Date: 2026-08-24 Time: 2:00 PM EDT Attendees: Blake Stewart (Founding Journey, note creator), Blake Emal (Megaphone Studio), Brentan Alexander (Roebling) Client: brentan-alexander (Roebling)
Summary
Content interview covering customer trust dynamics, process engineer risk aversion and tooling philosophy, AI confusion among enterprise clients (BP reorg), data security friction, fundraise context, and the white paper thesis. Brentan is traveling the week of Aug 31; next sync async. Blake to route news alerts directly to Brentan’s channel.
Decisions
- News alerts routing: Blake to adjust settings so relevant alerts go directly to Brentan’s channel; he’ll flag anything worth drafting.
- Next week async: Brentan traveling week of Aug 31; coordinate over Slack instead of a call.
Action items
- Blake Stewart: Route news alerts directly to Brentan’s channel (not just to Blake).
- Blake Stewart: Handle next week’s sync async (Brentan traveling week of Aug 31).
- Brentan Alexander: Check white paper outline status with Joshua (Joshua taking first stab).
Notable excerpts
Customer trust / risk aversion (Brentan’s framing):
“Okay information that will definitely arrive beats good information that’s only 50/50 likely to show up… no one gets fired for hiring McKinsey… the bad information gets punted further and further down the line until it becomes really expensive.”
Process engineer triple-checking culture:
“It’s not just double-checking, it’s triple-checking… They’re looking for fat-finger errors, not software bugs… The surprise for me was that that double and triple checking of the new software feature, it never stops.”
Roebling’s design response: expose the system’s built-in checks so engineers don’t re-run them manually. Math entry renders as visual equations (superscripts, division lines) rather than code syntax, enabling faster visual QA/QC.
AI mandate confusion at BP:
“They went through a really big reorg… around getting more efficiencies out of the team and AI being an important part of that… no one actually knows what the hell that means.”
Client challenge: should Roebling match existing reporting templates or push clients to rethink deliverables for the AI era? Client explicitly asked to be challenged when legacy deliverables no longer make sense.
Data security catch-22:
“You need to answer those questions to know what data to ask for… but until you have those controls in place, they won’t let you see the data.” “Everyone asks the same questions. Will you train on my data? Everyone knows the answer to that needs to be no.”
Fundraise VC targeting:
“Many VCs don’t understand physical infrastructure… When’s the last oil refinery we built in the country? I think it was like in the ’70s… that Vogtle nuclear plant was the first nuclear plant we’d built in decades and went super over budget.”
Targeting VCs with AI + physical infrastructure thesis, or corporate venture arms. Hard tech VC history is cyclical - VCs repeatedly underestimate build timelines.
White paper thesis (the riskiest argument):
“The whole argument that the FEL stage gate process and Class 3, Class 4, Class 5 estimating is outdated and wrong and harmful… these things came out of the ’50s, compute didn’t even exist… when you relax compute and some of these other restrictions that AI in terms of augmenting human capability, the conclusions of needing to stage gate… falls apart a little bit.” “Even our internal process engineers get a little uncomfortable… they like believe it, but also they’re like, I’ve been taught the opposite.”
Status: Joshua taking first stab at outline; Brentan to follow up.
See Granola for full transcript: meeting id 34442808-8e10-433e-8c67-a7b6b0c35ed8