# AI Limbo: Your Team Is Ready for AI Agents. Are Managers? > Workers are eager for AI agents, but few have been trained and managers are the missing link. What AI limbo costs Indian teams and how to fix it. - Source: https://bodhih.com/ai-limbo-managers-leading-ai-agents-india-teams-training/ - Publisher: Bodhih Training (Bengaluru, India; since 2008) - Last updated: 2026-10-13 - Contact: solutions@bodhih.com · +91 99000 11601 --- Priya runs a 14-person operations team at a Pune logistics firm. In March, her company rolled out an AI agent that drafts shipment-exception reports and chases carriers for updates. Six months later, three of her people use it every day, five tried it once and quit, and six have never opened it. Priya herself is not sure which group she should be in. Workers are eager for AI agents, but few have been trained and managers are the missing link. What AI limbo costs Indian teams and how to fix it. Key takeaways - HR professionals are even more candid than workers: only 18% say formal AI agent training has been delivered, and they believe 42% of employees are minimally prepared or not prepared at all. - The scarce resource is no longer access to AI. It is managers who know how to delegate to it, review it and develop people alongside it. - These are vendor and consultancy surveys, mostly from the US and UK. Treat the numbers as directional and test them against your own teams. Last updated 13 Oct 2026 · Bodhih Insights team Nobody has told her how to assign work to an agent, how to check its output, or what to do when it quietly gets something wrong. Her review template has no line for any of this. Her own manager asked last week, "How is the AI rollout going?" and she said, "Fine." Researchers have a name for where Priya's team is stuck: **AI limbo**. The tools have arrived and people are willing, but the training, rules and management habits that turn a tool into results have not. ## What "AI Limbo" Actually Looks Like AI limbo is not resistance. In the People Managing People research, workers were generally positive about AI. An EY survey from October 2025 similarly found 84% of employees eager to embrace agentic AI, even as 56% worried about their own job security. Limbo is what happens next. People want to use the tools. Leaders have announced them. But in between sits a layer nobody has equipped: the team manager. You can usually spot limbo by a few symptoms. - **Uneven use.** A few enthusiasts race ahead while the rest quietly opt out, and the gap becomes a status difference inside the team. - **Improvised standards.** Everyone has a different view of what "good enough" AI output means. - **Silent verification.** Nobody knows who checks an agent's work, so either everyone does (slow) or nobody does (risky). - **Stigma.** A SnapLogic report found 43% of employees believe using AI tools is seen as lazy or untrustworthy, so people hide their use. That is the same dynamic we described in our post on [shadow AI](https://bodhih.com/shadow-ai-employees-using-unapproved-ai-tools-india-training/). Many of these effects add up to the pattern we called [workslop](https://bodhih.com/workslop-ai-generated-work-india-teams-training/) and the fatigue we covered under [AI brain fry](https://bodhih.com/ai-brain-fry-mental-fatigue-ai-tools-india-teams-training/). Limbo is the management-side cause behind both. ## The Twist: Managers May Trust the Agent More Than the Intern Here is the part that deserves more attention than it is getting. The same SnapLogic research reported that many managers say they trust AI agents more than junior employees or interns. It is an understandable instinct. An agent is fast, never tired, and does not need a feedback conversation. But follow the thought through. If managers hand the routine, learning-rich tasks to agents, juniors lose the very reps that built the managers' own judgement. A few years on, the organisation has fewer people who can tell good output from confident-sounding output. This matters for India in particular. Entry-level and early-career roles are where much of our talent pipeline is built, and the [Great Flattening](https://bodhih.com/great-flattening-shrinking-management-layers-indian-managers/) is already giving managers wider spans to cover. Wide spans plus agents plus no feedback habits is how a team's learning quietly stops. ## Managing an Agent Is a Real Skill, and It Looks Familiar The encouraging news: leading human-and-agent work leans on skills good managers already have. They just have to be used deliberately. **Delegating clearly.** An agent, like a new joiner, does what you ask, not what you meant. Vague instructions yield plausible, wrong work. Writing a crisp brief with the goal, constraints, data and the definition of done is a management skill before it is a prompting skill. **Reviewing with a sceptical eye.** Managers need a habit of spot-checking: pick three claims, trace them to source. Fluent output is not verified output. **Deciding who owns the outcome.** The rule is simple and easy to forget: the human who sends it owns it. Say so explicitly. **Coaching, not just assigning.** The 56% figure on weak manager coaching is the quiet risk here. If managers cannot coach, they cannot help the person who tried the agent once and gave up. Building this habit is what [Coaching for Managers](https://bodhih.com/coaching-for-managers/) and [Giving and Receiving Feedback](https://bodhih.com/giving-and-receiving-feedback/) are designed for. **Protecting the learning loop.