80% Say AI Delivers ROI. Only 10% Are Past Stage 3.

⏱️ Your Weekly Brief

Welcome Back,

Last week, we started to dig into the year end State of AI reports and predictions for 2026. This week, three heavyweight reports landed on our desk from Anthropic, BCG, and Deloitte. The reports converge on an uncomfortable truth: Most organizations are mistaking motion for progress.

Let’s dive in 🎄

🔗 Anthropic's 2026 State of AI Agents Report

The GPTLDR

Anthropic surveyed 500+ technical leaders and found that AI agents have officially crossed from experiment to infrastructure. 80% report measurable ROI today—not projected, not piloted, actual returns. But the report also reveals that most organizations are just scratching the surface of what's possible.

The Details

Current Deployment Reality

• 57% of organizations now deploy agents for multi-stage workflows; 16% have progressed to cross-functional processes spanning multiple teams

• 86% have moved AI coding agents into production environments—enterprises lead at 91%

• 42% now trust agents to lead development work with human oversight

Impact Beyond Engineering

• Highest-impact use cases outside coding: data analysis and reporting (60%), internal process automation (48%)

• AI agents free up time across the entire development lifecycle—planning (58%), code generation (59%), documentation (59%), testing and review (59%)

2026 Trajectory

• 81% plan to tackle more complex use cases—39% developing multi-step processes, 29% deploying cross-functional projects

• Top barriers remain integration with existing systems (46%), data quality (42%), and implementation costs (43%)

• SMBs face an additional challenge: 51% cite employee resistance and training needs vs. lower rates at enterprises

The Workforce Shift

• Agents are redirecting human time: 66% more focus on strategic work, 60% on relationship building, 70% on skill development

• The message is clear: agents are handling execution while humans move to judgment and strategy

What Matters

• The proof point has landed. Eight in ten organizations seeing real economic returns means we're past the "will this work?" phase. The question is now "how do we scale what's working?"

• Hybrid builds dominate. 47% take a hybrid approach—off-the-shelf where it works, custom builds where it creates advantage. Pure build-or-buy strategies are losing.

• Data is the real bottleneck. Integration and data quality concerns top the barrier list. Your AI strategy is only as good as your data foundation.

What Matters

Here’s what leaders are doing differently at Frontier firms.

• Enable connectors for secure data access (25% of enterprises still haven't)

• Standardize workflows through reusable Custom GPTs

• Maintain executive sponsorship and clear mandates

• Invest in data readiness and continuous evaluation

• Deploy deliberate change management with embedded AI champions

🔗 BCG's AI Adoption Puzzle

The GPTLDR

BCG quantified the disconnect haunting boardrooms: 60% of companies are generating zero material value from AI despite substantial investment. The problem isn't technology—it's that organizations are measuring logins when they should be measuring workflow transformation.

The Details

The Five Stages of AI Adoption BCG identified a clear progression that explains the value gap:

1. Information Assistance (using AI like search)

2. Task Assistance (code snippets, calculations, simple outputs)

3. Delegation (assigning defined tasks like drafting emails)

4. Semiautonomous Collaboration (AI agents planning and executing with human oversight)

5. Fully Autonomous Orchestration (AI managing processes end-to-end)

Where Employees Actually Are

• 85%+ remain stuck at stages 2-3

• Less than 10% have reached stage 4—the inflection point where real value creation begins

• Many organizations mistake stage 1 activity as proof of adoption

The Five Personas Blocking Progress BCG segmented employees by their relationship with AI:

• AI Champions — Visible trailblazers who accelerate peer adoption

• Independent Explorers — Self-starters experimenting beyond official programs (54% use unapproved tools)

• Organizational Adopters — Follow structure, advance cautiously

• Passive Observers — Aware but hesitant; 38% find reviewing AI output "tedious"

• Cautious Skeptics — Experienced professionals wary of reliability and job implications

The Leadership Gap

• Only 25% of frontline employees say they receive sufficient AI guidance from leadership

• 62% of C-suite executives cite talent and AI skills shortage as their biggest value challenge

• Yet only 6% have begun upskilling their workforce in a meaningful way

What Matters

• Quality over quantity. Stop celebrating tool usage. Start measuring how fundamentally work is being reinvented.

