ChatGPT-4o comes with a new focus on enterprise needs

The latest iteration of OpenAI’s ChatGPT-4o is twice as fast as GPT-4, and its “o”—for “omni” multimodal capability—brings text, vision, and speech into the same neural network. Phil Fersht’s blog post Ten reasons why GPT-4o will pour fuel onto the GenAI smoldering platform explores the new capabilities this delivers. Here, we dive into what those capabilities mean for enterprise solutions.

4o meets the enterprise’s need for integration in workflows and processes

ChatGPT-4o responds to enterprise needs like no previous iteration. OpenAI has added protocols and interfaces to overcome some of the complexities of integration (see Exhibit 1) with third-party applications and services, including customer relationship management (CRM) systems, enterprise resource planning (ERP) systems, project management tools, and more. The upshot is that companies can now more easily integrate 4o into their custom workflows and business processes, improving access to real-time data from integrated systems and automating routine tasks by interacting with third-party tools and systems.

The improved integration capabilities include security features to ensure data exchange between ChatGPT and the enterprise systems is compliant, secure, and scalable.

Exhibit 1: Enterprise leaders identify regulatory compliance and security risks and integration among their most significant challenges—ChatGPT-4o addresses these and more

Source: HFS Research 2024; N= 550 enterprise leaders

Integration with CRM, ERP, project, and business intelligence platforms and tools is now easier—this is where the GenAI wars will be won

In CRM, 4o can integrate with platforms such as HubSpot or Salesforce to generate reports on customer interactions. It can integrate with SAP or Oracle to assist with inventory management, order processing, or financial reporting. It can connect with project management tools such as Jira, Trello, and Asana, track project progress, assign tasks, and send automated reminders. Similar integrations with human resources platforms such as Workday and BambooHR and business intelligence tools such as Tableau and Power BI are now more straightforward.

Integration with enterprise systems will likely be where the GenAI wars are won, ensuring that AI plays its most effective role without disrupting workflows. Microsoft has set its course with Copilot and Azure, and Google is bringing its Project Astra (a multimodal assistant program from DeepMind) to an Android device near you.

Custom instruction gives enterprise users more control over their bots

Custom instruction is another standout update in 4o. It gives enterprises more control over their 4o bots’ behavior, offering more precise alignment with operational needs, cultural context, and communication strategies and, in so doing, reducing the risk of offending through words or actions.

Custom instructions enable users to set behavioral guidelines for different contexts—for example, keeping a professional tone in business communications or adopting an open and friendly approach in customer service calls. Companies also will benefit from the new capability of setting preferred vocabulary and specific phrasing to align with branding. Custom instructions can include context-specific details such as references to ongoing projects, FAQs, or an individual customer’s preferences.

You can use the instruct function to ensure 4o meets compliance and sensitivity concerns, whether in response to regulatory demands or cultural expectations. In addition, anyone tasked with data analysis, report generation, or customer support can set instructions on how it should prioritize information, format responses, and handle specific queries.

Exhibit 2: ChatGPT’s Custom Instruction update instills some discipline into bot activity, giving enterprise leaders confidence the bots will behave as directed

Source: HFS Research and Dall-e, 2024

More control in customer support, report writing, and content creation

For example, your custom instruction in customer support could be: “When responding to customer queries, always address the customer by their first name, use a friendly and helpful tone, and be clear while giving step-by-step solutions.”

Or, if you are creating reports, you can give instructions such as: “When generating reports for the sales team, use formal language, include data visualizations where possible, and ensure all figures reference the latest available data.”

For more control over content creation, you could try: “When drafting social media posts, use an engaging and casual tone, incorporate industry-relevant trending hashtags, and include a call to action.”

Ten enterprise solutions the new 4o capabilities enable and enhance

New 4o capabilities such as multimodal inputs, custom instructions, improved multilingual support, advanced reasoning abilities, better integration, and enhanced security enable and enhance a whole range of enterprise solutions. Here are 10 examples illustrating these capabilities’ potential impact on the enterprise.

