Big announcements in the Agentic AI world at Google Cloud Next 2025. Agentspace is here. Build AI agents, operate AI agents and manage AI agents.

2 ways to get started:

✨ Agentspace! Here's how your employees can create custom AI agents with a chat-based interface. Get early access: https://lnkd.in/gdvX27j3. I'll be posting a demo tomorrow :-)

✨ Prefer to code? You can now build agents using Google's Agent Development Kit (ADK). An open-source, flexible framework to build & launch multi-agent systems. Start here: https://lnkd.in/g7ZXSRxG (SDK)

Book Signing Event [ April 17 ]

180 people signed up and we have 20 spots left for my Bay Area book signing event! Join us for dinner, a free book (will sign it live!) and fireside chat with Q&A.

AI Agent Certification for PMs in 2 weeks!

Get $300 off for the next 72h.

An extremely practical Agentic Bootcamp. Build 5 real no-code product agents. Led by Relay, Wayfound, Cassidy & Zapier executives.

WhatsApp Community!

I now have a WhatsApp Community for those interested in discussing AI while on the go!

Get 12-months of perplexity pro for free by upgrading to annual paid and then fill out the form

Subscribe now

Guest post by Adam Judelson, a former Palantir head of Product that owns a product agency helping companies achieve transformative visions, whilte hosting the Emergent podcast and newsletter.

AI agents can add superhuman capabilities to your product, but what is the intuition for when to incorporate them into your strategy?

Put simply, AI agents generate maximum value when we place them wherever modestly intelligent humans existed in workflows in the past. Agents gain power particularly when chained together, used in concert, or when forward-deployed. An agent can live on your computer to interact with your file system, on your social media account, deep inside a factory assembly line, on a satellite, or wherever that human-like quality is required for product success.

For those new to the buzz around agents, they are what they sound like: miniature GPTs that you can place inside complex architectures to perform cognitive tasks that machines have historically struggled with. They can operate autonomously, in clusters with other agents, or in a hybrid model combined with existing microservices. They leverage generative AI, so at a minimum, they can:

• Make a decision or judgment

• Summarize text or other content

• Analyze data or evaluate

• Generate text, images, or even systems

• Take actions as a real user (for example, make a purchase or send an email)

Generative AI is best for challenges requiring a modicum of judgment, whereas traditional services are superior for tasks that require certainty and precise instructions.

With this high-level heuristic in mind, let’s explore thinking models that could advance your product, moving from the least sophisticated to the most advanced. Along the way, we’ll provide examples and diagrams to illustrate how each might work across six varied use cases that we’ve named Inspectors, Conductors, Assembly Lines, Hydras, Missions, and Mobius Cycles.

Inspectors

Inspector agents excel at handling inconsistent human outputs, enforcing standards, or by providing qualitative critiques to ensure that work meets or exceeds expectations. In short, they can act as a quality-control gate. I like to think of this set of use cases as giving the agents a persona and then implanting them into a system or process where quality control or stress testing is needed. For example, a customer service system for a Fortune 500 company requires care and feeding across many dimensions:

• How technically competent were the representatives?

• Did they use appropriate soft skills in dealing with customers?

• Are they respecting company policies and processes?

A traditional human manager would have to tackle these challenges themselves, but a series of agents can operate on the manager’s behalf to spot concerning interactions (and eventually to monitor non-human agents too!). Imagine the following interaction among a hypothetical CVS pharmacy customer, a CVS customer service representative, and a Quality Control AI agent.

The above scenarios were scripted and analyzed 100% by AI agents, and you can see each taking responsibility for acting as a particular persona, providing valuable context for the customer service department in terms of how they want to train their human or AI agents to act. Just as we used to bake certain mores into a company culture, we now need to bake these into the AI to get the customer experience and personality we desire.

Stress testing is another strong use case for agents. Whenever a human or an AI comes up with a result, a stress-testing agent can act as a critical bystander trying to elevate the quality of what was produced. Take, for example, the following AI-generated summary of the Los Angeles Dodgers’ World Series win this year:

Let’s create and dispatch another agent as a sports news expert whose job it is to critique and improve on these clips. The agent correctly notices the lack of a score, no mention of the pitchers, very little sense of the actual action, and more. Here is an improved version after the writer agent stress tested and improved upon the summary. Unfortunately, our agent still needs some training, since the Dodgers did win much more recently than 1988, but the draft is substantially improved. A fact-checking agent is probably needed too!