** Decide deliberately which tasks juniors still do by hand, because that is how they build judgement. ## A Four-Question Check for Any Manager This Week You do not need a programme to start. Ask yourself these four questions about your own team. 1. **Who uses AI regularly, who dabbled, who never started?** If you do not know, that is your first finding. 2. **For each AI-assisted task, who checks the output, and how?** If the answer is "whoever notices," write down a real step. 3. **What does a good brief look like here?** Pick one recurring task and write the best instruction you can. Share it. 4. **Which tasks should a junior still learn without AI?** Name two, and protect them. Then spend 30 minutes at your next team meeting asking: "Where has AI saved us time, and where has it cost us time?" The honest answers will be more useful than any rollout dashboard. ## How Organisations Can Close the Gap Closing limbo is mostly about sequence and people, not software. **Equip managers before you scale the tools.** The fastest improvements tend to come from training the people who set daily expectations. [Leading Human + AI Teams Training](https://bodhih.com/leading-human-ai-teams-training/) focuses on exactly this: delegating to agents, setting review standards, redesigning workflows and keeping people's skills growing. **Build a shared base of AI skills.** When everyone has the same vocabulary and verification habits, the uneven-use problem shrinks. [AI Literacy Training](https://bodhih.com/ai-literacy-training/) builds that foundation through hands-on practice on real work. **Give senior leaders a clear point of view.** Teams read signals from the top about whether using AI is safe, expected or suspect. [AI for Leaders](https://bodhih.com/ai-for-leaders-training/) helps senior people frame strategy, governance and risk, and [Change Management Training](https://bodhih.com/change-management-training/) helps carry a team through the change instead of announcing it. **Strengthen the human judgement underneath.** [Critical Thinking Training](https://bodhih.com/critical-thinking-training/) trains the habit of questioning a smooth answer, which no tool can supply for you. ## The Bigger Idea Most AI adoption stories are told from the tool's point of view: which model, which agent, which licence. The research points the other way. Employees are largely ready. The constraint is a thin middle layer of managers who were never asked to learn this and are being measured on a rollout they cannot see inside. Organisations that treat the manager as the product, not the afterthought, will move out of limbo first. Priya, for what it is worth, has now written her first proper brief for the agent. Her team's exception reports are already getting shorter. ## Frequently asked questions ### What is AI limbo? AI limbo is the state where employees are willing to use AI agents and organisations have rolled them out, but the training, governance and management practices needed to use them well are missing. The term comes from a People Managing People report based on surveys of workers and HR professionals in December 2025. ### How many employees have been trained to work with AI agents? In the People Managing People research, 36% of workers said their employer had provided training on working with AI agents, 44% said they had received none and 20% said training was in progress. Only 18% of HR professionals said formal training had been delivered. These are surveys, so figures will differ by company and country. ### Why are managers the key to AI adoption? Managers set daily expectations: what gets delegated, how output is checked and whether it is safe to admit uncertainty. LifeLabs Learning's 2026 Mid-Year Workplace Trends Report found 56% of respondents rated their managers' coaching and feedback skills at 5 out of 10 or lower, which suggests many managers need support before AI rollouts can succeed. ### Should managers trust AI agents more than junior employees? Agents are fast and consistent, but they can be confidently wrong, and they do not learn the way a developing employee does. Handing all routine work to agents can also remove the practice juniors need to build judgement. A better approach is to assign work deliberately, verify agent output, and protect some tasks as learning opportunities. ### What skills do managers need to lead teams that use AI agents? Clear [delegation](https://bodhih.com/effective-delegation-techniques/), sceptical review of output, ownership of outcomes, coaching and feedback, and workflow design. Bodhih's [Leading Human + AI Teams Training](https://bodhih.com/leading-human-ai-teams-training/) covers these skills for corporate batches and can be customised to your tools and roles. ### Is this a problem in India specifically? Indian desk workers are among the fast adopters of AI. Slack's Workforce Index (December 2024, 1,029 Indian desk workers) found 61% already using AI at work, yet 40% had spent under five hours learning it. Most of the AI-agent training research cited here is from the US and UK, so Indian organisations should measure their own teams' readiness. ### Where do the statistics in this article come from? People Managing People's "AI Limbo" report (December 2025 surveys of 1,000 workers and 379 HR professionals), LifeLabs Learning's 2026 Mid-Year Workplace Trends Report, an EY survey (October 2025), a SnapLogic report on managers and AI agents, and Slack's Workforce Index (December 2024). All are self-reported surveys, many by vendors, and should be read as directional.