
Driving Change Every Day | Previously @Blackbaud + @Columbia faculty | Bestselling author of #DataDrivenNonprofits | Indiana University alum
Charleston, SC
Joined April 2008
- Tweets45K
- Following2.4K
- Followers7.7K
- Likes1.7K
Pinned Tweet
Wow! #1 Best Seller in Nonprofit Organizations & Charities on Amazon #DataDrivenNonprofits buff.ly/2cqvyuq
Offload high effort / low value tasks. Use the cognitive surplus to spend time on high value work.
The thing I avidly believe:
You can offload anything you want to AI. For example if you are writing a blog you can have it help you with research, do all the research, or write the entire thing.
Be VERY conscious about what you are offloading. Offload the things that let YOU get better/smarter/faster/more experienced. Don't offload the things you want to become, because if you do you will never become them.
If you offload writing your blog, it isn't your blog. It won't have your ideas, your perspective, your voice.
91
Prototypes are a concept car. Not a production car that hits the road. Very valuable for testing and validation.
i'm hearing this over and over again.
PM's are creating prototypes, everyone gets excited and wants to ship
design either:
1. becomes a blocker bc they push back
2. becomes clean up crew after being handed PM's prototype to tidy up
what advice would you give?
1
82
Good advice here. Design systems connected to AI can help others prototype. Design still leads on the big jobs. It’s a team effort.
I'm the maker of my own misery here because I built the tool our PMs use to prototype... here's what's helped me regain some sanity:
1. Do not become perceived as a blocker. As soon as this happens design is DoA. It sucks, but do the clean up and cover what's needed to feed the machine, get ahead, and get in front of the next situation.
2. Differentiate between tiers of product design work. For small scope updates, I want PMs to prototype and move ahead without much support from me most of the time. I want to protect design's time to focus on the biggest, most ambiguous projects where our skills move the needle the most (these are also most engaging for designers).
3. Do detailed critiques of UX gaps in any prototype I receive. Seeing a real UX teardown of a PM prototype quickly highlights deficiencies. Hold it to the quality bar you'd hold a designer to.
4. Show my work. People underestimate how much goes into most designs. Make it impossible to miss how the team's iteration informed a better output.
5. Inject myself earlier in the process. On my team, we have an initial product shaping stage where the PM writes an "Intent Doc." I must be aware early so I can influence that thinking, advocate for the right level of design involvement and jump in before things progress too far.
6. Make noise. There's a difference between speaking up and being a blocker. To do this most effectively, you need your team's trust and good relationships which is usually based in helping the team with #1.
7. Build better automated tools, processes, and guidelines. Unlike design systems of the past, these can now be operationalized automatically using AI. PMs don't know what they don't know about design. So the tools you give them have to account for this. Build your designers' knowledge into systems that make it hard for PMs to create artifacts that miss your quality bar.
If you're looking to scaffold your own AI prototyping environment, try the open-source one I built based on what I created for my team: itspatmorgan.com/design-stud…
github.com/itspatmorgan/desi…
1
80
Velocity matters on a racetrack. Navigating matters in a jungle.
A Ferrari is 10X faster than you on a racetrack, but you're infinitely faster than it crossing a jungle. The problem with the fast takeoff crowd in AI is that they think the world is a racetrack, when in fact it's more like a jungle.
73
“The gap between “works in a demo” and “everyone relies on it” is where most of the engineering happens.”
Somebody at your company vibe coded an internal tool six months ago that everyone now uses, and I'd bet they HATE it by now.
It goes the same way almost every time. The first version takes a day and gets things right about 80% of the time, which feels like magic.
Then people start using it for real work, and 80% stops being good enough.
They expect it to be right every time, and the person who built it is on the hook for every wrong answer.
Say the tool replaced a $1,000-a-month subscription. The builder now spends a day a week keeping it alive and costs the company $10K a month. That's about $2,000 of their time to save $1,000, before you count the day it breaks during month-end.
The gap between “works in a demo” and “everyone relies on it” is where most of the engineering happens.
To give you a sense of scale, on a patient-billing voice agent we built for a US healthtech startup, the test code is almost as big as the application: 343,080 lines of tests for 349,042 lines of code. We run 13,650 tests on every PR. The safeguards are in the code too: an identity check before a balance gets discussed, quiet hours enforced at send time, disputes routed to staff.
Nobody enjoys building that part, but it's what allowed the system to handle 61,246 calls.
I wouldn't ban vibe-coded tools. I'd give them a rule: once a tool has daily users, it has to graduate:
1. It gets an owner whose actual job includes looking after it, instead of somebody fixing it at night.
2. It gets a written spec of what “correct” means, and tests that check it.
3. The old way stays available until the new one matches it on real data.
4. Or you decide it isn't worth maintaining and go back to buying it.
Option 4 is a perfectly good outcome. Plenty of tools should have stayed a subscription.
1
118
Winning. Again. @IUHoosiers
73
Hoosiers do it again. @IUHoosiers
100
“What matters is consistency across teams, and across a whole company, and that is the human part of a Design System.”
What AI has changed in Design Systems, and what is still up to us uxdesign.cc/what-ai-has-chan…
1
90
Super Subs for @CelticFC
107
20th Big Ten Title!!! @IUHoosiers
68
“But no matter how robust a design system is, it can't take into account all the potential combinations of layout and functionality a software application needs to work.” — @LukeW
no plan survives contact with the battlefield and no design system survives contact with the product.
new article on how AI agents can fill the gap between design systems & product UI: lukew.com/ff/2170/design-fee…
Grande Kimi! @MercedesAMGF1
88
Home win. #HereWeGo
101
Homecoming Hoosier Hero! @IUHoosiers
18
Bravo! @LewisHamilton @ScuderiaFerrari
77
“How you do something is how you do everything.” — @CCignettiIU
Curt Cignetti talking about why he values academics when evaluating prospects. Good stuff here for anyone who wants to play college sports. Via IU Athletics. #iufb
90




























