February 2026

There’s a question I’ve been sitting with lately, and I suspect a few of you have too: at what point does “experimenting with AI tools” become professional negligence if you’re not doing it?

We operate in an industry that demands simultaneous fluency in technology, policy, acquisition law, international relationships, and operational realities. The cognitive load is immense. The pace is relentless. And the administrative drag of email, calendar management, meeting prep, and tracking action items across a dozen open threads quietly consumes hours that should be going toward strategy and relationships.

I’ve spent the last several months building a deliberate workflow around Claude, Anthropic’s AI assistant. Not tinkering. Not occasional use. A genuine restructuring of how my workday runs. This article is an attempt to share what I’ve learned: the specific mechanics, the surprising capabilities, and an honest accounting of what it means for how we think about talent and team-building going forward.

AI Proficiency Is Now a Leadership Competency

Let’s start with the uncomfortable framing.

We would not hire a VP of Business Development who refused to use a CRM. We would not trust a Director of Policy who didn’t read the Federal Register. And within the next few years (months?), I’d argue we won’t take seriously a senior executive who hasn’t developed genuine fluency with AI tools. Not familiarity. Actual fluency. In the linguistic sense. 

This isn’t hyperbole. It’s pattern recognition.

Every major productivity transition in the modern era, from email to smartphones to cloud collaboration, created a window during which early adopters extracted disproportionate advantage before the technology became table stakes. We are inside that window right now with AI. The defense tech sector, with its demanding communication requirements, proposal-heavy business development cycles, and the premium placed on executive bandwidth, is particularly well-positioned to benefit.

But fluency is the operative word. Fluency is different from familiarity. Familiarity is asking an AI to write a bio or summarize an article. Fluency is integrating AI deeply enough into your workflow that it fundamentally changes your throughput and the quality of your thinking.

Here’s how I’m building that fluency, and how you can too.

Note: I’m referencing Claude here because that’s what I use. I could have easily chosen ChatGPT or Gemini or something else, and you can too… they all have amazing functionality. The trick is to dive in and learn. 

The Architecture: Claude Connected to Your Actual Work

Most people interact with AI as if it’s a search engine with better grammar. They type a question, they get an answer, they move on. That’s fine for occasional tasks, but it misses the compounding value that comes from an AI that has context on your work, your priorities, and your ongoing threads.

Claude, when connected to Gmail, Google Calendar, and Notion, stops being a general-purpose tool and starts functioning as something closer to a context-aware chief of staff. The integration isn’t magic. It requires intentional setup and some initial investment in defining how you want information organized and surfaced. But once you’ve done that work, the returns are significant.

Here’s the stack I’m running and how each layer functions.

Layer One: Email Triage, Prioritization, and Drafting

Email is where executive attention goes to die. Most high-volume inboxes are a mix of 20% genuinely important communication and 80% things that can wait, be delegated, or be deleted. The problem is that the 80% is distributed randomly through your inbox, and the cognitive cost of sorting it is nearly as high as just responding to all of it.

Claude changes this equation.

Triage and prioritization. I give Claude access to my Gmail and ask for a morning digest: surface emails that require my personal attention, flag anything with deadlines or action items, and deprioritize newsletters, FYI threads, and anything where I’m CC’d but not a primary actor. The output is a ranked summary, not an inbox reorganization. I’m still in my inbox. I now just know where to start.

The key instruction is being specific about what counts as priority. For my work, that means: direct asks from government partners, time-sensitive contract and proposal-related threads, congressional and Hill staff correspondence, and anything from a short list of named relationships. Claude learns the pattern quickly and surfaces accordingly.

Drafting with context. When I need to respond to a complex email thread, whether it’s a government partner asking about a technical capability, a BD contact following up on a meeting, or a congressional staffer requesting background, I ask Claude to draft a response. I provide the thread, the goal of the response, and any constraints (tone, classification, what not to include), and Claude produces a near-final draft.

What makes this genuinely useful rather than a rough approximation is providing context upfront: your communication style, the relationship history, the strategic objective of the correspondence. When Claude has that context, drafts require light editing rather than rewriting.

I’ve started keeping a short “communication context” document in Notion that I can reference or drop into a prompt: my preferred tone, standing instructions for certain relationship categories, and recurring framing for common correspondence types. This reduces the setup friction on each drafting task considerably.

