Microsoft’s Copilot and a new wave of AI-native RPA tools are moving automation from IT backrooms onto every employee’s desktop. Winners will be platforms, validators, and the nimble — losers could be repeatable office tasks.
Illustration by IMF Alpha editorial · Reviewed by Pedro Marini
The headline is simple: automation is no longer a cottage industry for IT teams.
Today’s RPA doesn't feel like the brittle, scripted bots of a decade ago. It behaves more like a coworker who picks things up as it goes — sometimes impressively, sometimes imperfectly. Put large language models next to no-code automation — think Copilot-triggered flows in Power Automate or AI-aware connectors in modern RPA suites — and adoption inside finance, legal, HR, and sales starts to look different.
Why care? Because this changes who builds automations, how fast they spread, and who captures the value.
A short history to orient the change
It’s a bit like trading a fax machine for an email client that drafts, files, and files again for you. Different class of tool.
Real-world signs — concrete examples
These are not toy projects. They are repeatable playbooks for cutting cost and increasing capacity — though the human oversight piece usually stays front and center.
Where the money flows (and why investors care)
A finer point: pure-play RPA vendors face a choice. Some will pivot to AI-first value and stay relevant; others risk being relegated to a feature in a bigger suite and seeing margins compress. There’s room for both specialization and consolidation.
Risks and counterpoints — don’t buy the hype without a checklist
Governance will be where battles are won or lost. Teams that pair citizen development with central guardrails tend to capture benefits while avoiding the worst mistakes.
What CIOs, workers, and investors should do now
A human-centered reality check
The next wave of automation is less about mass layoffs and more about shifting the division of labor. Repetitive cognitive work gets cheaper; judgment, context and cross-domain synthesis grow scarce and valuable. That favors flexible teams, regulated industries that need traceability, and people who can combine domain expertise with product judgment. It also squeezes roles built on routine information wrangling.
So the real question for organizations is not whether AI takes tasks, but whether they will build the governance, skills and product strategy needed to capture the upside while limiting the downside.
Major AI projects are no longer starved for compute; they're starved for trustworthy, compliant data. Synthetic datasets are emerging as the fastest route to scale models and dodge regulatory landmines.
Firms are swapping raw tapes for engineered twins — cheaper, private, and faster. That changes who wins: cloud and GPU providers, data vendors, and the quants brave enough to trust simulations.
Chip advances, compact LLMs and privacy rules are pushing intelligence onto devices — what that means for apps, users and investors.