This article was written by Mark Palmer, a data and AI industry analyst for Warburg Pincus and a board member for six AI, data management, and data science companies. Time Magazine named him "A Tech Pioneer Who Will Change Your Life." Mark is a LinkedIn Top Voice in Data Analytics.
While 71% of companies use AI, it’s still early days with respect to automation with agents. The numbers tell a story of mass adoption of AI without mastery of its implications. But the winners are pulling away from the losers at breakneck speed with respect to using AI in general, and are experimenting aggressively with agentic computing. Over the past year, the gap in AI maturity between leaders and laggards has increased by 60%, meaning the best companies aren’t just getting better, they’re getting better faster.
But what separates those who are cracking the code from those still fumbling in the dark? It's not better technology or bigger budgets. It's something simpler and harder to copy. High performers are proactive. They don't wait for AI to work, they make it work and learn new workflows based on early experiments.
Here’s how high performers transition from being reactive with AI to being proactive about building agents that make decisions and automate new workflows throughout their organization.
Leaders Bake Strategic Context Into AI Agents to Frame Action
Context turns AI from a parrot into a partner. OpenAI doesn't know your strategy. Microsoft can't read your culture. Without context, you're paying premium prices for a very expensive echo chamber, and your teams won’t know what actions to take.
Companies know this, but most still use vanilla AI, so they get vanilla results. Smart leaders guide action by injecting context into LLMs during fine-tuning. Rewrite prompts. Retrain models. Feed AI your company’s playbook, not your competitor’s. This steers generative AI models to suggest actions that reflect the values, priorities, policies, principles, differentiation, and strategic intent of an enterprise.
Who creates this context? The C-suite starts it, but they hire new AI-specific roles: AI architects, analysts, compliance specialists, and ethics specialists to shape, synthesize, and share strategic context that guides action. This includes articulating corporate strategy in terms of OKRs, which some refer to as an “AI OKR copilot.”
Adobe uses an AI OKR copilot to automate objective creation as an advisor and track key results for over 60 OKRs and 1,000 key results. This helps them identify and resolve issues faster and coordinate OKR alignment across over 30,000 employees. Utilizing AI at scale to create and track OKRs is an effective way to align strategy and execution.
The key lesson to take from high performers is to be proactive in curating strategic context and sharing it broadly with your team, via customized LLMs, data, and best practices.
Leaders Use AI Agents as a Coach
The best coaches don't wait for you to fail — they prevent the failure. Traditional corporate training dumps information on employees once, then hopes it sticks. It doesn't. Research from 100 years ago shows that, without reinforcement, learners retain only about 30% of new information after 24 hours and as little as 10% after one month. A century later, little has changed; or has it?
Smart companies turn training into action with AI. By encoding training into personal AI coaching agents, lessons are reinforced and put into action. Unilever proved this works at scale. They deployed AI to thousands of workers across 190 countries, delivering just-in-time advice tailored to each person's role and experience level. AI coaching like this helps build employee trust, create a culture of continuous learning, and reduce escalations to senior management.
Here's why it works: Generic training is not only quickly forgotten, it also creates generic results. But AI agents adapt to you when you need it. They proactively monitor touchpoints with customers by tapping into CRM and employ agents to suggest resources and tools. New manager struggling with difficult conversations? AI provides scripts and scenarios specific to your situation. Veteran salesperson facing a new market? A coach reviews what’s actually worked with the top sellers in similar accounts, not some consultant's playbook.
The magic happens in the details. AI analyzes real coaching interactions to process unstructured data conversations from conversation transcriptions, CRM data, and detailed product positioning information. It combines these sources to correlate what works based on the client at hand. It studies successful conversations, failed approaches, and breakthrough moments. Then it customizes guidance based on your tenure, learning style, and current challenges. You get coaching that sounds like your company, not like everyone else's.
Agents deliver coaching insights to employees in real-time, when they’re needed. Traditional coaching requires scheduling, preparation, and follow-up. When a $2 million deal hangs on a client call that starts in an hour, you can’t wait for your coach to return from vacation. AI agents provide talking points instantly, exactly when you need them, drawing on their knowledge corpus, your goals, and your company context.
Companies that master AI coaching agents don't just train employees faster, they create a learning culture that adapts as quickly as business conditions change. While competitors rely on outdated training manuals, AI-coached teams learn from yesterday's successes and apply them to today's challenges.
Leaders Architect AI Governance Guardrails for Agents
Wayward AI decisions aren’t caused by rogue superintelligence; they're caused by missing guardrails.
