AI Automation Is Still Challenging

Highlights

Top Insights

After years of hype, organizations are realizing that deploying AI at enterprise scale is much harder than experimenting with it. The focus is shifting from excitement to real business value and operational integration.

AI agents (systems that can complete tasks independently) have generated massive excitement. But in reality: They still make mistakes and hallucinate. They are vulnerable to prompt injection and security attacks. Most organizations still require human oversight. Experts expect meaningful agent-driven automation within ~5 years, but not today.

Source: Action items for AI decision makers in 2026 (Ideas Made to Matter)

Top News

1. OpenAI announced GPT-5.4, with up to a 1-million-token context window and native computer-use capabilities.
2. Google introduced Gemini 3.1 Flash-Lite, its fastest and most cost-efficient Gemini 3-series AI model.
3. Google released a new Google Workspace CLI that unifies Gmail, Docs, Sheets, Drive, and other apps into a single command-line interface.
4. Researchers released Evo 2, an open-source large genome AI model trained on 8.8 trillion DNA bases that can identify genes, regulatory elements, splice sites, and mutation impacts.
5. Researchers introduced Merlin, a 3D vision–language AI trained on CT scans, radiology reports, and electronic health records that outperformed other models in classifying abdominal abnormalities.

Additional Insights

1. The AI Enterprise: Code Red (Bain)
AI is rapidly evolving from a productivity tool into the operating system of the enterprise, fundamentally reshaping competition, work, and strategy. As AI agents become capable of executing complex, multistep tasks across business systems, companies must rethink how they compete in a world where the marginal cost of intelligence is falling and productivity gains of 30 to 50 percent are possible. Success will depend less on scale and more on learning velocity, proprietary data, and trusted ecosystems that control customer interactions. Organizations therefore need to redesign workflows and workforces around human–AI teams, industrialize the development and governance of AI agents through repeatable “agent factories,” and adopt deliberate strategies to scale AI across the business. The companies that win will treat AI not as a technology initiative but as a full enterprise transformation that reshapes strategy, operating models, and how value is created.

2. Bridging the AI value gap: Are team dynamics the missing link? (Deloitte)
Organizations investing heavily in AI often assume small, agile teams closest to the technology will deliver the fastest results, but Deloitte’s research suggests stronger outcomes come from larger, more connected, and cognitively diverse teams. While 74% of surveyed organizations invested in AI over the past year, most spending still prioritizes technology itself, with only about 7% allocated to training and upskilling, leaving teams underprepared for the behavioral and collaborative changes required to use AI effectively. The research indicates that teams with more than 10 members use AI more frequently and report roughly twice the improvements in efficiency, problem-solving, and innovation compared with very small teams, likely because larger groups can integrate AI into workflows and distribute tasks more effectively. High-performing teams also prioritize cognitive diversity, actively hiring people with varied skills, experiences, and perspectives, which broadens how AI is applied and increases adoption. Equally important is team connectedness: teams that trust one another, learn collaboratively, and operate across functions report significantly greater benefits from AI. Overall, the findings suggest that AI value is realized less through isolated technical expertise and more through well-structured, diverse, and collaborative teams that can collectively adapt their roles and workflows around AI tools.

3. Six breakthrough business models reshaping global growth (McKinsey)
The article argues that a new wave of global growth is emerging from innovative business models pioneered largely in Asia, where companies combine large digital ecosystems, strong consumer engagement, and AI-enabled personalization to unlock asymmetric growth. It highlights six recurring archetypes, including creator- and network-driven commerce, ecosystem superplatforms, hyper-personalized offerings, and trust-centric service models that deepen long-term customer relationships. These models emphasize emotional resonance, community participation, and data-driven customization rather than simply scaling products or reducing costs. AI plays a central role by accelerating personalization, enabling dynamic pricing and recommendations, and coordinating complex platform ecosystems. The broader insight is that growth increasingly comes from redesigning how value is created and captured—through platforms, communities, and data—rather than from traditional product innovation alone. Companies outside Asia can adopt these architectures by building ecosystems, leveraging creator networks, embedding AI deeply into customer experiences, and designing business models around trust and personalization rather than transactions.

4. The Risks of Letting AI Direct Conversations (HBR)
AI systems ask questions differently from humans, and those differences silently shape decisions by steering what information gets explored. Compared with executives, LLMs tend to overemphasize interpretation and analysis while under-asking execution questions about resources, timing, and implementation, creating potential blind spots in decision processes. The risk is subtle: leaders may believe AI is helping them think better when in reality it is biasing the structure of the conversation itself.

Innovation Radar

1. AI Model Releases and Advancements