OpenAI is trying to move AI agents from developer workflows into the broader white-collar workplace, according to a TechCrunch report focused on the company’s ChatGPT Work push. The product, released last month and available on OpenAI’s lowest subscription tier at $20 a month, is being positioned as a way for non-engineers to hand multistep digital tasks to large language model-powered agents. TechCrunch describes ChatGPT Work as one of OpenAI’s biggest current bets. The aim is not simply to answer prompts, but to connect models to the tools professionals already use and let them perform tasks across those workflows. The report names accountants, investors, doctors and other computer-centered workers as the kinds of users OpenAI is trying to reach. The product is tied closely to the playbook that has already taken hold in software development. TechCrunch reports that ChatGPT Work is a modified version of Codex, OpenAI’s coding tool, adapted for people outside engineering. The strategic idea is straightforward: if engineers can use agents to complete software tasks, OpenAI wants to give other functions a similar interface for longer, more autonomous work. The trade-off is control. TechCrunch reports that Andrew Ambrosino, the lead engineer for OpenAI’s desktop app, has given the app access to his inbox, Slack account, phone, Notion, Figma and other tools as part of testing agentic workflows. Ambrosino acknowledged to TechCrunch that this creates a risk that an AI system could draw on private information in an inappropriate context, while saying he has accepted that risk for the work. That tension is central to the adoption question. For an agent to be useful, it often needs access to calendars, messages, files, work apps and other sources of context. For many users and employers, those are also the highest-risk systems to expose to automated software, especially when the agent is expected to act rather than merely draft. TechCrunch also frames the product push as commercially significant for OpenAI. Agents that run for longer and perform more complex work consume more tokens, which the report says can make them more lucrative per user. That makes expansion beyond software developers important for OpenAI and for the broader AI industry, which needs use cases large enough to support heavy spending on model training and compute. The competitive backdrop is vertical specialization. TechCrunch points to Harvey in law and Clay in sales as examples of companies pursuing specific professional markets with model-agnostic products, meaning they can use whichever AI systems work best at a given time. The report also cites Christian Catalini writing on Andreessen Horowitz’s Time to Build blog that if AI labs do not secure the complementary assets needed to scale in markets, value may accrue elsewhere. For now, the story is less that agents have already become universal than that OpenAI is trying to make them usable outside engineering. TechCrunch’s reporting suggests the next constraint is not only model capability, but whether ordinary professionals and their employers are willing to grant AI systems enough access to make autonomous work useful. Who benefits: OpenAI benefits if ChatGPT Work expands agent usage beyond developers and increases per-user token consumption. Professionals with repetitive computer-based workflows could benefit if the product can safely complete multistep tasks. Who's exposed: Users and employers are exposed to privacy and control risks when agents receive access to inboxes, messaging apps, phones and work tools. OpenAI is also exposed if vertical specialists such as Harvey and Clay win customer workflows with model-agnostic products.