Organizations without AI guardrails are like surgeons without protocols: dangerous, not daring. Adhering to surgical protocols isn’t just about safety; it enables teams to move quickly and react in real-time to changing conditions. Similarly, in the enterprise, AI guardrails don’t just ensure the safe use of agents; they also serve as a source of competitive advantage.
According to McKinsey, the CEOs of high-performing companies own AI oversight. No delegation. No excuses. CEO ownership is the top factor most strongly correlated with higher actual results, measured by earnings connected to their organization’s use of AI. This includes the policies, processes, and technology necessary to develop and deploy AI systems responsibly.
Governance guardrails prevent hallucinations, flag risky actions, and verify agent compliance. Frameworks like the NIST AI Risk Management Framework are a good starting point, but must be carefully applied to the actions agents take.
Guardrails guide more than nefarious use. Knowledge Harvesters are increasingly employed by leading learning and development teams (L&D) to capture, synthesize, and disseminate knowledge during AI post-training..
Geographic Sensitizers help large global organizations bridge diverse regional work styles, cultural norms, customer preferences, and regulations. By training agentic workflows on geographic differences, agents can suggest actions that consider these idiosyncrasies and cultural norms.
Leaders proactively embed governance guardrails into AI models at the system level, during post-training, and integrate them into agentic workflows and automation. They continuously monitor and improve them. Just as surgical protocols save lives by guiding what is and isn’t allowed, AI guardrails save companies from automated decisions that could kill their competitive advantage.
Leaders Fine-Tune AI Agents by Persona-Tuning
Your AI agents should sound like your customers, not like a robot. Empathy and connection stem from knowing your customer — not just their demographics, but also their daily frustrations, their language, and their fears.
AI agents can assume any voice. Lawyers, doctors, marketers. However, most companies don’t train AI about their audience, missing a significant opportunity to connect with customers as they communicate. “Persona assumption” is training AI agents on how to assume a “personality,” or the perspective of your customer. At home, you make AI sound like Yoda or Seinfeld. At work, make it think like your best customer.
If you sell to lawyers, AI can be fine-tuned with the perspective of the various personas that make up legal teams. An AI agent can better understand the questions and concerns of senior partners by analyzing a brief written by an associate from the perspective of a senior partner. It can add commentary to answer key questions the junior associate missed. It can review arguments for logical flaws and conduct thorough research on relevant, overlooked legal precedents.
To develop this persona-sensitivity, ingest real discussions, real meeting notes, and real examples to fine-tune agents about the terminology of their patrons, customize LLMs with this language, and then prompt agents to make decisions and recommendations from a specific point of view (for example, as a senior partner or associate).
The dimensions of persona assumptions are infinite and customizable to your business. Proactive companies recognize this and set out to fine-tune AI agents to always consider their audience, goals, and the context in which they’re made.
Leaders Are Proactive About How They Embrace Agents
According to Pew Research, 52% of employees loathe AI, and just 10% love it. As a result, employee resistance kills more AI projects than bad technology. Fear beats logic every time unless you plan for it.
That fear makes sense. AI agents threaten jobs. But smart leaders turn fear into curiosity. Instead of imposing AI ultimatums or heavy-handed AI rules, adopt an inclusive and collaborative approach. They emphasize human agency, privacy, and control when using agents and explain guardrails. Some employees resent when AI is “done to them, not for them.” To counter this perception, carefully discuss how agents augment work and improve existing workflows, rather than merely serving as a cost-cutting measure.
Reevaluate and incorporate AI enablement so that teams are informed about AI, how agents work, and how to use them. Explain how and why AI is monitored and how you’ve incorporated guardrails for ethics, bias, and privacy. This insight into the principles that guide your use of AI will help reduce resistance and set a tone of inclusion and positivity about AI.
But even with these good intentions, resistance will persist. Today, 31% of employees actively sabotage AI adoption efforts. In unreasonable cases, leaders need to make changes: Replace resistance with enthusiasm and culture changes.
Organizations change more slowly than the rate of change in AI technology, leaving some employees behind. Make this strengthening of culture deliberate, respectful, and proactive. Over time, fear will subside, and your culture will become more proactive.
AI Success With Agents Is About Proactive AI Leadership
Technology is easy. Thinking differently is hard. High-performing cultures think differently about encoding their strategy in AI agents, democratizing access to persona understanding, and sculpting organizational structure to embrace agents. The shift from reactive to proactive decision-making isn't about technology, it's about mindset.
Master this transition, and you don’t just decide faster. You use agents to implement strategies faster, coach employees in real-time, and turn AI skeptics into fans to make better decisions more quickly than your competitors.