Product marketer for AI and developer platforms: positioning, launches, enablement. Ex-Amazon, Fandom. 12 books on APIs and AI. Posts on PMM and dev tools.
San Francisco Bay Area, CA
Joined September 2025
- Tweets89
- Following34
- Followers2
- Likes4
Pinned Tweet
The three questions I ask before I touch any messaging on a technical product. (Spoiler: none of them are about the headline.) 🧵 1/5
I keep a running doc of the exact phrases customers use to describe their problems. Word for word, typos and all. It's not glamorous, but some of my best headlines started life as a frustrated sentence in a support ticket.
What can you say that a competitor can't say with a straight face? If that list is empty, that's not a marketing problem. That's a roadmap conversation, and probably an awkward one. 4/5
Quick positioning test: could a competitor paste your headline onto their homepage without changing a word? If so, congrats, you wrote a category description.
1. The per-employee performance here is wild.
2. Anthropic is going to be the blueprint for AI orgs going public in the next 12 months.
3. The "AI Bubble" fear is real - and this is going to reinforce the naysayers as well as the fans.
So are you going to invest in #ANT?
#ai
Anthropic has filed an S-1 registration statement with the SEC related to its proposed IPO.
Read more: anthropic.com/news/s1-sec
Legal disclaimer: A registration statement relating to these securities has been filed with the SEC but has not yet become effective. These securities may not be sold, nor may offers to buy be accepted, prior to the time the registration statement becomes effective. This post does not constitute an offer to sell or the solicitation of an offer to buy these securities, nor shall there be any sale of these securities in any state or jurisdiction in which such offer, solicitation or sale would be unlawful prior to registration or qualification under the securities laws of any such state or jurisdiction.
AI hot take: it makes you 10x across the board.
That means 10x faster. But also if you're bad at something, that means 10x more drift, 10x more frustration, 10x worse outcomes...10x faster.
#ArtificialIntelligence #IndustrialAutomation
What does it take to turn an AI prompt into a working industrial application? This #webinar features a live walkthrough of building a custom production monitoring application while managing permissions, deployment controls and shared standards. brnw.ch/21x6raF
Later today, I'm talking through what I think is the most interesting current use case for AI - industrial application building.
If you want to hear about the new industrial revolution - and all of its trappings - you have to come to this webinar!
onlinexperiences.com/scripts…
The key to effective product marketing is honesty. Really, that's it.
Also figuring out SEO, AEO, GEO, comparative analytics, competitor positioning, narrative strength, and a billion other variables.
But mostly honesty.
#productmarketing #marketing
An LLM audits a production flows.json and finds 16 crimes. Claude cleans up messy MQTT. Real-time OEE runs on a tea bag line.
Node-RED Con 2026: 12 talks from people running Node-RED in production. Free and online, Nov 3.
nrcon.nodered.org/
#nodered #industrialautomation
Industrial AI can't just be generalised AI - it needs to be built for industry. I legitimately think FlowFuse has one of the best solutions in this space (flowfuse.com/ai/) - and it's leagues better than just asking ChatGPT to build a ModBus bridge. #IndustrialAutomation
I've written 750+ articles and whitepapers and 12 books on API and application security. The same 7 mistakes keep popping up again and again.
Let's talk about them. 🧵
Mistake #7: Stop assuming every caller is human. More and more traffic is coming from agents. If your API can't tell user from agent, your audit trail and rate limits are worthless. Start paying attention - and dealing with it properly.
Mistake #6: If you don't know what APIs you are actually running, you're asking for trouble. Keep an inventory and actually pay attention. It can be a spreadsheet, but that's pretty silly - make an API directory and keep it constantly updated.
Mistake #5: Stop sending the whole object and just letting the frontend filter it. That whole object contains a ridiculous amount of info that can be used for horizontal escalation. Return only what the client needs - and do the work to figure that out first.
Mistake #4: Letting your tokens run everything. Tokens that can do anything are awful and you're basically asking to be attacked. One token with full scope is easy for you - but it's easy for an attacker too. Least privilege should be table stakes at this point.