The discipline of the prompt. The quality of email drafts scales directly with the specificity of your prompt. “Draft a response” produces mediocre output. “Draft a professional response that acknowledges their timeline concern, affirms our commitment to the Q2 milestone, and redirects the conversation toward the technical review meeting next week, three short paragraphs” produces something I can send with minimal revision. That specificity takes thirty seconds. The time savings are real.

Layer Two: Calendar Synthesis, Scheduling Intelligence, and Meeting Prep

Calendar integration is underutilized and underappreciated.

The average executive’s calendar is a data source, not just a schedule. Who you’re meeting with, how often, what’s clustered in a given week, what’s been rescheduled, what’s coming up that requires preparation: all of that is meaningful information that most of us process manually, inefficiently, and often not at all.

Weekly and daily synthesis. Each morning, I ask Claude to pull my calendar for the day and the week and give me a structured brief: what meetings are happening, what preparation is needed for each, what conflicts or back-to-back issues exist, and what blocks are protected for focused work. This takes approximately thirty seconds and gives me a clear mental map before I’ve opened an email.

For high-priority meetings, I’ll ask Claude to go further: pull the relevant Notion notes from the last interaction with that organization, surface any open action items, and generate a quick pre-read. It won’t always be perfect, but it is far better than walking into a meeting having only glanced at the invite.

Scheduling assistance. Claude can analyze multiple calendars for free time windows, draft scheduling emails, and help with the coordination overhead of complex multi-stakeholder meetings. For anyone who manages a high volume of meeting requests across government, industry, and investor audiences, this is material time savings. 

An important note here: Claude isn’t that good in this regard and still struggles with coordination across multiple calendars. I largely do this the old-fashioned way, but these tools are getting better and better. 

The deeper discipline here is maintaining good calendar hygiene so the AI has useful data to work with. Vague calendar entries like “call” or “meeting” are data deserts. Entries that include the organization, the purpose, and any relevant prep notes create a rich record that Claude can use to generate useful synthesis.

Layer Three: Notion as a Connected Knowledge System

Notion is where my institutional knowledge lives: meeting notes, project tracking, proposal status, relationship logs, contact context, and research archives. I’ve pretty much given up on other project apps (Asana, etc) given the flexibility and the one-stop-shop-ness of Notion. The challenge with any knowledge management system is that the value is only as good as the retrieval, and retrieval is where most systems break down.

Claude connected to Notion fundamentally changes the value proposition. But to get there, the underlying database structure has to be intentional. Here is how I’ve organized the four core databases and what each one enables when Claude can access them.

The Organizations Database. This is the backbone of external relationship management. Each record is an organization: a government agency, a congressional office, a contractor, an allied nation partner, an investor, or a prospective customer. Properties include organization type, current relationship status, primary point of contact, associated opportunities, and a running notes field that captures anything relevant that doesn’t belong in a formal meeting note.

The power of this database is longitudinal context. When I’m preparing for a meeting with an agency I’ve engaged several times over eighteen months, I don’t want to reconstruct that history from memory or dig through email threads. I ask Claude to pull the organization record and summarize where the relationship stands, what was discussed in the last interaction, and what’s unresolved. That summary takes thirty seconds to generate and fifteen minutes to reconstruct manually. At volume, that math matters.

The other use case is relationship health monitoring. Claude can scan the Organizations database and flag accounts where there hasn’t been a touchpoint in 60 or 90 days, helping me stay ahead of relationships that are drifting without requiring me to audit the database myself.

The Contacts Database. Related to but distinct from Organizations, the Contacts database tracks individuals. In defense tech BD, you’re managing relationships with dozens of program managers, contracting officers, congressional staffers, think tank analysts, flag officers, and industry peers simultaneously. Most of these relationships are episodic rather than continuous, which means the context degrades quickly if you’re not capturing it.

Each contact record includes their current role, the organization they’re associated with, how we met, the last interaction date, notes about their priorities or communication preferences, and a flag for follow-up cadence. The last field is critical: some relationships warrant a monthly touchpoint, others quarterly, others only when there’s a specific reason to connect.

Claude can work across this database in ways that would take significant manual effort otherwise. Before attending a conference or industry event, I’ll ask Claude to pull all contacts associated with attending organizations and surface anyone I should prioritize reconnecting with, along with context on where we left off. That preparation used to take an hour. It now takes a few minutes.

The Meetings Database. Every substantive meeting gets a Notion record: the date, attendees, organization, meeting type (introductory, technical deep-dive, debrief, proposal review), a summary of what was discussed, and a clearly labeled action items section. That last element matters. Action items need to be consistently structured so Claude can retrieve and synthesize them reliably. A notes field full of prose is harder to query than a dedicated action items property with owner and due date.

