OpenAI case study says Merge uses its models across business tasks

OpenAI published a customer case study on 5 November 2026 describing how integration platform Merge uses GPT‑6 Astra, ChatGPT and Codex for software development and internal business workflows.

The publication matters because it promotes OpenAI’s models as infrastructure for both customer-facing products and internal operations, while offering only vendor-reported figures on reliability and time savings with no independent measurement.

According to the case study, Merge “helps companies build connected software and AI products,” offering customer‑data integrations through Merge Unified, secure tool access for agents via Merge Agent Handler, and model routing and cost controls through Merge Gateway.OpenAI Blog1 OpenAI writes that Merge for Workforce extends these capabilities inside companies by managing which models, tools and business systems employees can access.OpenAI Blog1

OpenAI states that Merge uses GPT‑6 Astra, ChatGPT and Codex “for product reviews, marketing, and business development planning.”OpenAI Blog1 Inside the company, CEO and co‑founder Shensi Ding is described as using Codex to review product code, including work on a setup interface for configuring tool access in Merge for Workforce, while colleagues use GPT‑6 Astra for product testing and marketing.OpenAI Blog1

Reliability and dinner‑planning claims

In a section on rollout, OpenAI reports that Merge launched “a tool for adding connectors to AI products,” and quotes Ding as saying the company is “pretty proud of the quality.” Ding is quoted claiming that “tool calls through Merge succeed 99.9% of the time, compared to public MCP servers, which are around 60%.”OpenAI Blog1 The comparison in OpenAI’s write‑up does not describe the methodology, sample, definition of “success” or measurement period for either figure.

The case study also describes a “dinner bot” workflow used to plan executive dinners for customers and prospects. For one Dallas dinner, OpenAI writes that the bot pulled potential guests from customer relationship management systems and Apollo, checked previous invitations and responses, helped prepare outreach from Ding’s email, maintained an attendee tracker, proposed a seating chart and created calendar notes where sales and customer success teams could add context about each guest.OpenAI Blog1

OpenAI states that the model handled “reasoning and drafting,” while Merge’s connectors provided access to business systems. Ding is described as reviewing the bot’s output, rewriting anything that “sounds off,” and curating the guest list using relationship context the bot did not have, with the team handling hosting, relationship‑building and on‑the‑ground logistics.OpenAI Blog1

In October 2026, Merge reported through the OpenAI case study that preparing guest lists and outreach and maintaining the attendee tracker had gone “from dozens of hours across the company to about 30 minutes per event,” based on the same planning tasks measured over the preceding three months.OpenAI Blog1 OpenAI’s account presents this time‑saving figure as Merge’s own measurement and does not add independent checks.

The case study states that Merge “ran 16 executive dinners in Q3, up from eight in Q2, and plans to host more than 30 in Q4,” without specifying the year.OpenAI Blog1 OpenAI places these figures in a section dated October 2026, implying the quarters refer to 2026 but not stating this explicitly.

Internal examples and leadership guidance

Beyond product engineering, OpenAI reports that “Charlie Lynch on Merge’s sales leadership team built reusable AI workflows for deal approvals, forecast updates, and next steps after renewal calls,” using those workflows to prepare material before reviewing recommendations and deciding what to do next.OpenAI Blog1 The document does not name which model or models those workflows use.

OpenAI writes that Merge’s social media team uses GPT‑6 Astra in ChatGPT to find posts the company should respond to, and that the marketing team used GPT‑6 Astra to clarify messaging for a Merge Gateway Evaluations launch.OpenAI Blog1 Ding is quoted saying, “It’s been especially great for team members who aren’t in engineering, because now they can build anything.”OpenAI Blog1

In a section titled “Leadership lessons,” OpenAI presents its own headings such as “Lead by using the tools” and “Give people space and time to tinker.” Under those headings, the case study quotes Ding telling executives, “If you’re an executive, you are the ceiling for AI fluency in your organization,” and describing experiments on a new product as a lower‑risk setting for AI use.OpenAI Blog1 OpenAI characterizes Merge’s policy as letting each team member choose tools and “go wild and make mistakes,” including time spent testing ideas that do not pay off, but does not present that phrasing as a direct Ding quotation.OpenAI Blog1

Promotional context and agent governance background

OpenAI labels the piece “Merge builds product and business tools with OpenAI technology” and places it in a customer‑story section of its site, positioning it as promotional content highlighting GPT‑6 Astra, ChatGPT and Codex as building blocks for partner products and internal workflows.OpenAI Blog1 Blockchain.News, summarizing the same case study, describes Merge as an integration‑infrastructure company founded in 2020 by Shensi Ding and Gil Feig, and reports that Merge launched Agent Handler, described as a model‑agnostic protocol for secure AI‑agent tool execution, in October 2025.Blockchain.News3

In a separate, undated post on its own blog, Merge outlines Merge for Workforce as “one governed MCP endpoint” for OpenAI “dots,” which it defines as personal agents in ChatGPT that can use plugins and manage ongoing responsibilities.Merge blog2 That post states that IT teams can choose which company tools an employee’s dot may use, apply security rules to each action, and record every tool call routed through Merge, with options to log, redact or block actions when policy rules trigger.Merge blog2

OpenAI’s case study concludes with Ding saying, “Anything that I want to do, I can just tell AI to do it, and it can just do it,” and states that Merge is continuing to use OpenAI technology “to expand what its teams can build,” framing the described deployments as an ongoing effort by the company.OpenAI Blog1