While these agents are presented together in this example, in a live production system each might occupy different space in different or complementary systems. The choice is up to the product leaders and the system architects and will depend in large part on how often each agent might be reused in different contexts.

The inspector model excels when:

• The results of an ongoing process have high variance in quality

• The quality is best critiqued by a human, and therefore encoded in an agent

• The results need to conform to certain standards, but where those standards require ongoing judgment

Conductors

Increasingly sophisticated product systems can rely on AI agents to serve as the Conductor in Chief, orchestrating and selecting actions based on the content of the request. Conductor use cases tend to be ones in which the product receives a natural-language request from a user, and the conductor agent must determine, based on the information supplied in the request, where to route it, and then recompile the outputs from each part of the system into one coherent response for the end user.

One company I advise, Danti.ai, provides customers with a search interface for finding hard-to-access content related to places on Earth—for example, satellite imagery, the locations of shipping vessels, property reports, and much more. Each data source is accessed differently, and rarely is an LLM-based search the right answer for queries where high precision is required in the response, so a conductor agent is needed to determine when to invoke traditional, pre-existing tools that do not use generative AI or to dispatch other agents.

For example, when a user asks a question like “What homes were damaged in Tampa in the recent storm?” the conductor must use its judgment to infer that news data, satellite imagery, and severe-storms data sources are ideal for answering this question. The conductor must make that selection (Anthropic calls this paradigm “tool use”), request the news, use it to find the storm in question and its date and exact location, and then form a query to find satellite imagery based on dates and latitudes and longitudes so the user can receive a result set that answers the question. Meanwhile, the conductor must also meaningfully combine the disparate answers into one user-facing “result” that summarizes what it found.

Screen capture from danti.ai during a search for a dam collapse incident in Ukraine showing news, satellite imagery, shipping information, and much more

The conductor model excels when:

• Your inputs are natural-language requests from humans or other LLMs

• The answer or result requires accessing many different resources but frequently not all of them, and where judgment is required to determine which ones

• The result requires judgment to recombine

Assembly Lines (aka Extended Conductors)

Sometimes the task is finite, such as in the search example above (query in -> result out); however, there is a powerful variant of the conductor model that I like to call the “assembly line” because it involves an open-ended set of actions toward an ongoing persistent goal. It’s easiest to imagine this paradigm as an agentic system that is responsible for keeping an assembly line on track. Let’s take the following literal example of a drone production assembly line:

1. A pre-existing quality-control automation throws “Error 5521” while a drone is being built.

2. The agent checks the production history archives, finds the error code, and learns that there is a 63% chance that this is due to a loose cable.

3. The agent decides to capture an image of the issue and deploys a camera robot to verify the situation.

4. The agent receives the image and sends it to a quality-control plus computer-vision agent, which assesses that the issue is the result of poor attachment of the balance plug to the power distribution board.

5. An assembly-line-routing agent re-routes the assembly back to the power technician.

6. The agent creates a defect ticket and assigns it to the manufacturing engineer.

An assembly-line procedure from First Resonance’s ION product, which acts as an operating system for advanced manufacturing factories

The assembly-line model excels when:

• There is an ongoing goal that must be maintained over the course of a complex process (in this case, keeping the assembly line running)

• The process can be monitored in/at one or many places

• Issues will arise in pursuit of that goal that require judgment to resolve

• There are tools available to the agent to pursue resolution

Hydras

The AI agent world often mimics the human world, and, much as humans specialize and gain knowledge from their experiences, so too can agents. Similar to humans, often a superior outcome is achievable only when we collaborate, so the “hydra” refers to use cases when training multiple personalities that each act like humans with different interests helps us to sort out the best plan of action. This can be particularly advantageous when the user of the product would benefit from seeing how the sausage is made and might want to see the spikiest viewpoints and not just the final summary or “average” result.

Take, for example, a Board Meeting preparer system. A naive approach to preparing for such a presentation is to toss your deck into an LLM such as ChatGPT and ask for feedback or ask it to play a few roles. Much more powerful is tailoring a specialized investor agent, a marketing agent, a sales agent, a product agent, an engineering agent, and so on, and then having each critique the plans with data and experience to back up their specialities trained over time. Better yet, in addition to this critique, it can be advantageous to set the agents up to have a conflict with one another that they then have to also resolve, giving you crucial insights into how to navigate these treacherous waters and thread a needle through the solution.

