Bank AI job postings climb 49% to 139,819 as agent orchestration grows fastest

Work the 49% backward and Draup's 2025 count comes to about 93,800 bank AI postings, so the year added roughly 46,000 listings ^[1][17][18]. Draup builds the count from public job ads and from platforms such as LinkedIn ^[9]. A job ad shows that a bank intends to hire. The hire, the start date and the payroll cost come later, if they come at all.

The orchestration figure means the skill appeared about 18.2 times as often this year as last ^[2][19]. The CNBC report does not include the base count, so the fastest growth rate in the study could rest on one of its smaller numbers. The counts that were published favour control work. Governance-related skills drew more than 16,000 references, against roughly 8,400 for training, deploying and running models, or about 1.9 for every one ^[8][20].

That split shows which staff banks are trying to hire. The earlier wave of bank AI hiring was mostly engineers and data scientists who built models or adapted them to company data. The current wave adds people who embed AI inside business lines ^[11]. A deployment typically strings together specialised agents: one to inspect raw data, another to analyse a document and a third to check regulatory compliance ^[3]. According to Swaminathan, the forward-deployed engineer decides which agents are needed, what each one does, which technology to use and when a human overseer steps in ^[4]. He said even automating staff vacation approvals turns up a web of edge cases and exemptions ^[5]. "There is a lot of complexity in an enterprise," Swaminathan said. "Sometimes these complexities are visible, but many times they are hidden. It takes a long time even to automate a simple process." ^[6]

Security teams are working to stop third-party tools and external model connections from creating systemic vulnerabilities ^[14]. "There is a lot of focus on making sure that the third parties that we are using in these products are not going rogue from a cybersecurity standpoint," Swaminathan said ^[13].

CNBC takes the listings to show banks moving past chatbots toward agents that handle a growing share of the work ^[10]. The postings could play out in three ways. In the first, banks turn them into agents that take over repetitive back-office work, and the pay premium for generative AI roles ^[15] is recovered through lower headcount elsewhere. In the second, recruiters relabel existing integration jobs with the new vocabulary, and the percentages measure word choice more than spending. In the third, control work keeps growing faster than build work, and agents stay under close human supervision for longer than the pace of hiring suggests.

I think the first and third are both happening. On the counts that were published, supervision is the larger hiring line. Swaminathan said the specialised roles remain hard to fill ^[16], so banks pay the premium now and collect any labour saving later. That view is wrong if the count behind the orchestration figure turns out to be larger than the 16,000 governance references ^[8]. In that case most of the new payroll is going to building agent teams, and less of it to watching them.