February 21, 2026
7 min read
AI Implementation
Most organizations have adopted AI. Frontier firms have made it instinctive. Here's what CIOs need to build to close that gap.
Your organization probably uses AI. According to McKinsey, 88% of organizations now do. Wharton's latest research shows 82% of enterprise leaders use generative AI weekly, with 46% using it daily.
But there's a difference between using AI and being built around it. Most companies are still in the "deliberate" phase. Someone identifies a use case, builds a business case, runs a pilot, and maybe gets to production. Harvard Business Review calls this reflective decision-making: "deliberate steps such as data gathering, analysis, and synthesis, layered with experience and judgment." It works. It's also too slow for what's coming.
The companies pulling away aren't deliberating about whether to use AI for a given task. They're reaching for it the way you reach for a keyboard. Microsoft calls these organizations "frontier firms." HBR calls the behavior "reflexive," meaning immediate action that "leverages real-time data, often through AI and automation, to enable fast, context-aware responses."
MIT Sloan Management Review and BCG surveyed 2,102 executives across 116 countries and found this shift is already happening: 76% of executives now view agentic AI as "more like a coworker than a tool." That's not a technology statement. It's a behavioral one.
The gap between reflective and reflexive is where the CIO's next job lives.
What a Frontier Firm Actually Looks Like
Microsoft's 2025 Work Trend Index, based on a survey of 31,000 knowledge workers across 31 markets, defines the frontier firm as an organization "structured around on-demand intelligence and powered by hybrid teams of humans and agents." A follow-up IDC study of 4,000+ business leaders puts numbers on the difference.
Only 22% of organizations qualify as frontier firms today. But those 22% report 4x better outcomes than slow adopters across brand differentiation (87%), cost efficiency (86%), top-line growth (88%), and customer experience (85%). They achieve returns 3x higher than their peers.
What separates them isn't a bigger AI budget. It's how broadly and deeply they deploy. Frontier firms use AI across an average of 7 business functions: customer service, marketing, IT, product development, cybersecurity, and more. Compare that to organizations still running isolated pilots in a single department.
The results show up in production deployments, not just surveys. Capital One built a multi-agent Chat Concierge for auto dealerships that is 55% more successful at converting leads into buyers (Fortune). Salesforce's Agentforce platform has closed 18,000 deals since its October 2024 launch, generating over $500 million in product sales. PepsiCo deployed agents across technology, customer service, and employee experience, and found that AI agents identified technical gaps that humans missed.
The Agentic Shift: From Assistant to Coworker
The behavioral shift goes deeper than "using AI more." It changes what AI is to the organization.
Bain's Technology Report 2025 found that companies using agentic workflow automation were twice as likely to say AI exceeded their goals compared to those using it in assistant mode. Leaders who moved from pilots to scaled agentic workflows achieved 10% to 25% EBITDA gains. The difference between those numbers and the single-digit productivity bumps most organizations see comes down to one thing: whether AI is embedded in how work gets done or sitting alongside it.
MIT SMR and BCG's survey found that 66% of organizations adopting agentic AI expect fundamental changes to operating models, roles, and career paths within three years. Organizations are beginning to "onboard, train, evaluate, and even retire AI agents much like employees." The researchers call this "HR for AI."
Microsoft describes the journey in three phases: AI as an assistant (removing drudgery), then agents as "digital colleagues" (taking on tasks at human direction), then humans setting direction for agents that run entire business processes. Most organizations are somewhere in Phase 1. Frontier firms are building for Phase 3.
Why Most Organizations Are Stuck
If the results are this clear, why aren't more organizations reflexive?
The data points to five gaps.
The ROI gap. PwC surveyed 4,454 CEOs across 95 countries and found that 56% report getting neither revenue nor cost benefits from AI. Only 12% report both. Forrester's survey of 1,400+ AI decision-makers found that only 15% see positive EBITDA impact. ROI confidence is declining: in late 2024, 81% of firms reported 5%+ ROI; by mid-2025, that fell to 62%. The returns most organizations expected haven't materialized, and patience is running out.
The deployment gap. Capgemini surveyed 1,500 executives at $1B+ organizations and found stark maturity numbers: only 2% have deployed AI agents at scale, 12% at partial scale, 23% have pilots, and 61% are still exploring. Compare that to the 40% of enterprise apps Gartner projects will have embedded agents by 2026. The infrastructure is about to assume AI-first workflows, and most organizations aren't ready.
The awareness gap. Microsoft's Work Trend Index found that 67% of leaders report familiarity with AI agents, but only 40% of employees do. PwC's Global Workforce Survey puts it more sharply: only 14% of workers use generative AI daily. You can't build reflexive behavior in a workforce that doesn't understand what's available to them. This is a CIO problem. If your people don't know what agents can do, they won't reach for them.
The trust gap. Capgemini found that trust in fully autonomous AI agents is actually declining, dropping from 43% to 27% in a single year. Ninety percent of executives view human involvement in AI-driven workflows as positive or cost-neutral. The reflexive shift doesn't mean removing humans. It means redefining the handoff between human judgment and machine speed.
The governance gap. Only 18% of organizations have an enterprise-wide AI governance council (McKinsey). Only 21% have mature AI governance (Deloitte). Meanwhile, 83% of AI leaders report major or extreme concern about generative AI, an 8x increase in two years (Gartner). And here's the critical finding: PwC found that CEOs with Responsible AI frameworks are 3x more likely to report meaningful financial returns. Governance isn't a brake on AI value. It's a multiplier.
BCG's research reinforces the point. 70% of AI barriers are people and process-related. Only 20% are technology problems. The CIO who focuses on platforms alone will miss 70% of the problem.
