AI is Carbs

I need an AI detox. I’ve been noisy about the fact that I’m working too much – usually by the time I get to this point, I can actually ramp it back, but this time around, the non-stop work cycle has been longer and harder to stop – and I blame AI:

• You CAN do everything all at once (and it doesn’t mean you should). The fact that stuff you had to do painstakingly over many hours is now just half an hour with AI and then repeatedly hitting ‘enter’ for 45 minutes is both amazing, and basically addictive. Leads to the temptation of “just doing this one last thing” and the next thing you know, it’s 2AM and you’ve done 17 “one last things.” This is bad.
• Human context window, not AI’s is the problem: I am routinely doing five or six things concurrently. Which is neat, but I think it’s also probably one reason why I committed .env this week. I’m talking about this stuff because I want folks to do better as they ramp up their AI use! So here are some suggestions (that I will also try to do):

1. Scope by outcome, not by task category. “I’m gonna refactor isFreemium” works and is doable in one sitting because you can test your self to a stopping point (i.e. there are no more references to isFreemium). “I’m gonna clear as many bugs as I can” doesn’t work as well because there’s 1000 of them and no particular stopping point.
2. Navel-gaze your AI workflow. I have a script that does daily look-backs on my workload. AI is not shy about telling me that I’m trading sleep for more PRs and it’s frankly rather judgey about it. It’s good to manage yourself even (or especially?) when the missing management is staying within bounds of sustainable effort.
3. Let AI do AI, and figure out where you need to do more Human. Speccing is a human thing. PR reviews are only a human thing if you’re catching stuff that AI can’t – and figuring out how to make it catch more is a human thing.

AI is allowing us to ship whole new features in a single release. That’s real, worth doubling down on, and worth protecting by not burning ourselves out getting there.

Go forth and AI safely! – Bomee

Metrics

• Unique Buildings: 13,749 (Real time, via Admin panel!)
• Annual Run Rate: Unchaged

Development

• Bug Count (P-0 and P-1): 25, but 20 of them already have PRs, 7 of which are going out with Rel 117 this week.

This Week’s Highs and Lows

• Yesterday’s NYCA demo for the NYC Mayor’s Office of Climate and Environmental Justice (MOCEJ) and others in preparation for the launch went well! Kudos to Marc and Jon for the great prep! (JMB)
• Successful NYCA in-person pre-launch meeting and demo. We presented key features including scope builder, adding buildings, creating projects, creating RFPs, sending invitations, receiving and reviewing bids, and responding to bids and setting up profiles as a service provider. It was quick and without hiccups - thanks to a LOT of hard work behind the scenes to make sure the flow was smooth and without bugs. MOCEJ is excited and working to nail down the date for the public launch in late March or early April. (JB)
• We’re running yellow lights ahead of NYCA’s launch. Yes, the releases are a bit bumpy, but they’re loaded with good stuff - both features and system hardening. (FRH)
• NYSERDA seems finally ready to expand use of the Building Data Platform internally. It took a lot of work under Robin’s leadership to establish trust, but we’re getting there! (FH)

Any missing highlights? Please share in Discord comments.

Crow’s Nest

Looking out for icebergs: What are the risks on the horizon that we’re watching for and navigating around?

Jason’s List

Repeats, with tweaks:

• Improving clarity around our roadmap and prioritization process–I’ve started to meet with a PM coach to help me. But I’m back in the cycle of dealing with near-term stuff, so it may be a while before I make a lot more progress on this.
• We may have won some minds, and we’re winning some wallets, but will we win their fingers? Which users will actually use our software instead of just looking for the outputs and expecting consultant-style service?
• What is our moat in a world and industry where data wants to be free, especially when some of our largest customers are city and state agencies?
• Honing our delivery process AI is helping us write new code quickly, but we haven’t unblocked the whole flow. We need to automate more of the testing to keep quality and velocity up, while keeping in mind that we are humans who need rest.

François’ List

Repeats:

• To get ready for NYCA launch, we need to project ourselves to a near future were a lot of variables are multiplied by 10-100x. Max number of concurrent users. Size of database backups. Occurrences of Sentry issues. Probability of a malicious actor trying to make us look bad. Let’s be creatively paranoid and prepare the ship!
• We have a lot of “almost there” large features or complex clean-ups that need to land soon: EPA Portfolio Manager v2, new BKB supporting buildings beyond Multi-Family (shipped!), Building Attribute Guesser for NYSERDA BDP (shipped!), improved tariff support, etc.

Bomee’s List

Jon’s List

• NYCA Willdan support We continue to work with the Willdan team in getting ready to provide NYCA users with hands-on case management. Their team seems engaged and smart but not super experienced. How can we support them to ensure the human side of the program is a success?

On Deck for Next Week

• We’re going to be giving the Department of Buildings’s (DOB) Sustainability Bureau a Momentum tour on Wednesday. (JMB)

Upcoming Releases

Release 118

• Program registration and onboarding flows going live!
• Service provider profiles
• Surfacing LL97 pathway in the UI
• Hopefully surface rebate explanation in the UI as well
• Security, stability, and speed improvements (the triple S!)

Please Leave Feedback

Please note your reaction to this update in the Discord channel. It helps us to know what is resonating, what is unclear, etc. Thanks!

• What are your highlights / low-lights?
• Did we miss a highlight? Something else you want to react to?