Data Agent ChatGPT Work: data analysis without coding for SMEs

Introduction: the bottleneck that slows down business decisions

In Italian SMEs, data analysis is often a slow and fragmented process. The data exists — in CRMs, Excel spreadsheets, e-commerce platforms — but turning it into insights takes time, technical skills, and almost always, the involvement of specialized personnel. Consequently, strategic decisions arrive too late to seize market opportunities.

Most small to medium-sized businesses struggle to leverage the advantage of data due to a lack of internal analytical resources.

The Data Agent in ChatGPT Work , officially announced by OpenAI in the official announcement of September 10, 2026 , aims to bridge this exact gap. Specifically, it lets you ask questions about your company data in plain English and get instant answers as charts, tables, and interactive dashboards.

How the Data Agent works: architecture and operating logic

The Data Agent integrates directly into the environment ChatGPT Work , the enterprise version of ChatGPT designed for business contexts. How it works is divided into three main steps.

• Connecting data sources: the agent supports integration with corporate databases, spreadsheets, CRMs and other platforms via native connectors or APIs. Data migration to third-party platforms is not required.
• Natural language querying: the user asks questions as they would with a colleague. For example: "Which products had the lowest margin in the last three months?" or "Show me sales trends by region in Q2."
• Interactive output generation: the system returns charts, tables, and dashboards that can be shared with the team, exported, or automatically updated as the underlying data changes.

Plus, the Data Agent can suggest proactive analysis. So, it doesn't just answer direct questions, but also spots oddities, trends, and connections on its own that you might miss doing things manually.

From a technical standpoint, OpenAI developed this agent as a direct evolution of the Coding Agent already applied to research and martech , extending computational reasoning capabilities to a non-technical audience.

Real-world use cases for Italian SMEs

The relevant question for a mid-sized company is not "how does it work technically" but "where does it save me time and where does it help me make better decisions." Below are three high-frequency scenarios in Italian SMEs.

Monitoring of sales performance

A sales manager can query the company CRM by directly asking which customers haven't made purchases in the last 60 days, or which geographical areas show a drop in revenue. Subsequently, the Data Agent generates an updatable dashboard that can be shared with the sales team without going through IT.

Margin analysis by product line

Manufacturing and distribution companies often struggle to cross-reference sales data and operational costs in real time. The Data Agent makes it possible to connect these sources and get a consolidated view of the margin by SKU, channel, or customer. As a result, pricing decisions become faster and better informed.

Automated reporting for management

Instead of spending hours every week preparing manual reports, the Data Agent can automatically generate a summary of key KPIs. Similarly, it can update dashboards in real time, reducing the operational workload for administrative and marketing departments.

Comparison with traditional BI tools

Tools like Tableau, Power BI, or Looker offer advanced analytical capabilities. However, they require a significant learning curve, a well-structured data infrastructure, and often, dedicated personnel for model maintenance. For an SME with 20-100 employees, this represents an investment that is difficult to justify.

ChatGPT Work's Data Agent positions itself differently. It doesn't replace an enterprise BI platform in highly complex scenarios. Instead, it covers the vast majority of a small to medium-sized business's daily analytical needs with a much lower barrier to entry.

In this context, it is worth noting how other players are moving in the same direction. For example, the comparative analysis between GPT-6 Astra and Claude Fable 5.1 highlights how competition between models is accelerating structured reasoning capabilities, with direct benefits for applications like this one.

Trade-offs and limits to consider before adoption

No tool is without limitations. The Data Agent is no exception, and an honest evaluation requires considering a few critical aspects.

• Input data quality: the system only spits out accurate output if the source data is clean and well-structured. Duplicate, incomplete, or messy data leads to misleading insights. First things first, you need to audit your existing data sources.
• Governance and privacy: Connecting sensitive corporate data to a cloud platform requires careful evaluation of security policies. OpenAI has stated that enterprise company data is not used for model training, but every organization must verify compliance with its own regulatory framework, including GDPR.
• Dependency on how questions are phrased: output quality depends on the user's ability to formulate relevant questions. Therefore, a minimum of internal training remains necessary to fully leverage the tool.
• Integration with legacy systems: not all Italian management and ERP systems have native connectors. In some cases, an intermediate integration layer might be needed.

Even so, the tradeoff between perks and setup hassle still makes sense for most mid-market SMBs. You'll actually notice you're spending less time putting reports together by hand within just the first few weeks of using it.

AI agent automation: a rapidly expanding ecosystem

The Data Agent is not an isolated product. It fits into a broader ecosystem of AI agents that are redefining business processes across the board. Similarly to what is happening with the automated visual asset management using Gemini Spark , data analysis is also shifting towards models where AI performs complex operations based on user instructions, without requiring specific technical skills.

This trend has direct implications for those working in Digital marketing and strategic planning. In fact, the ability to quickly access insights about your customers, acquisition channels, and campaign performance becomes a concrete competitive advantage, no longer reserved for large companies with dedicated data science teams.

We at SHM Studio watch this evolution closely as part of our work on applied artificial intelligence services . In particular, we are evaluating how to integrate tools of this type into the workflows of clients who manage multi-channel campaigns and need consolidated reporting in a short time.

To explore the landscape of AI agents and their impact on business processes, the section dedicated to automation and AI agents gathers the latest analyses published by our team.

Recommended decision: when and how to start

The practical question for a marketing manager or business owner is: is it worth adopting the Data Agent now, or is it better to wait for further maturation of the technology?

The answer depends on three variables. First of all, the volume of data the company manages daily. Secondly, the frequency with which manual reports are produced. Finally, the availability of internal technical resources to handle the initial integration.

For companies that produce weekly or monthly reports with data from multiple sources — CRM, e-commerce, Google Ads or LinkedIn campaigns — adoption makes sense even at the current stage. Experimenting with these tools early allows for the development of internal skills that become a competitive differentiator in the medium term.

On the other hand, for companies with highly fragmented data or outdated management systems, it is advisable to invest in a data cleaning and integration project first. Otherwise, the risk is getting unreliable outputs that lead to bad decisions.

Who manages google ads campaigns or LinkedIn campaigns will find particular value in the Data Agent's ability to cross-reference performance data with CRM data, obtaining a unified view of return on advertising investment. Furthermore, those who invest in SEO you can use the tool to analyze correlations between organic traffic and conversions in a much more detailed way than traditional tools.

In summary, ChatGPT Work's Data Agent represents a significant step towards the democratization of data analysis. It doesn't eliminate the need for a solid data strategy, but it substantially lowers the technical barrier to access it. For Italian SMEs looking to accelerate their decision-making process, now is the right time to start experimenting.