Open Questions
1. How do organizations manage the transition from assistant-mode AI to agentic workflows without disrupting existing operations? Bain's four-level maturity model (assistants, task-doers, supervised workflow agents, multi-agent constellations) suggests a progression, but the path from Level 2 to Level 3 involves fundamentally different governance, security, and organizational design requirements. Most organizations have no playbook for this.
2. Will declining trust in autonomous AI agents slow the reflexive shift? Capgemini's finding that trust dropped from 43% to 27% in one year runs counter to the adoption enthusiasm in other surveys. If trust continues to erode as organizations encounter agent failures at scale, the reflexive behavior that frontier firms depend on may plateau.
3. What happens to AI budgets when the ROI gap persists? With 71% of CIOs expecting budget cuts if targets aren't met by mid-2026 and 25% of planned AI spend at risk of deferral to 2027, the window for proving value is narrowing. Organizations that haven't demonstrated measurable returns may lose the investment runway to reach frontier firm status.
The CIO's Playbook: From Adoption Champion to Organizational Architect
The CIO's role is shifting. Getting people to use AI was last year's job. The next job is making AI usage instinctive, effective, and governed.
Build governance before you build scale. PwC's data is definitive: Responsible AI frameworks deliver a 3x ROI multiplier. Reflexive AI usage will spread faster than you can govern it after the fact. With Gartner projecting 40% of enterprise apps will feature task-specific AI agents by 2026, the governance framework needs to be in place before the infrastructure forces the issue. Deloitte's finding is telling: 85% of organizations expect to customize autonomous AI agents, but only 21% have the governance to support it.
Move from assistant mode to agentic workflows. The 2x performance gap Bain identified between agentic workflows and assistant-mode AI is the clearest signal in the data. Don't keep adding copilots to existing processes. Identify the workflows where agents can own end-to-end task execution with human oversight at defined checkpoints. Capital One and Salesforce didn't build chatbots. They built multi-agent systems that run complete processes.
Close the awareness gap deliberately. The 27-point gap between leaders (67%) and employees (40%) on AI agent familiarity won't close on its own. Closing that gap requires structured enablement: putting AI tools in the workflow where people actually work and showing them what's possible in their specific context. Not a lunch-and-learn. Embedded experience.
Prove value in months, not years. With 71% of CIOs facing potential budget cuts by mid-2026, the luxury of multi-year transformation timelines is gone. Target high-impact, low-risk use cases that deliver wins quickly. Pharmaceutical companies are using agent-based adverse event processing to free up 40% of team capacity. Mortgage companies are projecting that even a 1-2% increase in completed applications represents material bottom-line impact. Start where the ROI is fast and measurable.
Treat digital labor as workforce planning. Microsoft reports that 82% of leaders expect to use digital labor to expand their workforce in the next 12 to 18 months. Ninety-five percent of frontier firm leaders are hiring for AI-specific roles, compared to 78% overall. MIT SMR and BCG found that 45% of agentic AI leaders would reduce middle management positions while 29% expect fewer entry-level roles. This isn't an IT initiative. It's a workforce architecture decision. The CIO who frames it that way earns a seat at the workforce planning table.
The Window Is Narrowing
The data points in one direction. Organizations that figure out the behavioral and organizational side of AI first will compound their advantages. Those that stay in reflective mode, deliberating, piloting, studying, will find the gap harder to close with every quarter.
PwC's CEO survey found revenue confidence at a five-year low: only 30% of CEOs feel very or extremely confident about growth in the next 12 months. Forrester reports that ROI confidence in AI is declining, not improving. And Bain warns: "If you're still piloting, you're dangerously behind."
Microsoft put it plainly: "Every company will become a frontier firm, or be disrupted by one."
For CIOs, the question isn't whether your organization will adopt AI. It already has. The question is whether you'll be the one who makes it instinctive.
Nova Group provides Fractional CIO, AI Readiness & Strategy, and Advisory services for mid-market organizations making this transition. If you're a CIO or IT leader who sees the gap between where your organization is and where it needs to be, and you want a structured path to close it, that's the conversation Nova Group was built for.
Sources
Microsoft WorkLab, "2025: The Year the Frontier Firm is Born" (2025). microsoft.com/worklab
Microsoft / IDC, "Bridging the AI Divide: How Frontier Firms Are Transforming Business" (November 2025). blogs.microsoft.com
Harvard Business Review, "Companies Are Using AI to Make Faster Decisions in Sales and Marketing" (June 2025). hbr.org
McKinsey & Company, "The State of AI: How Organizations Are Rewiring to Capture Value" (2025). mckinsey.com
Deloitte, "The State of AI in the Enterprise: From Ambition to Activation" (2026). deloitte.com
Gartner, "Gartner Predicts 40% of Enterprise Apps Will Feature Task-Specific AI Agents by 2026" (August 2025). gartner.com
Boston Consulting Group, "Where's the Value in AI?" (2024). bcg.com
Wharton School, "2025 AI Adoption Report: Gen AI Fast-Tracks Into the Enterprise" (2025). knowledge.wharton.upenn.edu
MIT Sloan Management Review / BCG, "The Emerging Agentic Enterprise" (November 2025). sloanreview.mit.edu
PwC, "29th Annual Global CEO Survey" (January 2026). pwc.com
Capgemini Research Institute, "Rise of Agentic AI: How Trust Is the Key" (July 2025). capgemini.com
Bain & Company, "Technology Report 2025: AI Leaders Are Extending Their Edge" (2025). bain.com
Forrester, "The State of AI, 2025" (2025). forrester.com
Fortune, "2025 Was the Year of Agentic AI. How Did We Do?" (December 2025). fortune.com
CIO.com, "How to Get AI Agent Budgets Right in 2026" (2025). cio.com
Microsoft / IDC, "Generative AI Delivering Substantial ROI to Businesses" (January 2025). news.microsoft.com