The shift from AI as a single-query tool to AI as an autonomous agent capable of completing multi-step tasks is the most significant escalation in automation capability since large language models became widely available. When someone asks ChatGPT a question and gets an answer, one task was automated. When an agentic AI system is given a goal and independently plans, executes, checks, and iterates across a series of steps to achieve it, entire workflows are automated. This distinction matters enormously for understanding which jobs are exposed and on what timeline.
Agentic AI systems like Claude with tool use, OpenAI's Operator, and Google's Project Mariner are moving from research demonstrations to production deployment in 2026. The jobs most immediately affected are those where the primary value was coordinating a series of individually routine steps: research coordination, data pipeline management, customer service workflows, and multi-step analysis tasks.
What the research says
92M
jobs displaced by 2030
WEF Future of Jobs 2025
170M
new roles created by 2030
WEF Future of Jobs 2025
41%
of employers plan AI-driven headcount reductions
WEF 2025
55K
job cuts explicitly attributed to AI in 2024
Challenger, Gray and Christmas
Agentic AI refers to AI systems that can take sequences of actions autonomously to complete complex tasks, rather than just answering a single question. An agentic system can browse the web, write code, call external APIs, create files, and check its own work across multiple steps. Examples include Claude with tool use, OpenAI's Operator, and Google's Project Mariner.
Regular AI tools respond to a single query and return a result. Agentic AI can be given a goal and will plan and execute the steps to achieve it independently. This means AI can now handle workflows that previously required human oversight and coordination across multiple steps and systems.
Agentic AI has the most immediate impact on roles that involve coordinating multi-step processes: research analysts, project coordinators, executive assistants, data analysts, customer service workflows, and software developers working on well-defined tasks. Any role where the primary value is orchestrating a series of individually routine steps is exposed.
Yes, because it expands the automation surface beyond single tasks to entire workflows. A standard AI tool might automate writing a report summary. An agentic AI can gather the data, analyse it, write the summary, format it, and email it to the right people without human intervention. Full workflows become automatable.
Focus on elements of your role that require genuine human judgment, stakeholder relationships, or accountability that cannot be delegated to a system. Workers hardest to replace by agentic AI are those who own the decisions, not just the execution of steps. Understanding what agentic AI can and cannot do helps you position yourself to direct these systems.
Free Assessment
The Fossil Score assessment includes agentic AI exposure. Free, 4 minutes.
Get My Fossil Score