OpenAI’s finance team has demonstrated how artificial intelligence coding tools can transform traditionally manual processes, cutting reporting timelines from days to hours while sparking broader questions about the future role of automation in corporate finance.

The finance function has long been characterized by meticulous manual work-reconciling data, preparing reports, and ensuring accuracy across countless spreadsheets. But a recent experiment at OpenAI suggests that the landscape is shifting dramatically. When Kyle Kober, director of product finance at OpenAI, embarked on a mission to automate tedious workflows within his department, he found himself in unfamiliar territory: writing code. Using Codex, an AI coding agent developed by his own company, Kober and his colleagues managed to transform one of their most cumbersome monthly-close reporting processes, reducing the time required from approximately five days to just five hours.

This transformation represents more than just an isolated efficiency gain. It signals an emerging trend in corporate finance where AI-powered coding tools are empowering finance professionals to build custom solutions without relying on IT departments or software engineers. As finance teams across industries grapple with increasing data volumes and pressure to deliver real-time insights, the ability to automate complex workflows through AI coding could redefine what’s possible for the profession.

The specific challenge Kober’s team tackled centered on OpenAI’s computing capacity costs-a particularly labor-intensive area that required reconciling product-usage data with accounting records before transforming that information into meaningful analysis and reporting. This process exemplifies the kind of repetitive, detail-oriented work that consumes significant time in finance departments worldwide.

Before implementing AI coding solutions, the team spent roughly five days each month working through this reconciliation and analysis process. The work involved cross-referencing multiple data sources, identifying discrepancies, and preparing reports that would inform strategic decisions about resource allocation and capacity planning. While essential, the process left little time for higher-value activities like strategic analysis or forecasting.

The adoption of Codex changed this equation fundamentally. According to Kober, the AI coding agent caught on quickly among OpenAI’s engineering teams in late 2024, and finance professionals followed suit just a few months later. “I think most engineers had their Codex moment in like November or December of last year; it was only a couple of months later for us,” Kober explained in an interview with CFO Dive.

By leveraging Codex to automate significant portions of the workflow, the finance team achieved a dramatic reduction in processing time-from five days to approximately five hours. This tenfold improvement freed up resources for more strategic activities and demonstrated the potential for AI coding to transform finance operations.

Kober’s automation project fits within a broader strategic initiative at OpenAI. In a blog post published in early 2025, OpenAI CFO Sarah Friar outlined her ambitious plan to build an “AI-native finance function.” This vision encompasses multiple transformative goals, including moving toward a “zero-day close”-where financial results are available immediately at period end-and implementing automated, continuously updated forecasting capabilities.

These objectives represent a significant departure from traditional finance operations, which typically operate on monthly or quarterly cycles with substantial time lags between period end and reporting. The AI-native approach envisions a finance function that operates in real-time, providing decision-makers with current information rather than historical snapshots.

Kober played a central role in this transformation, working to rebuild manual processes that had become bottlenecks in the organization’s financial operations. The computing-related workflow was particularly challenging due to the volume and complexity of data involved, making it an ideal candidate for automation through AI coding.

The broader push reflects OpenAI’s recognition that its finance function must evolve to match the pace of innovation in its core business. As the company scales its AI infrastructure and product offerings-including a recent announcement of $110 billion in new investment-the finance team requires tools and processes capable of keeping pace with rapid growth and change.

While AI-driven coding has rapidly gained adoption among software engineers, its expansion into finance represents a significant development. Research firm Gartner has predicted that by 2028, 90 percent of enterprise software engineers will use AI code assistants, up dramatically from less than 14 percent in early 2024. This explosive growth trajectory reflects the technology’s ability to accelerate development processes and reduce the technical barriers to creating custom applications.

For finance teams, this democratization of coding capabilities opens new possibilities. A June 2025 report from Boston Consulting Group highlighted how AI coding agents can enable finance professionals to build targeted applications for tasks such as analysis, matching, and anomaly detection “without waiting for long development queues.” This independence from IT departments represents a fundamental shift in how finance teams can respond to emerging needs and opportunities.

The trend aligns with broader patterns in AI adoption across finance functions. According to research from McKinsey, 44 percent of finance professionals reported using generative AI for over five use cases in 2025, up from just 7 percent the previous year. This rapid acceleration indicates that finance teams are increasingly comfortable experimenting with AI technologies and integrating them into core workflows.

Moreover, the finance industry’s reliance on data-intensive processes makes it particularly well-suited for AI applications. As IBM research notes, AI tools can process large volumes of data quickly and accurately, enabling finance teams to address challenges that would be impractical to tackle manually. The combination of AI coding with other AI capabilities creates a powerful toolkit for transforming finance operations.

While the productivity gains from AI coding are compelling, they come with important economic considerations. Gartner has warned that AI coding expenses are rising as token consumption increases and vendors shift toward consumption-based pricing models. Tokens represent the basic units of data processed by AI systems, with greater token consumption generally resulting in higher costs.

The research firm has made a striking prediction: by 2028, AI coding costs could surpass the average developer’s salary. This forecast suggests that while AI coding tools can dramatically improve productivity, organizations must carefully manage their usage to avoid runaway expenses. The shift from subscription-based pricing to consumption-based models means that costs can scale unpredictably if usage is not monitored and optimized.

