OpenAI's Operator Agent: A Delicate Dance Between Retail Giants and AI

OpenAI's Operator Agent: A Delicate Dance Between Retail Giants and AINews broke on March 20th about OpenAI's Operator, an AI agent launched in January, and its partnership with DoorDash as a launch partner. Operator assists users with tasks like trip planning, report writing, and online shopping

OpenAI's Operator Agent: A Delicate Dance Between Retail Giants and AI

News broke on March 20th about OpenAI's Operator, an AI agent launched in January, and its partnership with DoorDash as a launch partner. Operator assists users with tasks like trip planning, report writing, and online shopping. However, DoorDash raised a crucial question: if all website traffic consisted of AI bots instead of human users, the company's business model would face serious challenges. This concern highlights the potential threat posed by agents like Operator to consumer applications.

Recently launched agents from OpenAI, Anthropic, and Google can automate various tasks by taking control of a user's browser, such as applying for jobs, finding doctors, and online shopping. This presents an unprecedented challenge for website operators: sophisticated agents could lead users to completely bypass company applications, positioning AI companies as powerful intermediaries.

DoorDash, for example, relies onalbeit relatively smallbut growing revenue from restaurant ads on its website. If users access DoorDash indirectly through agents, the advertising value would plummet, potentially causing restaurants to reduce ad spending and ultimately impacting DoorDash's revenue. A source involved in Operator's development confirmed this as a central discussion point during OpenAI's collaboration with DoorDash.

Significantly, larger players with even more substantial advertising businesses, like Uber and Instacart, also agreed to be launch partners for Operator. Yet, the widespread adoption of browsing agents poses a disruptive threat to the business models of all these companies.

This "disintermediation" isn't new; the rise of social media and search engines similarly impacted the news publishing industry. However, Andrew Lipsman, a retail media consultant, suggests the AI threat to advertising might not materialize. He points to early technologies like Amazon's Alexa as examples where predictions of industry upheaval ultimately failed to fully materialize. He argues that many similar predictions have proven inaccurate.

The discussion surrounding the impact of agents on retail is becoming increasingly public. In February, Amazon CEO Andy Jassy, during an earnings call, addressed the impact of agents on Amazon's e-commerce business, stating that most retailers will establish cooperating rules with agents, and Amazon will be no exception.

Some companies have already started taking action. Reddit, for instance, has banned Operator from accessing its website.

Addressing concerns from retailers like DoorDash, OpenAI argues that browsing agents can be a significant source of referral traffic for retail websites. Launch partners like DoorDash receive aggregated data on Operator usage, providing insights into how it drives traffic. DoorDash executives have also stated that participation helped them understand the technology better. However, this also highlights the potential conflict between retailers and AI developers.

Hari Vasudev, CTO of Walmart US, believes one way for retailers to counter the threat is by developing their own agents to interact with users. Ali Musa, a DoorDash spokesperson, stated that the collaboration with OpenAI allowed both companies to learn from each other, exploring the potential of agents as an additional customer channel, and suggested the impact on their advertising business wouldn't be threatening but rather additive and complementary. OpenAI and Instacart spokespeople declined to comment on the meeting or widespread website owner concerns, and a spokesperson for Uber did not immediately respond.

Even if browsing agents don't become as ubiquitous as OpenAI anticipates, the company already provides increasingly important referral traffic to retail websites through ChatGPT. Adobe reports that while traffic from generative AI applications to retail websites is still far less than from paid search or email, it's rapidly growing. An Adobe survey found 39% of respondents used generative AI for online shopping tasks.

ChatGPT's rapid growth gives OpenAI significant leverage, empowering them when demanding website publishers or retailers allow agents access to their applications and websites. Many retailers have existing business relationships with OpenAI, paying to use their AI models to power features in internal or consumer applications. This dependence might make it harder for them to oppose OpenAI's launch of products like Operator.

Agent technology is a hot area in AI. OpenAI released tools earlier this month to help developers build their own browsing agents. While these agents have many flaws, their prominence is rapidly growing. For example, Manus in China has gained attention for automating the generation of in-depth stock analysis reports.

The rise of browsing agents is forcing businesses to rethink their advertising and product discovery strategies. Vasudev suggests retailers can mitigate the threat by developing their own agents to interact with users, offering recommendations or supplementary product information. He envisions that browsing agents and Walmart's hypothetical agent would "build a shared understanding of customer needs," with Walmart's system needing to provide information on all its in-store items to the browsing agent.

However, this future remains some time away. Currently, Operator is only available to a limited number of ChatGPT Plus subscribers in certain countries, who pay $200 per month for a more powerful version of the chatbot capable of handling more computationally intensive queries. Furthermore, some users complain about Operator's slow response times and difficulty handling complex web pages. This demonstrates that despite the immense potential of agents, the technology is still in its early stages and faces significant technical and commercial challenges. The ongoing interplay between retailers and AI companies will shape the future of e-commerce and online services.