The Agent-Era Org Chart Is Being Rewritten From the Middle Out

Three org charts, one question

Authors Hemanth Asirvatham and Elliott Mokski propose three AI-integrated organizational models in an OpenAI piece: a Beneath scenario in which a new basement of AI labor reports to individual contributors, a Beside scenario in which AI joins the org chart as a colleague working directly for human managers, and a Between scenario in which AI coordinates and monitors work between humans. They argue AI adoption is likely to start at the ground level because those workers are most familiar with the tasks, and that cheap prototyping enabled by AI may reduce the need for traditional planning-intensive red tape. m:…6491

The risks they name are as specific as the models: AI consultancy can centralize power at the top of an organization, pay disparity may increase if judgment and scope become the primary differentiators while AI gives high-quality advice to everyone, and AI could lead to smaller firms or sprawling AI holding companies. AI monitoring can also increase trust and flexibility in remote work by focusing on output rather than performative productivity, the authors argue, which reframes surveillance as a management design choice rather than a technical inevitability. m:…6491

From service desk to enablement engine

At n8n, head of internal AI and automation Romina da Costa is moving the automation team beyond the service desk model, after the previous model reached a ceiling as the company grew and centralized teams created downstream bottlenecks by shifting work rather than fixing processes. The team now manages more than 12,000 internal workflows and prioritizes building the builders over direct workflow production, establishing a network of technical champions in each department and a shared documentation template for all workflows. m:…0677

Two disciplines define the new mandate: the team requires a baseline comparison against current human performance before starting automation projects, and it avoids using language models or agents when simpler, more predictable tools suffice, while prioritizing buying proven software tools over building internal ones to avoid maintenance burdens. Success is now measured by how many teams no longer require direct assistance, and the team is developing an AI fluency framework and working to integrate AI fluency into employee performance evaluations, which turns enablement into a tracked capability rather than a slogan. m:…0677

Expertise and deployment get re-priced

On the Salesforce Admins Podcast, Stacy Jacobs argues certifications help professionals stay updated on best practices for security models and native features, and warns AI tools can hallucinate or pull in inappropriate data when users lack foundational knowledge of the platform. Consultant-level certifications improve the ability to ask better questions and uncover underlying business problems, she says, while human admins provide the common sense and historical context that AI currently lacks, even as AI can help study for certification exams. m:…4444

Jasper is scaling the deployment version of the same idea, expanding its forward-deployed model by placing more engineers and marketing experts inside customer teams, pairing engineers with marketers to define what good looks like and where human judgment is required. The company says a global enterprise cut translation timelines from 22 to 13 days, launched content in nine languages in 11 calendar days with projected annual savings of approximately 300,000 dollars, and a global bank increased citations across leading language models by 114 percent within two weeks, moving from seventh to top three on Perplexity with 77 percent brand presence in measured AI-generated answers. Enterprise marketing teams often lack a written, agreed standard for good work, Jasper argues, and the forward-deployed team exists to write that standard into repeatable evaluations. m:…1172