# Finance teams remain wary of letting AI make routine decisions, PEX survey finds

> Only 28% of finance teams trust AI to make routine decisions, PEX's survey of 687 leaders found. Interest runs at 66%, but accuracy fears and system gaps keep adoption at 31%.

- Type: AI news
- Page: https://completeaitraining.com/news/finance-teams-remain-wary-of-letting-ai-make-routine/
- Published: 2026-09-12
- Updated: 2026-09-12
- Author: Jeroen Erne
- Jobs: Finance
- Reading time: 2 min

Only 28% of finance teams are comfortable letting AI make routine finance decisions, according to a survey by spend management platform PEX, signaling a trust gap that blocks wider automation even as two-thirds of leaders express interest in the technology.

The survey of 687 finance and operations leaders found that while 31% already use AI today, a much larger 66% are interested in doing so. The 35-percentage-point gap between interest and actual use comes down to a single issue, said PEX CEO and founder Toffer Grant: "Trust."

Finance teams want to maintain control over financial processes. Integrating automation into those workflows creates a "level of tension" that individuals need to work through, Grant said in an interview. He pointed to the fear of errors on high-stakes tasks. "I think a lot of the folks who are kind of on the early end of their adoption are not going to let a tool like PEX just pay a $15,000 invoice," he said. "You have the vendor relationship to manage. You need to know that the system isn't going to pull $150,000 instead of $15,000."

#### Trust in accuracy remains the top barrier

Trust in AI's output accuracy was the top barrier to implementation, cited by 36% of respondents. That wariness widens a "want-versus-have" gap across seven AI capabilities the survey tracked, including cash flow forecasting, audit documentation automation, and AI-generated financial reports.

For AI-generated financial reports, 61% of respondents expressed interest, but only 14% said they currently use the technology - a 47-percentage-point gap. The second-largest barrier was a lack of interoperability with existing systems, cited by 20% of respondents.

Grant pointed to "easy wins" that can build the foundation of trust. Using AI to wade through a 20-page credit card statement or simplify receipt adjudication frees up time and builds comfort with routine automation. Rather than an accountant chasing down an employee for a missing receipt, the tool handles it. "Now, what somebody is doing is getting very comfortable on relying on a platform to do the annoying work of being the enforcer, and also the annoying work of doing the chase down," Grant said.

#### Inertia and the crawl-walk-run approach

Beyond trust, simple inertia slows adoption. Many companies still rely on manual expense reporting - employees taping receipts to paper and submitting them physically. The survey recommends a "crawl, walk, run" framework to break that inertia. Companies at the earliest stage should focus on automating one low-effort capability. The report noted that 77% of "crawlers" run nothing at all, so a single first step is where progress begins.

The payoff is measurable. Of the 340 respondents piloting or using AI, 69% said they reduced time spent on manual review, and 51% reported cutting their close time. For finance leaders exploring [AI for Finance](https://completeaitraining.com/tag/finance/), these early wins can demonstrate value without handing over final decision authority on large transactions. Those looking to build a strategic foundation can explore structured learning paths like [AI for CFOs](https://completeaitraining.com/learningpath/ai-for-cfos-chief-financial-officers/), which covers forecasting and automation strategy for finance departments.

#### Why this matters for finance professionals

The survey makes clear that AI adoption in finance is not blocked by a lack of tools or interest, but by a deliberate caution around accuracy and control. For finance teams, the practical path forward is not wholesale automation of decisions, but targeting repetitive, low-risk tasks - receipt tracking, statement reconciliation, policy enforcement - where errors are easily caught and trust can accumulate. The 69% of early adopters who cut manual review time show that even narrow automation delivers concrete efficiency gains without ceding control over material financial decisions.

---

This is the Markdown form of https://completeaitraining.com/news/finance-teams-remain-wary-of-letting-ai-make-routine/. The whole site is mapped at https://completeaitraining.com/llms.txt.