Key Takeaway

The arrival of GPT‑6 Astra makes the debate around artificial general intelligence less abstract. The practical question for organizations is no longer only whether AI can reason at a high level, but what happens when it can also operate computers, select tools, execute multi-step workflows and absorb part of the coordination work traditionally performed by managers. International companies are already reducing layers of small-team management. For Georgia, this does not imply an automatic collapse of middle management. It does imply a new test: which management roles create real decision value, and which mainly transmit information, monitor status and coordinate routine work?

The practical shift matters more than the AGI label

A recent Wall Street Journal CIO Journal analysis linked GPT‑6 Astra to the possible beginning of the AGI era. OpenAI President Greg Brockman noted that there is no single accepted definition of AGI and that the boundary is inherently fuzzy. For businesses, the more useful question is therefore capability, not terminology.

OpenAI describes Astra as a major advance in computer use, browsing, software engineering and professional work. It can fill forms, update CRM records, organize calendars, conduct research, prepare documents, install and test software, analyze data and execute other multi-step workflows. The distinction is important: a chatbot produces an answer, while an agentic system increasingly carries out the work required to reach the answer.

That has organizational consequences. If status collection, reminders, reporting, routine monitoring and workflow coordination can be partly automated, the amount of human managerial time required for coordination may fall. A technology shift therefore becomes an organizational-design question.

Why large companies are cutting micro-team management

According to the Wall Street Journal, Uber plans to nearly halve the number of “micro-teams” with only one or two direct reports. Its broader restructuring includes about 3,300 job cuts and an eventual reduction of roughly 20% in managers. Google previously cut the number of managers overseeing small teams by 35%, while Axon Enterprise eliminated many small-team manager roles and moved some supervisors back into individual-contributor positions.

These restructurings should not be attributed to AI alone. Companies were flattening hierarchies for cost, speed and accountability reasons before the current generation of AI. But AI can reinforce the trend by automating parts of coordination and monitoring, potentially increasing a manager’s span of control.

Gartner predicted in 2024 that by 2026, 20% of organizations would use AI to flatten their organizational structures, eliminating more than half of current middle-management positions. Gartner also warned of the trade-offs: manager overload, employee anxiety, resistance to change, and weaker mentoring and development pathways for junior workers. A separate 2026 Gartner survey found that 95% of organizations had implemented AI in some capacity, but only one in five had achieved significant or transformational value. Adoption alone is therefore not organizational redesign.

Georgia must ask a different question

Georgia’s business structure is very different from that of large U.S. technology companies. According to Georgia’s National Statistics Office, Geostat, there were 280,811 active economic entities as of July 1, 2026. Of these, 192,717 were individual entrepreneurs and 80,153 were limited liability companies. These figures do not measure company size directly, but they illustrate how fragmented Georgia’s business landscape is. Many firms are already too small to have several layers of management.

For Georgia, AI-driven flattening is therefore primarily relevant to larger and mid-sized organizations: banks, insurers, telecom companies, retail networks, developers, large service firms, universities, public institutions and rapidly growing technology companies. In a small business, the first effect may be different: not removing a manager, but reducing the administrative burden on the owner or director.

What AI can absorb – and what should remain human

BTU’s research and analytical framework for agentic management emphasizes a critical distinction: technical capability is not the same as permission to act. An AI system may be capable of completing a task, but the organization still has to define mandate, decision rights, responsibility, controls and stop conditions. Fewer operational managers should not mean weaker governance. In many cases, it requires the opposite: clearer rules about who – human or agent – is allowed to decide what.

The right metric is not layers, but coordination value

A management layer is neither automatically useful nor automatically bureaucratic. The real question is whether it contributes decision quality, accountability, knowledge transfer or people development in a way that another mechanism cannot replace.

If a manager spends most of the day collecting status updates, scheduling meetings, forwarding instructions, updating spreadsheets and compiling reports, an agentic system can potentially reduce a large share of that load. If the manager makes difficult judgment calls, develops people, resolves conflicts, protects quality and assumes responsibility under uncertainty, that role may become more valuable in the AI era, not less.

The manager of the future may therefore become less of an information relay and more of a decision architect. The team may be larger, but the manager will increasingly work alongside AI agents that handle specific processes, monitoring and information flows.

Three organizational models Georgia may see

What companies should not do

The most dangerous reaction would be to copy foreign restructuring examples and cut managers before an AI-enabled operating model has actually been built. Gartner’s 2026 findings show that broad adoption does not automatically translate into significant value. If data are poor, responsibilities are unclear, workflows are undocumented and employees do not trust the system, removing a management layer may create an organizational vacuum rather than efficiency.

A second mistake would be to ignore the developmental role of managers. A small-team manager often trains new employees, gives feedback, prepares them for future roles and protects team culture. If the layer disappears, the organization needs a replacement mechanism for mentoring and professional development.

A third mistake is granting AI excessive autonomy simply because it is technically capable of acting. Organizational agents need explicit mandates, real approvals, escalation rules and stop mechanisms.

What Georgian companies should do now

• Measure what each management layer actually does: decision-making, coordination, control, people development or simple information transfer.
• Identify low-risk work that AI agents can absorb: status collection, reporting, scheduling, documentation, monitoring and routine follow-up.
• Do not remove a management layer until the new AI-enabled workflow has proved itself in a real pilot.
• Define an AI agent’s mandate, decision rights, data access, escalation and stop rules before deployment.
• Shift manager KPIs away from administrative activity toward decision quality, team development, outcomes and risk management.

BTU Researchers’ Assessment

According to an assessment by BTU researchers, Georgia’s near-term organizational shift is unlikely to be the mass disappearance of middle management. A more realistic transition is the redistribution of work inside management itself: routine coordination moves to agentic systems, while human roles concentrate on judgment, accountability, people development and exception handling. Where a management layer mainly transfers information, its need may genuinely decline. Where it creates trust, professional judgment and accountability, AI may make the role more important.

Conclusion

The exact date on which the AGI era begins may remain debatable for years. For businesses, that is not the key issue. What matters is that systems such as Astra are moving from answers toward action: using computers, applications and tools and executing increasingly complex multi-step work.

The question for Georgian companies is therefore not “How many managers can we remove?” but “Which management functions still require a human, and which can become software-based coordination?” The answer to that question will determine whether AI simplifies organizations or merely adds another technology layer on top of old bureaucracy.

Data and Main Sources

The Wall Street Journal, CIO Journal – “Yes, We’re Entering the Era of Artificial General Intelligence” and “Corporate America Is Axing the ‘Micro-Team’ Boss,” September 2026; OpenAI – GPT‑6 Astra: A new generation of intelligence, 2026; Gartner – Top Predictions for IT Organizations and Users in 2025 and Beyond; Gartner – AI automation and entry-level hiring survey, 2026; National Statistics Office of Georgia – active economic entities as of July 1, 2026.