For finance teams adopting AI coding, this creates an interesting paradox. The tools promise to reduce the time and resources required for various tasks, but without proper governance and oversight, the cost of the tools themselves could offset those gains. Organizations will need to establish clear policies around AI coding usage, monitor consumption patterns, and ensure that the technology is deployed strategically rather than indiscriminately.

Despite these cost considerations, the overall trend toward AI adoption in finance continues to accelerate. Nearly 70 percent of financial executives are incorporating AI into their operations, with 85 percent of North American companies either having adopted or planning to adopt AI technologies soon, according to industry research. This widespread adoption suggests that organizations view the benefits as outweighing the costs, even as they work to optimize their AI investments.

Alongside economic considerations, the integration of AI-generated code into production systems raises important questions about security and risk management. A March 2025 paper published by the Cloud Security Alliance, a nonprofit dedicated to cybersecurity, noted that organizations are integrating AI-generated code at scale into production systems despite documented security risks.

These risks stem from several factors. AI coding tools may generate code that contains vulnerabilities or follows insecure patterns. The speed at which AI can produce code may also outpace traditional review processes, potentially allowing flawed code to enter production systems. For finance functions handling sensitive financial data and regulatory compliance requirements, these security considerations carry particular weight.

Boston Consulting Group has emphasized that finance leaders bringing AI coding into their teams need clear guardrails in areas such as auditability. The ability to trace how decisions were made, understand the logic behind automated processes, and demonstrate compliance with regulatory requirements remains essential, even as automation increases.

This need for oversight has led to collaborative approaches like the partnership between PwC and OpenAI announced in May 2026. This initiative aims to develop an “AI native finance function” that combines AI agents with human supervision, enabling finance teams to leverage AI capabilities while maintaining appropriate governance and control. The collaboration focuses on real-world applications and continuous improvement, ensuring that finance professionals can effectively adapt to new AI capabilities while managing associated risks.

Perhaps the most significant aspect of AI coding in finance is how it expands the capabilities available to finance professionals. Traditionally, finance teams requiring custom tools or applications had to submit requests to IT departments, join development queues, and wait for engineering resources to become available. This dependency created bottlenecks and limited finance teams’ ability to respond quickly to emerging needs.

AI coding tools change this dynamic fundamentally. As Kober noted, “There’s now the opportunity to just do it yourself.” Finance professionals with domain expertise can now translate their understanding of business processes and analytical needs directly into functional tools, without requiring deep programming knowledge or external technical support.

This self-sufficiency enables finance teams to become more agile and responsive. When a new reporting requirement emerges or an analytical opportunity presents itself, teams can quickly develop targeted solutions rather than working around limitations in existing systems or waiting for IT support. The ability to iterate rapidly on tools and workflows allows finance functions to experiment, learn, and optimize in ways that were previously impractical.

The shift also has implications for finance talent development and career paths. As AI tools handle more routine coding tasks, finance professionals can focus on the strategic and analytical aspects of their work-understanding business needs, designing effective solutions, and interpreting results. At the same time, some level of technical literacy becomes increasingly valuable, creating opportunities for finance professionals to develop hybrid skill sets that combine financial expertise with technological capabilities.

As AI coding continues to gain momentum in finance, organizations face the challenge of balancing aggressive transformation with prudent implementation. The potential benefits-dramatic efficiency gains, enhanced analytical capabilities, and greater agility-are substantial. However, realizing these benefits requires careful attention to governance, security, cost management, and change management.

Successful adoption appears to require several key elements. First, organizations need clear strategies for where and how to deploy AI coding, rather than pursuing automation for its own sake. The most effective implementations focus on high-value use cases where automation can free up significant time for strategic work, as demonstrated by OpenAI’s computing reconciliation project.

Second, appropriate guardrails and oversight mechanisms must be established. This includes processes for reviewing AI-generated code, monitoring security implications, ensuring auditability, and managing costs. The combination of AI capabilities with human supervision-as emphasized in the PwC-OpenAI collaboration-appears to be emerging as a best practice.

Third, organizations must invest in training and change management. Financial institutions are increasingly emphasizing upskilling and collaborative change management to overcome barriers to AI adoption such as integration complexities and data security concerns. Helping finance professionals develop the skills and confidence to work effectively with AI coding tools is essential for realizing their full potential.

The experience at OpenAI and similar pioneering organizations suggests that AI coding represents a genuine inflection point for finance functions. The technology offers the promise of transforming finance from a primarily retrospective, reporting-focused function to a real-time, strategically engaged partner in organizational decision-making. As more finance teams experiment with these capabilities and share their experiences, best practices will continue to evolve, helping organizations navigate the opportunities and challenges that AI coding presents.

The path forward will likely involve continued experimentation, learning, and refinement. Finance leaders who approach AI coding with both enthusiasm for its potential and rigor in its implementation will be best positioned to capture value while managing risks. As the technology matures and organizations gain experience, AI coding may well become as fundamental to finance operations as spreadsheets are today-a powerful tool that, when used thoughtfully, enables finance professionals to work at levels of speed and insight that were previously unimaginable.