The BIG productivity gap AI trained staff vs untrained staff AI agents are no longer a demo in someone’s slide deck – they’re quietly becoming your team’s most relentless new colleagues. And that’s exactly why AI literacy just went from “nice to have” to “management responsibility.”
Over the past year I’ve seen a clear shift with clients: we’re not just talking about chatbots that help you draft an email. We’re talking about agents that understand goals, plan steps, call tools, and take actions across systems – updating your CRM, chasing tasks, assembling reports and even triggering workflows while you sleep.
In other words, work is being reorganised around AI, not just sprinkled with it. From copilots to colleagues A couple of years ago, “AI at work” meant opening a chat window, pasting some text in, and hoping for the best. Today, many teams are working with: • Agents embedded directly into their project tools and CRMs • Domain specific “skills” (e.g. extract action items, build a status report, triage a ticket) • Workflows where humans set the direction and agents handle a big chunk of the execution
If you’re a project manager, that might look like starting Monday with an automatically generated status pack: key risks flagged, overdue actions highlighted, draft emails ready for stakeholders, and a suggested agenda for your stand up. Nobody got “hired” to do that work. But it’s happening anyway.
The big leap: productivity and management information This is the biggest leap I’m seeing in the workplace right now: agents and skills quietly taking ownership of the boring but important glue work. Things like: • Pulling data from multiple tools to assemble management information • Writing first drafts of board or steering group papers • Chasing updates from owners and updating trackers • Preparing frontline agents with suggested responses and next best actions
For leaders, this is both a gift and a risk. On the one hand, your reports arrive earlier, richer, and more customised. On the other, they are increasingly written by systems that your people don’t fully understand, can’t properly interrogate, and may be too busy (or anxious) to challenge. If we’re honest, many organisations are already “managing by AI output” without admitting it. The uncomfortable truth: our people aren’t ready
Here’s the uncomfortable bit: the technology has sprinted ahead faster than our skills, norms and governance. Most employees were never trained to: • Work productively with agents rather than tools • Decide what to delegate and what to keep • Check the reasoning behind AI generated recommendations • Know when to override, escalate or simply say “this isn’t good enough”
Instead, they’re told some version of: “Here’s a powerful AI feature, go experiment.” That might be fine when we’re talking about drafting an email. It’s not fine when agents are shaping decisions, money flows and people’s careers. This is where AI literacy comes in – and I don’t mean “prompt tips” and a list of clever hacks. What AI literacy really needs to mean now
When I talk about AI literacy with clients, I frame it as the capability to work alongside AI colleagues safely and effectively. For me, that includes: • Understanding: A practical grasp of what today’s agents and skills can and can’t do – in plain language, not maths. • Delegating and supervising: Knowing how to brief an agent, iterate on its work, and set clear boundaries of responsibility. • Judgement: Being able to interrogate outputs, spot weak logic or bias, and bring domain expertise back into the loop. • Governance: Knowing what is acceptable in your organisation – data, compliance, regulatory expectations – and behaving accordingly.
If you wouldn’t let a new human hire make certain decisions without training and oversight, you shouldn’t let an agent do it either. The difference is that your new AI “hire” can be rolled out to thousands of people at once.
Why this is now a leadership issue, not an IT issue A lot of organisations still treat this as a technology conversation: platforms, integrations, vendors. But AI agents change the nature of work itself. They affect: • How information flows to leadership • What managers spend their time on • How frontline employees make daily decisions • How accountable you are for AI shaped outcomes
That makes AI literacy a leadership and culture topic. Leaders need to be able to ask better questions of AI generated reports. Managers need to know how to run teams where human and agent contributions are blended. Employees need confidence that they’re not going to get punished for using (or not using) AI in the “wrong” way. Simply rolling out more AI features without investing in AI literacy is like buying a fleet of high performance cars for people who’ve only ever ridden bicycles – and then wondering why the accident rate is creeping up. Where I see organisations starting well
The organisations that look most “agent ready” to me are doing three things:
- Designing AI first workflows They don’t just bolt AI onto existing processes. They redesign workflows around what agents can reliably own, and where humans genuinely add value.
- Investing in role based AI literacy They recognise that leaders, managers and frontline staff need different skills. They train people in realistic scenarios with their real tools, not in abstract workshops.
- Putting clear guardrails in place They pair enthusiasm with boundaries: guidance on data, transparency about how AI is being used, routes to escalate concerns, and alignment with emerging regulation.
The net effect is that people feel safer experimenting, and value shows up much faster. Why I care about this (and what we do at AI Advisers) As founder of AI Advisers, I spend a lot of my time with leaders who feel caught between two pressures: • “We need to move fast on AI – our competitors are.” • “We can’t afford to break things, scare our people, or fall foul of regulators.”
My strong view is that AI literacy is the bridge between those two. It’s how you unlock the upside of agents and skills without turning your organisation into a black box experiment.
We help organisations: • Identify high value, realistic agent use cases • Design workflows where agents and humans complement each other • Build practical AI literacy programmes that give people the confidence and judgement to use AI well
If you’re experimenting with agents, or you can see them coming into your core tools and you’re not sure your teams are ready, I’d love to chat. Do your people know how to brief, supervise and challenge their new AI colleagues – or are you hoping the technology will quietly teach them? Contact www.aiadvisers.co.uk