The hydra model excels when:

• Customers need to make important decisions where extra time is worth the tradeoff

• Where many subject matters come together to impact the decision

• Where data and experience exist to inform multiple “personalities” and their biases

The Mission

Mission use cases are ones in which a system must take on a new or novel goal each time it is used and needs to make a series of planning and execution decisions to achieve this. Imagine trying to offload the planning and booking of an entire vacation end-to-end to an agentic system. Or think about a robotic arm in space that needs to fix a broken solar panel on a space station, and tomorrow it needs to unload a new shipment of resources. A more “down to Earth” example is autonomous drone reconnaissance. Suppose we want to ask a drone, “Is there any new construction along the river within one mile of here?” This is a highly specialized mission. Agents might operate as follows:

1. An agent to plan the mission and recruit the other agents below

2. An agent to interpret the question and parse out the location and target

3. An agent to plan the flight path within the constraints and communicate with the drone’s flight software

4. An inspector agent to test the flight path against flight rules from the FAA

5. An agent to check if the weather is suitable and to select the time for the mission

6. An agent to trigger the drone’s camera when the drone is along the river

7. An agent to interpret the images and identify construction

8. An agent to make time-lapse image comparisons

9. An agent to summarize the results

This setup highlights how multiple specialized agents work in unison to achieve complex outcomes—each agent doing what it does best while collectively achieving the mission goal.

Diagram of a Skydio autonomous drone inspection breakdown

The mission model excels when:

• An open-ended task must be completed that requires planning and coordination

• There are a large series of steps that often do not proceed in a specified order

• Possibly an interplay between hardware and software

Mobius Cycles

A final variant is what we’re calling Mobius cycles, named to communicate the ongoing cyclical nature of the tasks, and yet they do not always arrive back at the same starting point. These are use cases where we might cycle through the same or very similar sets of tasks, and will do so repeatedly to achieve a goal. This requires thoughtful agent orchestration where multiple agents interact seamlessly, like a symphony of specialized players. Writing code without intervention from a human is a great example, and Devin showcases this pattern wonderfully. Devin is an agentic system that takes a basic prompt for what to build with software, puts it together in code, tests it, corrects issues, deploys it, and even finds bugs and improves the code over time. It is constantly returning to the same cycle of actions, but it is exercising judgment about which order and when and is even evaluating the results continuously.

Thinking in specifics, we can consider a coding example in which a bug-fixing agent corrects faulty code while an additional quality-control agent checks and enhances the readability and performance of that fix. By stringing these together, each agent takes responsibility for a specialized part of the work, making the output more robust and reliable. Then the code is periodically exercised and run, the output is assessed against the specification, and the process is repeated to continue to improve it.

Visualization from the robotic arm and computer vision company picknik.ai/space

The Mobius model excels when:

• Repeated cycles of tasks are required, with iterative refinement and improvement over time

• Continuous processes exist, such as coding pipelines or testing loops

• The task demands dynamic judgment about task order, timing, and evaluation of results but within a semi-defined construct

A word of excitement, and caution

We’ve explored a lot of nuanced and powerful use cases for agents. Remember, they are far more than simply using a single LLM in your product—they are new workflows that are empowered by the ability to chain together complex structures of autonomous agents with existing systems to achieve a particular goal that demands some amount of elevated reasoning at multiple points in the system. They are often forward-deployed to do the work, and they excel when used in concert, or where a modicum of human judgment is required.

As a result of these advances, this is the most exciting time for product since the invention of the internet. However, AI agents are just that—they are AIs trained on what humans before them have done, and some of those humans were not the most trustworthy creatures. Agentic systems carry the promise of massive productivity gains, but they also carry the near guarantee of accidental information leaks, wild execution of decisions without a human in the loop, data corruption, or truly just about everything envisioned in every dystopian robot movie. So please proceed with caution, and do invest in protecting sensitive data, people, and critical infrastructure as a first-class citizen when experimenting with these amazing creations.

🚨 AI vs. Earthquakes: What OpenAI's Deep Research Predicts for Greece's 380+ earthquakes near Santorini the past days - An area in Greece has increased seismic activity (380!) in recent days, so I d…

Read more

Turning a generative AI feature into a revenue stream involves choosing a model that matches your product’s value, your audience, and your business goals. Here’s a rundown of top pricing models, exam…

Read more

AI Products at Meta connect & what this means for PMs.