1. Multichannel interaction in customer support: 4o brings multimodal inputs into one neural network. This enables processing and interpreting text, images, and other data types (potentially voice and video calls). Customer inquiries could be handled across various channels and include visual troubleshooting. For example, a customer could share an image or a video of a faulty item during a chatbot conversation, and the chatbot will analyze the content and offer end-to-end resolution without the need for human interaction.
2. Personalized responses in customer support: 4o comes with custom instructions (described above) that enable tailoring responses to meet company policies and align with the brand voice. Your bots will be able to deliver consistent and personalized customer interactions, potentially with more focus than a human could provide.
3. Real-time language translation: 4o is faster and offers better multilingual support. You now can enable global customer support teams to communicate with customers in their native languages, reducing friction and improving customer experience.
4. Complex data interpretation for analysis and insights: With its advanced reasoning capabilities, ChatGPT-4o can analyze and interpret more complex datasets, as well as spot patterns, correlations, and anomalies. It can synthesize information from diverse data sources and combine structured and unstructured data to deliver a more holistic view of enterprise operations. Users can also converse with the data using natural language. For example, marketing teams can identify emerging trends faster, segment audiences more effectively, and optimize marketing campaigns for better ROI. Finance departments can apply 4o’s advanced reasoning to predict financial performance, assess investment risks, and provide insight into cost-saving opportunities. Operations can use it to analyze supply chain data, help optimize logistics, and predict inventory needs. Human resources can apply advanced reasoning to identify factors influencing employee satisfaction and retention and recommend strategies for improving workplace culture.
5. Automated report generation: This is among the many wins from the integration dividend described earlier. Improved integration capabilities make the automation and accuracy of report generation from raw data and multiple sources more efficient.
6. Intelligent task automation in project management: With improved memory and context handling, you can automate repetitive tasks by maintaining and referencing longer and more complex conversation threads. The integration dividend also supports this solution.
7. Enhanced and globalized collaboration: Better integration with tools and APIs enables 4o to function more effectively with collaboration platforms, offering real-time suggestions and supporting communication among team members in different locations. The improved language capabilities will particularly benefit project teams that don’t speak the same language.
8. Sales forecasting: The advanced reasoning capabilities referenced in point 4 can help sales teams adjust their real-time strategies based on accurate forecasts. Integration with platforms such as HubSpot and Salesforce inevitably contributes to this performance upgrade for sales teams.
9. Streamlined recruitment process in HR: 4o’s improved context handling, combined with the new custom instruction capability, means recruitment tasks such as screening resumes can be largely automated. And what about initial interviews? If you can instruct the bot to represent the best of your brand through a video interview, why shouldn’t it be your potential employee’s first point of contact? Though it may not be suitable for all roles, it could help enterprises cast their nets wide enough to avoid missing out on talent while enhancing the candidate experience.
10. Regulatory compliance monitoring: 4o’s new enhanced security, delivered through encryption, secure data handling, and access controls, supports regulatory compliance monitoring in many regulated industries. For example, in healthcare, 4o can monitor access to patient records, ensure communications meet US HIPAA requirements, and create reports on what data has been accessed, by whom, and for what purpose. Financial services teams can apply 4o to check that customer data is processed per GDPR guidelines. It can provide access when customers submit requests and take care of data deletion on demand. In corporate governance, 4o can monitor internal controls, document financial transactions accurately, and ensure reporting is transparent and accurate to meet compliance with US SOX legislation.

The Bottom Line: ChatGPT-4o update suggests the GenAI supply side is maturing; leaders must prepare to take advantage.

At our NYC Spring Summit in May, HFS identified that generative AI was still a smoldering platform awaiting fuel that would light a fire under slow-responding enterprises. This technology is moving fast. There’s no doubt that 4o pours fuel on the smoldering platform. Is it enough to trigger a conflagration? Some enterprise hesitancy is due to both integration and control concerns. The fact that 4o addresses these issues AND enables a rack of new enterprise solutions suggests the GenAI supply side is maturing. More of the same will follow from OpenAI and its rivals.

4o turns up the heat and shortens the time left to act. If you aren’t yet convinced to move to GenAI, take another look. If you’ve already embarked on this transition, start taking advantage of these new capabilities and ChatGPT’s focus on the enterprise.