Generative AI Is Eating RPA — Here's Who Wins and Loses

Why this matters now

RPA vendors sold the promise of automating repetitive work. What arrived was reliable template-matching: bots that cut labor costs but demanded constant upkeep. Now generative AI can read messy text, reason about context and draft prose. That difference matters — it makes automation adaptable instead of brittle, and procurement teams at large firms are already changing where they spend. Vendors who thought they were finished selling the same product are having to rethink fast.

The short story

• RPA handled structure; generative models handle context. Put them together and automation starts to act like a junior analyst rather than a fixed script.
• Major platform players, notably Microsoft, are baking copilot-style AI into low-code tools so business teams can assemble smarter automations with less IT gatekeeping.
• Hardware providers such as Nvidia capture a lot of the economic upside because enterprises need significant compute to fine-tune and run these models.

What's interesting is how these three forces interact: smarter software reduces the need for brittle exceptions, and heavy compute creates a new recurring revenue channel.

Concrete examples

• Invoice processing used to be about rigid templates and long exception lists. Now an LLM can pull fields from odd layouts, call out anomalies and even draft a follow-up message — which reduces the number of manual exceptions.
• Customer support bots no longer only match FAQs. They can draft personalized answers and summarize the case for a human agent, cutting handle time and improving first-contact resolution. Not perfect in every scenario, but meaningfully better.

Market implications — winners and losers

• Winners will be platforms that combine orchestration, governance and models — think large cloud providers and GPU makers. They capture recurring cloud spend and higher-margin services.
• Losers are niche vendors that sell brittle, rule-heavy bots and have no credible plan to integrate AI. Their customers will either upgrade or consolidate; some vendors will pivot, others will fade.

Expect uneven outcomes. Some incumbents will adapt; others will be picked off or acquired.

Jobs and skills — not just layoffs

Yes, repetitive jobs will be displaced. At the same time demand will rise for:

• Automation architects who design hybrid AI+RPA flows
• Data and model ops specialists who monitor drift and compliance
• Business analysts who turn legal and policy nuance into test cases

These are higher-skill, higher-pay roles. The shift will be messy and sector-dependent.

What to watch next

• Signals: growing enterprise spend on AI-augmented low-code tools and lower bot churn.
• Deals: big cloud vendors partnering with or buying RPA firms.
• Rules: privacy and model-transparency regulations that could slow deployments in finance and healthcare.

Quick take for investors

• Favor companies showing rising AI services revenue and improving customer retention. Volume alone won't move the needle; margins do.
• GPU and inference-engine suppliers are a leveraged way to play enterprise AI adoption.
• Industrial automation companies that adopt AI to reduce downtime and optimize processes can benefit even if they aren't pure software businesses.

A human note

The debate around automation tends to polarize — doom or utopia. Reality sits somewhere in between. Companies will get more efficient, some roles will disappear, and new, often better-paid jobs will appear. The winners will be the firms that treat automation as a product to be maintained and improved, not a one-off project.

If you listen to this quarter's earnings calls, pay attention to three words: AI-enabled, low-code, stickiness. Those mentions will hint at which vendors are turning RPA from a cost-cutting exercise into a competitive capability.