Following a rocky initial presentation, OpenAI returned to the stage at DevDay 2026 for “Take Two” of their highly anticipated “Dots” demo. The presentation aimed to prove that these AI agents—referred to as Dots—are not just conceptual, but highly functional tools capable of managing complex, real-world tasks simultaneously.

The live demonstration showcased an unfiltered, unscripted workflow, providing a compelling look into how professionals can use a Dot as a personal assistant, project manager, and data analyst.

Seamless Travel Management

The demo began with a classic, high-stress scenario: a last-minute business trip[cite: 9]. The presenter needed a flight to Los Angeles for an important meeting in East Hollywood by 3:00 p.m. the following day, needing to leave the office no later than 10:00 a.m[cite: 9].

The requests given to the Dot were layered and highly specific: find a company-policy compliant, refundable flight, locate hotel options in East LA (with screenshots to review the aesthetics), and provide map locations to check proximity to contacts[cite: 9].

The Dot handled the complexity with ease. It initially suggested a Southwest flight landing at 12:55 p.m., but when asked if that left enough time to navigate LA traffic, the AI adjusted its strategy[cite: 9]. Factoring in LAX exit times (30-45 minutes) and driving time (45-75 minutes), the Dot proactively recommended arriving closer to noon to build a safer buffer[cite: 9]. It then provided visual options for hotels like the Silver Lake Pool & Inn and Cara, allowing the presenter to choose the former while the AI handled the booking and cancellation terms[cite: 9].

Complex Data Analysis and Workflow Optimization

While the Dot was booking the travel arrangements in the background, the presenter seamlessly pivoted the conversation to a completely different domain: analyzing user feedback for software development[cite: 9].

The presenter requested a concrete workflow to capture user feedback from Slack channels, identify the biggest themes, highlight pressing issues lacking solutions, and pinpoint outstanding gaps[cite: 9]. The Dot immediately summarized the existing manual workflow—which included batching messages, grouping them into a master list, and linking them to Linear tickets and Pull Requests (PRs)[cite: 9].

The AI agent and the presenter then strategized a new system. The Dot was instructed to prioritize large issues without credible fixes[cite: 9]. More impressively, it was tasked with a proactive communication loop: when a PR lands that addresses a major user complaint, the Dot is to go directly into Slack, respond to the users, link the PR, and ask them to test the fix[cite: 9].

When asked for the current biggest issue based on the last 24 hours of data, the Dot correctly identified reliability—specifically, users being unable to tell if their own Dot was working, missing replies, or experiencing tool access failures[cite: 9].

Managing Communication Overload

As a final task in the multi-threaded interaction, the presenter addressed internal communication[cite: 9]. With internal teams questioning the relationship between Dots and other product launches (like ChatGPT in Slack and Teams), the presenter asked the AI to pull together their top unanswered Slack messages on the topic and draft responses[cite: 9].

A Glimpse into the Future of Work

The DevDay 2026 “Take Two” demo effectively proved OpenAI’s core promise for Dots. Throughout the presentation, the AI agent maintained context across drastically different tasks—acting simultaneously as a travel agent, a product manager analyzing user data, and a communications assistant drafting internal replies[cite: 9].

By handling the logistics, data sorting, and communication drafting, the Dot freed the presenter to focus on higher-level decision making—exactly the kind of AI-human collaboration OpenAI is hoping to bring to the modern workplace.