Founders want hybrid PM-engineers
What happened
- OpenAI, Anthropic and Notion are hiring product engineers and forward-deployed builders to turn AI agents into recurring workflows inside customer tools. - Notion posted an AI Product Engineer role for Custom Agents on April 23, while Anthropic seeks engineers to codify repeatable deployment patterns. - The hiring shift tracks agent rollouts moving from pilots to production workflows. (deloitte.com)
Why it matters
Founders and AI companies are hiring for one job shape again and again: people who can define the workflow, wire the tools, and ship the product. (openai.com) (indexventures.com) (anthropic.com) Notion published an AI Product Engineer opening on April 23 for its Custom Agents team, which automates recurring work like filing tasks, writing reports, and answering questions from a knowledge base. The role asks for someone who can prototype features, scale asynchronous workflows, and work across artificial intelligence, infrastructure, and product teams. (indexventures.com) Anthropic is hiring a Forward Deployed Engineer in Applied AI to build production applications inside customer systems and deliver artifacts like Model Context Protocol servers, sub-agents, and agent skills. The job description also says the engineer should identify and codify repeatable deployment patterns and feed those lessons back to product and engineering teams. (anthropic.com) OpenAI is hiring a Product Manager for API Agents in San Francisco to guide how developers build agentic applications on top of its models. OpenAI said on April 22 that its new workspace agents in ChatGPT are designed to handle long-running workflows across tools and can be shared inside an organization. (openai.com 1) (openai.com 2) The common thread is not a classic product manager writing requirements for engineers on one side and operators on the other. The common thread is a builder who can map a process, decide what the agent should do, connect the system to real tools, and keep a human review step where the model is likely to fail. (openai.com) (anthropic.com) That job shape follows the way companies are deploying agents. OpenAI’s practical guide says teams should start with workflows where the value is clear, the steps are defined, and the agent can operate safely and predictably. Anthropic’s engineering guidance says the most successful teams usually use simple, composable patterns instead of elaborate agent frameworks. (openai.com) (anthropic.com) Consultants are describing the same shift from another angle. Deloitte wrote in December 2025 that companies are struggling when they merely layer agents onto human-designed processes, and that the gains come when operations are redesigned for agent-compatible workflows. (deloitte.com) Deloitte also cited Gartner forecasts that 15% of day-to-day work decisions will be made autonomously through agentic artificial intelligence by 2028, up from none in 2024, and that 33% of enterprise software applications will include agentic AI by 2028. In the same chapter, Deloitte said only 11% of surveyed organizations were actively using agentic systems in production. (deloitte.com) Recruiters are already packaging that demand as a hybrid search. Bamboo X described a seed-stage artificial intelligence startup that spent more than six months looking for a PM-engineer hybrid before filling the role in five weeks with a $160,000 base salary, $5,000 sign-on bonus, and equity. (bamboo-x.com) The pitch from employers is getting more concrete than “use AI.” They want people who can turn a messy manual process into a repeatable system, decide where the model should act, and then measure whether the workflow actually saves time or reduces errors. (anthropic.com) (indexventures.com) (openai.com) That is why the market keeps converging on the same hire: not a pure manager and not a pure coder, but someone who can make an agent useful inside real work. (openai.com) (deloitte.com)
Key numbers
- Notion posted an AI Product Engineer role for Custom Agents on April 23, while Anthropic seeks engineers to codify repeatable deployment patterns.
- (openai.com) (indexventures.com) (anthropic.com) Notion published an AI Product Engineer opening on April 23 for its Custom Agents team, which automates recurring work like filing tasks, writing reports, and answering questions from a knowledge base.
- OpenAI said on April 22 that its new workspace agents in ChatGPT are designed to handle long-running workflows across tools and can be shared inside an organization.
- (openai.com 1) (openai.com 2) The common thread is not a classic product manager writing requirements for engineers on one side and operators on the other.
Quick answers
What happened in Founders want hybrid PM-engineers?
OpenAI, Anthropic and Notion are hiring product engineers and forward-deployed builders to turn AI agents into recurring workflows inside customer tools. Notion posted an AI Product Engineer role for Custom Agents on April 23, while Anthropic seeks engineers to codify repeatable deployment patterns. The hiring shift tracks agent rollouts moving from pilots to production workflows. (deloitte.com)
Why does Founders want hybrid PM-engineers matter?
Founders and AI companies are hiring for one job shape again and again: people who can define the workflow, wire the tools, and ship the product. (openai.com) (indexventures.com) (anthropic.com) Notion published an AI Product Engineer opening on April 23 for its Custom Agents team, which automates recurring work like filing tasks, writing reports, and answering questions from a knowledge base. The role asks for someone who can prototype features, scale asynchronous workflows, and work across artificial intelligence, infrastructure, and product teams. (indexventures.com) Anthropic is hiring a Forward Deployed Engineer in Applied AI to build production applications inside customer systems and deliver artifacts like Model Context Protocol servers, sub-agents, and agent skills. The job description also says the engineer should identify and codify repeatable deployment patterns and feed those lessons back to product and engineering teams. (anthropic.com) OpenAI is hiring a Product Manager for API Agents in San Francisco to guide how developers build agentic applications on top of its models. OpenAI said on April 22 that its new workspace agents in ChatGPT are designed to handle long-running workflows across tools and can be shared inside an organization. (openai.com 1) (openai.com 2) The common thread is not a classic product manager writing requirements for engineers on one side and operators on the other. The common thread is a builder who can map a process, decide what the agent should do, connect the system to real tools, and keep a human review step where the model is likely to fail. (openai.com) (anthropic.com) That job shape follows the way companies are deploying agents. OpenAI’s practical guide says teams should start with workflows where the value is clear, the steps are defined, and the agent can operate safely and predictably. Anthropic’s engineering guidance says the most successful teams usually use simple, composable patterns instead of elaborate agent frameworks. (openai.com) (anthropic.com) Consultants are describing the same shift from another angle. Deloitte wrote in December 2025 that companies are struggling when they merely layer agents onto human-designed processes, and that the gains come when operations are redesigned for agent-compatible workflows. (deloitte.com) Deloitte also cited Gartner forecasts that 15% of day-to-day work decisions will be made autonomously through agentic artificial intelligence by 2028, up from none in 2024, and that 33% of enterprise software applications will include agentic AI by 2028. In the same chapter, Deloitte said only 11% of surveyed organizations were actively using agentic systems in production. (deloitte.com) Recruiters are already packaging that demand as a hybrid search. Bamboo X described a seed-stage artificial intelligence startup that spent more than six months looking for a PM-engineer hybrid before filling the role in five weeks with a $160,000 base salary, $5,000 sign-on bonus, and equity. (bamboo-x.com) The pitch from employers is getting more concrete than “use AI.” They want people who can turn a messy manual process into a repeatable system, decide where the model should act, and then measure whether the workflow actually saves time or reduces errors. (anthropic.com) (indexventures.com) (openai.com) That is why the market keeps converging on the same hire: not a pure manager and not a pure coder, but someone who can make an agent useful inside real work. (openai.com) (deloitte.com)