Finally seeing AI’s true potential, and how AI PMs should think about putting together experiences now that AI's capabilities & hardware are robust enough to support our vision across companies like Google, Meta, NVIDIA, and so on.

Key Takeaways:

✨ Big focus on multimodal. This is more than just using speech as an input to AI. It’s about all forms of human interaction—whether it’s speech, vision, or text— coming together in interactions with AI, in the same way we do with each other, but better - like editing an image via text or live audio to audio translation. Great to see many interactions across modalities enabled: voice-to-image, image-to-enhanced image, language-to-language, and even emerging interfaces like wrist-based neural devices. Meta’s Llama AI models also get multimodal and a stack but that warrants a whole new post.

✨ Full holographic glasses 🔥— this is a dream come true for anyone who’s worked in this space. Years ago, I was working on AI experiences for glasses ages ago, and watching this today is the wild future I envisioned.

✨ Great launches like: memory on smart glasses (I find that memory is often overlooked— as it's so complex to get right. I actually explored memory in the past and think it's a breakthrough for human-like interactions) or new Avatar styles and photorealistic rooms that was fantastic to see finally coming to life.

Thoughts for you: This shift towards multimodal AI is opening up a world of possibilities, regardless of surface. It means you can now design experiences and AGENTS where technology interacts as seamlessly and naturally as humans for richer & emotionally impactful experiences. Remember, AI itself isn’t the product—the experience is. When we talk about 'user impact,' we’re talking about creating real emotional connections, just like the awe people demonstrate in this video, that's where the true magic of AI lies.

I'm soon hosting a free lesson on creating your own AI agent that covers some of the above: https://lnkd.in/g66zKbve.

AI PM Jobs & Discount to Grace Hopper Conference

Read more

𝗧𝗵𝗲 𝗔𝗜 𝗣𝗿𝗼𝗱𝘂𝗰𝘁 𝗗𝗲𝘃𝗲𝗹𝗼𝗽𝗺𝗲𝗻𝘁 𝗟𝗶𝗳𝗲𝗰𝘆𝗰𝗹𝗲

Building an AI product is more than just training a model and somehow rolling it out—it’s about integrating it into an experience that has a (validated) purpose.

I kick off my AI PM 101 course with the AI Product Development Lifecycle concept, here’s an (overly simplified) breakdown of this process for my followers:

✨ 1. Ideation:

It all starts with mapping out AI’s unique capabilities to the right problem. What pain points can AI uniquely solve? Use AI where it adds the most value, and avoid using it for problems that simpler solutions can handle.

✨ 2. Opportunity:

Assess the market and determine if the idea is an experience your target persona & market need, and whether it supports a scalable business model. This is where you evaluate not just the experience, but also how AI’s ability to personalize, predict, or automate can create differentiation. Also consider AI’s infrastructure demands—will you need specialized hardware, or can it scale?

✨ 3. Concept & Prototype:

Work with your xfn partners to scope out the details.. data requirements, the right model(s), architecture, experiments, even design and map out the end-to-end experience. This step also involves assessing trade-offs i.e. balancing technical feasibility vs delivering early user value.

✨ 4. Training & Development:

Model training! Train the model and integrate it onto an experience to assess whether it meets the MVQ. Think about performance & product metrics, and dogfood to validate if the experience adds immediate value from day 1.

✨ 5. Testing & Analysis:

Testing in AI involves continuous learning/evaluating model outputs. Validate against real-world data, monitor for bias or drift, and use this feedback to iterate.

✨ 6. Roll-out:

If everything looks good, launch by making the experience accessible to users. The models might require ongoing tuning and monitoring post-launch to adapt to new data so you may need to retrain or tweak the model as you gather more real-world usage data, and this might involve further iterations before finding the right product-market fit.

Join my free webinar on how to nail your next product review that will dive into the best practices on how to lead the discussion and decisions for the stages above.

The AI Product Development Lifecycle (AIPDL) © 2023 by Marily Nika is licensed under CC BY-ND 4.0

As AI weaves its way into more and more products, companies are realizing they need dedicated product leaders to guide the way. Enter the rapidly evolving role of the AI Product Manager. 

But how exactly does the AI Product Ladder change as your org climbs the career ladder?

Let's dive in, as per image above, at the most junior level, the Associate Product Manager (APM) or Product Manager (PM) (L3/4) primarily focuses on working hand-in-hand with AI/ML eng teams to shepherd the development and deployment of AI models. Key tasks include monitoring model performance, ensuring data pipelines are in place, adhering to AI governance and ethics guidelines. The scope is tactical - making sure that AI "machinery" is integrated into an AI-powered experience, for example owning the implementation of an AI chatbot for Zappos.