• One-size-fits-all fails. Different personas need different interventions. Champions need amplification; skeptics need peer demonstrations; observers need protected learning time.

• Manager multipliers. Frontline managers shape whether adoption sticks. One manager who gave a team permission to miss a deadline to experiment with AI saw usage surge.

🔗 Deloitte’s Divergent Thinking Advantage

The GPTLDR

As agentic AI moves from concept to enterprise infrastructure, Deloitte argues the competitive edge won't come from better models—it'll come from who's in the room building them. Organizations that cultivate divergent thinking and neurodivergent p

erspectives will build more resilient, trustworthy AI.

The Details

The Agentic AI Acceleration

• By 2028, 33% of enterprise software applications will include agentic AI—up from less than 1% in 2024

• These systems won't just assist; they'll orchestrate procurement, compliance, medical decisions, and more

Why Different Thinking Wins Deloitte makes the case for cognitive diversity in AI development:

• Creativity in framing problems — Divergent thinkers ask questions others don't (e.g., "What if energy grids worked like ecosystems?")

• Breadth of collaboration — As agents proliferate and interact, designing these systems benefits from varied perspectives

• Rigor in quality assurance — Neurodivergent analysts often excel at testing, error detection, and spotting edge cases

The Performance Data

• Organizations fostering high innovation-related attributes (including diverse thinking styles) outperformed peers on ROI metrics

• Workers who believe their organization overlooks diversity of thought are 60 percentage points less likely to use AI tools daily

• Some corporate neurodiversity programs have shown productivity gains up to 140% in testing roles

The Perception Gap

• More than half of managers believe their teams embrace differing viewpoints

• Only a third of employees agree

• Rigid interviews, AI-driven screening, and consensus-rewarding cultures may be filtering out the difference organizations need

Why It Matters

• Your AI reflects your team. Homogenous teams produce systems that plateau or fail in ways designers can't imagine. Cognitive diversity catches blind spots.

• Trust is cultural, not just technical. Workers connect AI adoption to whether their organization values different ways of thinking.

• Recruit for difference deliberately. Rewrite job descriptions, adapt interview formats, and reward analytical contributions—not just bold presentations.

Stop Drowning In AI Information Overload

Your inbox is flooded with newsletters. Your feed is chaos. Somewhere in that noise are the insights that could transform your work—but who has time to find them?

The Deep View solves this. We read everything, analyze what matters, and deliver only the intelligence you need. No duplicate stories, no filler content, no wasted time. Just the essential AI developments that impact your industry, explained clearly and concisely.

Replace hours of scattered reading with five focused minutes. While others scramble to keep up, you'll stay ahead of developments that matter. 600,000+ professionals at top companies have already made this switch.

Join them today, for free.

📚 Interesting Reads

• Perplexity's first AI agent study finds 36% of tasks are productivity-related

• VentureBeat declares "build vs. buy" dead as AI has collapsed building costs so dramatically that companies should prototype first, then decide whether to purchase.

• BNY's OpenAI partnership has 20,000 employees actively building agents and 125+ use cases live

• CIO argues the role is evolving into "Chief Autonomy Officer" responsible for defining how human and AI judgment coexist

• Wired on the three milestones that will make AI agents ubiquitous

 ➜ Until Next Week

The research converges on a simple point: AI adoption is a transformation problem, not a technology problem.

The organizations seeing real returns have figured out that how people use AI matters more than whether they use it, that managers shape adoption more than tools do, and that cognitive diversity helps teams build better systems.

80% of leaders say AI agents are delivering financial value today. The gap between them and everyone else is narrowing—but it's not closed yet.

Stay curious,

—The GPTLDR Team

AI, simplified for Decision Makers.