Hi {{first_name|there}},
This week comes down to one question: does your approach to AI actually match what your company stands for? Delong Lu of Proton takes it on from inside a company built entirely on privacy, today's sponsor In Parallel shows what happens when your AI tools stay grounded in your real goals and owners instead of guessing, and we round up the events bringing the network together in person this summer. Scroll on.
Delong Lu has spent nearly a decade at Proton, growing from one of its earliest product managers into Chief of Staff. He describes the role as a "loose radical on the management team," and that seat is exactly why he ended up leading how a privacy-first company adopts AI.
Ask Delong Lu what he does as Chief of Staff at Proton and you get a unique response. He calls himself a “loose radical” on the management team. He sits on a roughly ten-person senior team reporting to founder and CEO Andy, holds a primary focus area with real targets, and otherwise roams the organization, absorbing context and pulling teams back into line when they drift. The role is not fixed. His focus has changed multiple times in three years.
"Having had a long track record and context of the organization, I can see smokes or fires or potential areas for leverage and opportunity from a distance, and then I can go book calls to start pulling at the thread."
That freedom only works because his principal genuinely backs him, and it is also what makes him effective. Because he roams, he sees the gaps a functional leader, heads down in one unit, would miss. During a recent quarterly review, he caught one. The CEO had a clear vision for an AI-enabled future, but the work happening on the ground did not match it. Most people would let that slide. Delong surfaced the disconnect and turned closing it into one of his active work streams.
That work runs on two fronts, and both are instructive. The first is the product. Proton built its AI assistant, Lumo, on the same promise as everything else it ships: it runs on Proton's own infrastructure with zero-knowledge encryption, so the company cannot read what you tell it. That architecture is what lets a customer bring a genuinely sensitive question, about their health or their finances, to an assistant at all. The second front is harder, and it is where Delong actually spends his time: how do you let 600 people use AI internally without quietly eroding the thing customers pay you to protect?
His answer is not a tool. It is a principle.
"AI right now is a tool. You are the software engineer, and you are still the person responsible and accountable for whatever comes out of this."
In practice that means an engineer has to understand and defend every line, nothing reaches production without human oversight, and no customer data goes into outside tools. Delong is clear about the failure mode the whole industry is racing toward. AI can produce code ten times faster, but somebody still has to review it, and "who is actually reviewing this" is the question most companies have not answered yet. Proton would rather move deliberately than ship work nobody can stand behind.
Other Chiefs of Staff should steal his framing here. He does not treat AI adoption as a tooling decision, he treats it as a question of what the company already believes.
"If we really were to even try and force AI at all costs, we don't care, I think half the company will probably walk."
Culture eats strategy for breakfast. At Proton the policies flow from values, not the reverse, which is why his advice for Chiefs of Staff at less settled companies is to start with the values conversation, not the software. So the harder question is the one Delong keeps asking: when your principal's vision and your organization's reality have quietly come apart, who is actually responsible for noticing?
Walk in already knowing
You know this already — chiefs of staff like you don’t win by getting more time with execs — they win by surfacing the right risks and decisions early. “How is my project doing?” is an increasingly tough question.
Think of In Parallel as your newest, and strongest team member. It listens to your existing meetings and tools and keeps goals, KPIs, and action owners continuously up to date.
Ask Claude, Copilot, or ChatGPT — and with In Parallel, the AI answers are instantly grounded in your reality, every time. No chasing updates. No new tool to adopt. And no more split realities across the team.
New — the In Parallel MCP. Ask your AI “How are my projects doing?” and get an answer straight from your live plan, not your inbox. Now inside Claude, Copilot, and ChatGPT. Same question, same model — a grounded answer instead of a guess.
SOC 2 certified — and it makes the AI you already use 10x more relevant.
Don’t take our word for it — ask your own AI.
Paste this into Claude, Copilot, or ChatGPT (or open in-parallel.com/agents). It already knows your role, calendar, and inbox — so it can tell you what In Parallel would mean for you:
What would In Parallel mean for me and my team? Take a look at https://www.in-parallel.com/agents and tell me honestly — including the case against.
👉 Get in touch: Book a 30-min meeting
👉 Learn more: In Parallel MCP
📆 Upcoming Events
📍 This Week
Next 2 Weeks
June & July
|
Stop Surviving the Role. Start Leading It.
Most Chiefs of Staff learn the role by surviving it. There's a better way.
The CoS Launchpad is a 6-week certification program built for people who want to operate at the highest level from day one. You'll learn the frameworks, build the instincts, and get the credentials that signal you're serious about the role.
Whether you're stepping into your first CoS position or looking to sharpen what you've built over years in the seat, Launchpad gives you the structure most CoS never get.
Not ready for a cohort-led program? Try Chief of Staff Fundamentals first!
That's the week. If the Delong conversation lands with someone on your team wrestling with AI, forward it their way.
best,
Scott
Founder, Chief of Staff Network & BizOps Network
