Thomas Tobin

A couple of times I’ve tried to get sparklines in a product, where a small way of getting historical context makes sense. Then an a PM you get arguments over charting libraries, developer time, and get a compromise. Now, the PM drives the agent until it’s done.

I am low key convinced MCP was a plot to drive up token usage in LLMs by 50x

@thuddwhirr The bridge itself isn’t in green… just SF and Marin.

Waymo'ing to sonoma was not something I'd have expected this decade...

@mrp @medinism @paid_ai @jasparcjack @Artisan @manojganapathy @WorldBizBuz @ifs @lightspeedvp What did they say?

AI agents replace both software & labor. How will you price it? @medinism @paid_ai @jasparcjack @Artisan @manojganapathy @paid_ai @WorldBizBuz @ifs

@levie Do you think this ends the “scientist pm” who tests and moves in small experiments to optimize their way to success? Can you succeed like that faced with the many possible variables in how to do a job or task plus what to deliver as context and the ways this could impact the AI?

When doing product management for AI agents, the biggest shift is that the user you think about making successful is the agent. They are effectively your new “customer”. And the whole game is to figure out how to get the agent the most relevant context possible to execute their task, which is why context engineering is such a big deal. By default, AI agents know absolutely nothing about the work that they're supposed to do. You have to take this raw form of intelligence and mold it directly to the workflow that you're trying to automate. Imagine telling a super smart person (with no domain expertise by default) that just joined your company that you want them to review a contract in line with all your other contracts, or write code for a new product, or come up with a medical diagnosis about a patient, or estimate risk about a particular initiative, etc.  That new person has near infinite permutations of things they *could* go off and do to solve that task. That person would likely need instructions about how to perform the task, they would need a clear set of goals and “rules” that you expect the output for, and then they would need to talk to colleagues, review internal documentation, review past work in a similar area, and so on. Well, the issue is the same for the AI agent. Except they're at a huge disadvantage: they can't "easily" talk to colleagues, they often don't have access to the same resources a person does by default, and they by default don't know which part of their knowledge to tap into for a task.  But most importantly, they have an inherent limit of the information they can consume in each step of the process (size of the context window). And when they get too much data they perform worse at the task (context rot).  Entire companies will win or lose based on their ability to get the *right* context to the agent. Thus, AI agent PMs will require a very different type of expertise: * Deeply understand the domain that you're building agents for. In an ideal world you actually have worked in that space, but if you haven't then a good chunk of your time should be studying the actual people that work in that space. Understanding every single step of their work is going to be critical. * You have to think through essentially what would a human need to know to do the task across instructions, rules, existing data sources, best practices, etc. And then figure out how you actually get this data to the model (context engineering). * Figure out what end-user UX and features are needed to supply the agent with the right context to perform the task. Things like task queues, connecting to data sources, how to re-prompt the user during a long running task, how the user can review the output work and make modifications, and so on. * Doing evals on the AI agents with every tweak of the agent's instructions, model improvements, and any other variable that changes in the process. Especially figuring out how these changes affect the real-world customer environments is critical. Lots of work changes for engineers and PMs in a world of building AI agents and we’re just at the start of thinking through what this looks like.

@shcallaway Exposure Package™ (for your LinkedIn) Title: “Founder” (title case); Cap table status: not that. LinkedIn polish: We provide a monthly founder-thought-leadership prompt Auto-endorsements: “Product Strategy,” “Zero-to-One,” and “Founder Energy” from at least three board members.

Someone please remind me to do an April Fool’s post about hiring for a new role: “Founder” No need to start your own company when you can be a Founder (in name only of course) at mine

@shcallaway Isn't it: - 1970s: Single Terminal - 1980s: Multiple Terminals - 2000s: Terminal with 5 buttons - 2010s: Multiple flying Terminals - 2025: Single Terminal

Evolution of Computer Interfaces - 1970s: Terminal - 1980s: GUI - 2000s: Touch Screens - 2010s: AR/VR - 2025: Terminal

@thuddwhirr x.com/superjoshln?s=…

Who is responsible for this?!

@shitshowdotinfo @ESYudkowsky However, on average women do better than men at higher temperatures: journals.plos.org/plosone/articl…

Printing this out and pasting it above every thermostat at work

@simonmxtthews @DJSnM And venting something continuously?

@DJSnM i think it’s tumbling :(

And for the first time in a long time Starship is in its planned (sub)Orbit

@DefenderOfBasic @KevinF_26 You need ngrok.com to spin up the magic of a public IP and port forwarding.

@KevinF_26 Oh yes, I see what you mean! I think you're right

no seriously, why can't my mom right click on a file and have it be on the internet????

@wingod @DJSnM @timoelliott your meme is still going!

This should win science cartoon of the year award!

@petergyang Google and outlook both have this built in now: support.google.com/calendar/answe… - but some edge cases are missing like group events

What's the best alternative to Calendly?

@Kellblog Does this work the other way around? When doing European expansion do US founders and operators feel there’s no experience ?

These People Aren't Gods: How European Founders Can Stop Making The #1 Mistake in US Recruiting #recruiting buff.ly/3WLKKeg

@Kazanjy Segment by Twilio?

What were the best-timed ZIRP exits? - Divvy - Mailchimp - Chorus - SalesLoft - Drift - Slack - Afterpay Who else?

@LuizaJarovsky Can you show your prompt to it?

🚨 BREAKING: Claude AI did not pass my privacy & reputational harm test. I gave it my LinkedIn profile and asked it for my bio. It's nearly 85% wrong, including working at Google and Anthropic. To help protect against privacy harm, it should refrain from providing AI-generated bios or reports about people. Please revise.

@Kellblog Two other milestones: Google using page links as the way to rank page content: 1996 OpenAI giving up on links and using Reddit posts with >3 karma for link quality: 2019 (noted in gregoreite.com/drilling-down-…)

The assassination of the hyperlink continues Born: 1966 Popularized: 1990 (invention of WWW browser) Died: 2024 Murdered by: Twitter, LinkedIn, et alia.

@Kellblog But #goodie !

Career Development:  What It Really Means to be a Manager, Director, or VP #oldie buff.ly/2wQqu0x

@lilianedwards Reddit says this but their content is polluted with works that appear copyrighted: e.g. reddit.com/r/SongLyrics/c… . Posts are owned by the poster, so Reddit can both license use and disclaim ownership.

Opt out for copyright in large models is happening. But if you're a large player like Reddit you get to negotiate to opt in for $$. What about the small artists?

@RuneKek @ednewtonrex The other part of this is that companies don’t do this in a vacuum. Somebody is going to drive this model and claim credit and get paid. Those people still get paid, they just have more tools.

@ednewtonrex What if its a company that scales by hiring more humans?

OpenAI just released a model that can generate 1-minute videos. You simply cannot argue that these models don't / won't compete with the content they're trained on, and the human creators behind that content. What is the model trained on? Did the training data providers consent to their work being used? The total lack of info from OpenAI on this doesn't inspire confidence. Across the AI industry, people's work is being exploited without consent to build products that compete with that work. This must be controlled by regulators.