how our job will evolve and even exist than now with the emergence of AI Agents. But let me be upfront that AI tools don’t change the elemental job of the PM, which is to discover the essential problems to resolve and guide the perfect ideas to implementation. AI Agents can definitely augment and, in some cases, replace certain activities, and that’s a great thing.
Don’t give in to alarmist narratives of how your job might be negatively impacted. Each PM role is exclusive. While we share common features: create product concepts, define requirements, iterate with customers, GTM, the day-to-day work of a social media PM may be very different from the work of a cloud infrastructure PM, requiring different features to be automated. Because the mini-CEO of your product, only you choose what is required for fulfillment. So you need to be the one to choose how your job will evolve to make your product successful. You might be in the driving force’s seat to decide on what to reinforce or automate with AI agents to perform your job higher. A recent Stanford research paper defines a useful framework for making these decisions and divulges that employee desire for automation is more of a defining factor for successful adoption than simply technical feasibility.
The Human-Centric Framework for AI Adoption
The Stanford study sheds light on ways AI agents can profit work. It introduces the Human-Centric Automation Matrix, a 2×2 plotting Employee Desire against AI Capability, to assist prioritize AI automation of PM tasks. Highlighting that staff need to automate tedious, repetitive tasks but are deeply concerned about losing control and agency. An amazing majority of staff within the study anxious about accuracy and reliability of AI, with fear of job loss and lack of oversight as other concerns. A living proof in highlighting the risks of full autonomy is the recent issue with Replit wiping out a whole database of an organization, fabricating data to cover up bugs and eventually apologizing (See FastCompany).
This trust deficit logically rules out full autonomous AI for high-stakes communication with customers or vendors communications. The preference is clearly for AI taking a partnership or assistive role. The paper introduces the Human Agency Scale (HAS), to measure the degree of automation (cf. levels of autonomy in self-driving cars):
- H1 (no human involvement): The AI agent operates fully autonomously.
- H2 (high automation): The AI requires minimal human oversight.
- H3 (equal partner): Human and AI have equal involvement.
- H4 (partial automation): The AI is a tool that requires significant human direction.
- H5 (human involvement essential): The AI is a component that can’t function without continuous human input.
Most staff are fairly comfortable with the H3-H5 range, preferring AI to be a partner or a tool and never a alternative. The choice for the PM isn’t just what to automate but in addition to which degree we must always hand over control to the AI Agent.
The concept is explained higher with a 2×2 matrix with Automation Capability on the X-axis and Automation Desire on the Y-axis. The 4 quadrants are classified as:
- Green Light Zone: High automation desire and high capability
- Red Light Zone: Low desire and high capability
- R&D Opportunity Zone: High desire but low capability
- Low Priority Zone: Low desire and low capability
The framework helps determine which jobs are possible and still have a high likelihood of getting adopted within the workplace.
Putting the Framework into Motion
As a substitute of blindly following mandates to “use AI Agents” PMs should do what they do best – think strategically on what’s best for the business. Use this 2×2 to discover the areas ripe for automation that can have probably the most impact and keep your team happily productive.
- Green Light Zone: These can be the highest priority. Automating market insights, synthesizing customer feedback, and generating first drafts of PRDs are tasks which might be each technically feasible and highly desired. They save time and reduce cognitive load, freeing you as much as do higher-level strategic work.
- Red Light Zone: Proceed with caution. AI has the flexibility to robotically generate marketing collateral, manage customer communication or take care of vendor contracts, but PMs aren’t ready to offer up control on these high-stakes tasks. An error can have serious consequences and augmentation (H3-H4 on the HAS scale) could also be the proper option.
- R&D Zone: Have to innovate to get the tech able to automate the job. While there’s a high desire for automation however the tech isn’t ready, more investment is required to get us there.
Most significantly, take charge. The PM-to-engineer ratio isn’t improving anytime soon. Adding agentic capabilities to your toolkit is your best bet for scaling your impact. But drive with caution. To thrive and make yourself indispensable, you need to be the one shaping the long run of your role.
Key takeaways
- Prioritize Desire Over Feasibility: The Human-Centric Automation Matrix is a robust tool. It enhances traditional tools (e.g., Impact/Effort, RICE, Kano) by considering adoption and trust, and not only capability. True success is in constructing AI tools that your team actually uses.
- Think Agency and Not Just Automation: Use Human Agency Scale (H1-H5) to find out the extent of automation. Data-heavy and repetitive PM tasks (e.g., market insights discovery, data-based prioritization) fall into the “Green Light” zone resulting from high employee desire and readiness for AI. These are also inputs to decision making, so vital checks and balances are already in place in subsequent steps. Others may fall into just H4, as just being a tool. This approach is helpful in managing risk and constructing trust.
- Concentrate on augmentation in high-stakes areas: Creative, strategic, or customer-facing tasks (aka “Red Light” opportunities) match well with augmentation strategy. While AI will generate options, analyze data and supply insights, final decisions and communications must remain with humans.
- Core PM Skills Are More Beneficial Than Ever: AI Agents will handle more of the information-focused activities. We want to further develop our uniquely human skills: strategic considering, empathy, stakeholder management, and organizational leadership.
The long run of product management might be shaped by the alternatives of forward-thinking PMs, not by just the AI’s capabilities. Essentially the most successful and adopted approaches might be human-centric, specializing in what PMs really want to excel. Those that master this strategic partnership with AI won’t only survive but in addition define the long run of the role.
References
[1] Y. Shao, H. Zope, et al. (2025). “Way forward for Work with AI Agents: Auditing Automation and Augmentation Potential across the U.S. Workforce.” arXiv preprint arXiv:2506.06576v2. https://arxiv.org/abs/2506.06576
[2] S. Lynch (2025). “What staff actually need from AI.” Stanford Report. https://news.stanford.edu/stories/2025/07/what-workers-really-want-from-ai