As you step up to the Senior Product Manager level (~L5), the role broadens. Now you're not just focused on shipping AI features, but defining the holistic AI product roadmap and strategy. You'll work xfn with design, data science, and go-to-market teams to actually identify and prioritize the most impactful AI use cases to bring to life. Evangelizing the product vision to both internal and external stakeholders becomes critical. Here, the scope expands to weaving multiple AI capabilities together into a cohesive product offering, for example launching a personalized recommendations feature for Netflix.

Move up to the Principal Product Manager or Group Product Manager level (~L6/7), and the responsibilities get meatier. You're setting the multi-year vision for where AI can take the product, and rallying the entire team around that north star. Securing buy-in and resources from executive leadership is part of the job. You'll also be more involved in defining the AI infrastructure and data strategy to enable cutting-edge innovation. The scope covers the full AI lifecycle of taking major AI initiatives from concept to market success, for example leading a discovery platform at Pfizer.

Finally, at the Product Director, VP and CPO level, the focus zooms all the way out to defining the company's overarching AI strategy and roadmap. 

• What AI-powered products and businesses should we be betting on to secure our future? 

• What investments, partnerships and acquisitions do we need to make that happen? 

Responsibilities span influencing board-level decisions, driving key strategic initiatives, and positioning the firm as an AI thought leader. The ultimate scope is leveraging AI to disrupt industries and shape the company's long-term competitive advantage, for example overseeing AI across Google’s product suite or shaping Amazon’s 10-year vision.

While the day-to-day of an AI PM may look quite different from level to level, there are common threads across the board: a passion for transformative innovation, a commitment to responsible AI practices & the ability to lead teams around an ambitious vision while keeping them motivated throughout an experimentation-first culture towards continuous iteration for long-term success.

🚨 Do you want to learn more?

JOIN MY FREE LIVE WEBINAR

SIGN UP TO MY COURSES, NEW COHORTS KICKING OFF IN 2 WEEKS:

My course, Maven’s #1 AI Product Management course & certification kicks off in 2 weeks! Advanced course, 12-week AI PM Bootcamp or signature 3-week AI PM 101 course

Note: This representation of the AI Product Management ladder reflects my personal perspective from my 10+ years in Big tech and does not correspond to any specific company or team. The structure and responsibilities can vary widely across organizations. What are your thoughts?

As AI advances more and more, I often get asked whether we can hand product development fully over to the algorithms and data models - and my answer is no.-

✨While AI enables us to automate certain analytical tasks, it is critical to keep the (product) human in the loop.
✨Purely autonomous systems can easily lose sight of the overall experience - the customer needs, they can ignore ethical considerations, and fail to account for real-world constraints.
✨As PMs, we must actively oversee the AI development process, providing ongoing guidance and supervision. I always take a "human in the loop" approach when my teams are training models. That does not mean micromanagement, that just means being aware of model training and making sure people can count on your input/direction/guidance
✨Adding to that, that cheesy phrase 'AI can't read the room' is actually true - there is a massive need to understand cultural contexts, market trends and other social cues that in all honesty AI struggles to get.
✨AI's decision making, while powerful, can miss the subtleties that (product) human intuition and experience brings to the table i.e. turning raw data into a narrative, telling a story that aligns with the market & user needs.

How can Product Managers stay in the loop?
✅ By continuously monitoring and evaluating AI recommendations
✅ By looking for potential biases
✅ By making judgment calls/trade-offs on when to rely on the tech vs. our own expertise
✅ By making sure MVQ (minimum viable quality) is reached when an experience is put together and prior to launching.

We can't treat AI like a black box - it's here to augment PMs rather than replace us. By keeping the (product) human in the loop, we benefit from AI's analytical powers while ensuring our products resonate with human values. AI may lack emotional intelligence, but combined with human creativity, empathy, and the trade-offs/judgment PMs bring, we can develop amazing solutions.

The future requires harmony between humans and machines in product management. I hope you enjoy this visual representation about the process of AI Product Management!

Would you like to support my work? Here are some ways:

• Share this post on LInkedIn

• Becoming a paid subscriber to my newsletter

• Subscribe to my YouTube channel

• Join any of my upcoming cohorts