where are we missing capabilities? Where are our workforce capabilities?

• Inventory current AI use cases that are currently being used across your organization. Get an understanding of what initiatives are already taking place

• Ensure ethics and responsible use" guidelines are established

2. Establish an up-skilling approach

• Determine the enterprise tools that will be made available

• Gain business leader and executive alignment and endorsement across functions, where applicable

• Establish a communication strategy for communicating across the stakeholders. Consider guidelines, examples, expectations and ultimately outcomes. Ensure the stakeholders understand the vision and value.

  • Consider adding to individual performance targets

• Identify a set of "AI champions" across the functional areas or departments who will serve as peer mentors

• Establish a process and forum to showcase learnings; both wins and instructive failures

• Top support in execution and adoption, consider launching more focused teams that can focus on direct AI tasks such as "AI learning sprints"—structured 2-week periods where teams tackle real business problems with AI assistance

3. Scale, embed, improve

• Identify the top 5 high-impact use cases emerging from experimentation and allocate resources to develop them

• Launch communities of practice around key themes (prompt engineering, AI-assisted analysis, workflow redesign, etc.)

• Update performance reviews and job descriptions to include AI literacy expectations

• Establish continued credential-training paths and hands on up-skilling

• Create communication and governance to continue recurring pathways. Some examples:

  1. Monthly: "AI What's New" sessions where teams share latest capabilities and experiments

  2. Quarterly: Revisit your AI maturity matrix—what's now possible that wasn't 90 days ago?

  3. Annually: Comprehensive skills assessment to identify emerging gaps and evolving training needs

How we can help

We have deep expertise in building and delivering successful digital capabilities biotech manufacturing. Our capabilities span the full transformation journey; from developing comprehensive transformational strategies to designing innovative digital solutions grounded in GxP-compliant workflows. We understand that digital success in regulated environments requires more than impressive demos; it demands a business-centric view grounded in current digitalization maturity, data accessibility, and compliance requirements. Our experts have helped several Fortune 500 biotech and pharmaceutical companies successfully navigate these challenges. We specialize in manufacturing digitalization. We bring battle-tested frameworks for overcoming the unique barriers biotech manufacturers face. We can help you achieve meaningful business value and transform with purpose

Article citations

1. https://www.pfizer.com/news/articles/a_new_frontier_for_ai_helping_scientists_develop_potential_new_medicines

2. https://www.businessinsider.com/citi-bank-ai-accelerators-volunteers-2025-12

3. https://www.figure.ai/news/production-at-bmw

4. https://www.cmu.edu/tepper-news/news/stories/2021/december/moderna-ai-academy.html

Related posts