The Meetings database is where the daily and weekly synthesis really comes to life. Claude can pull all meetings from the past two weeks, extract open action items across them, and give me a consolidated list organized by owner or due date. What was previously a manual review of a dozen individual meeting notes becomes a thirty-second prompt.

It also enables after-action analysis that most executives never have time to do. “Across my meetings with the Coast Guard this year, what commitments have we made and which ones have we followed through on?” or “What themes are coming up repeatedly in customer conversations that might be worth addressing in our positioning?” These questions are answerable when the meeting data is structured and accessible.

The Projects Database. This is the operational tracker for active work: proposals, contracts, government programs, internal initiatives, and partnership negotiations. Each record includes the project name, status, associated organization, key milestones, the last update, and a current blockers field.

The Projects database is where Claude earns its keep most visibly on a daily basis. When I ask for my morning summary, I want it to include a status sweep across active projects: what’s due this week, what’s blocked, and what needs my personal attention. Without a structured Projects database, that synthesis requires manually checking multiple trackers. With one, it’s a single prompt.

The deeper value is cross-database synthesis. A project record links to an organization, which links to meetings, which links to contacts. Claude can traverse those relationships and answer questions like: “What’s the current status of the Project X proposal, who are the key stakeholders, and what was discussed at the last meeting with that agency?” That kind of connected recall is what transforms Notion from a filing cabinet into something that actually functions like institutional memory.

Building for retrieval, not just storage. A few structural principles make all four databases work better with Claude. Use consistent property names across databases so relationships are queryable. Write a brief one or two sentence summary at the top of every record so Claude can get the gist without reading every detail. Tag records with a status property that reflects current state. And link records to each other where the relationship is real: a meeting record should link to the organization and the relevant project, not just reference them in prose.

The investment in structure pays dividends proportional to the volume of your work. For a defense tech executive managing multiple active opportunities, a bunch ongoing agency relationships, and a pipeline that requires constant tracking, a well-structured Notion system becomes a genuine force multiplier.

Active recall rather than passive storage. With this structure in place, instead of navigating through nested pages to find a specific decision or conversation thread, I can ask: “What did we discuss with [agency] at our last meeting, and what were the action items?” or “What’s the current status of the [project name] proposal and what’s outstanding?” Claude surfaces the answer from the notes rather than requiring me to find the note.

This changes how I use the system. When I know I can retrieve information conversationally, I invest more in capturing it. The relationship between input and output becomes reinforcing rather than extractive.

Proposal and document scaffolding. When working on a proposal or a major document, I’ll ask Claude to pull the relevant research, prior proposals, and meeting notes from Notion and use them to develop an outline or first draft. The output isn’t submission-ready by any stretch. It never is. But having a structured starting point that incorporates institutional knowledge rather than starting from a blank page dramatically changes the pace of production.

Layer Four: Research and Analysis

Defense tech business development requires continuous research: understanding agency priorities, tracking legislative developments, synthesizing open-source technical developments, monitoring competitor positioning. This is knowledge work that doesn’t fit neatly into meetings or proposals but directly shapes strategic decisions.

Claude is a genuine research partner for this kind of work.

Synthesis of complex documents. Budget J-books, RFPs, congressional testimony, GAO reports: these documents are dense, long, and require careful reading to extract relevant information. Most people procrastinate review.I give Claude the document and a specific question: “What does this RFP’s technical evaluation criteria emphasize most, and what are the disqualifying factors?” or “How does this agency’s stated priorities in the FY26 budget request compare to their stated priorities last year?” The answer takes seconds and is usually actionable.

Competitive and market intelligence. I have to track a complex competitive landscape, and Claude can synthesize publicly available information across sources, compare positioning, and flag developments worth watching. The key discipline is treating outputs as starting points for your own judgment, not as conclusions. Claude is very good at gathering and structuring information; the strategic interpretation still belongs to you.

Briefing preparation. Before a Hill meeting, a customer meeting, or an industry conference, I’ll ask Claude to prepare a briefing note: who I’m meeting with, their recent public statements or priorities, likely areas of interest or concern, and suggested talking points. This takes five minutes of setup and produces a brief I would previously have spent an hour preparing manually.

Layer Five: The Daily Operating Rhythm

This is the integration layer, where the individual capabilities compound into a genuine workflow change.

Morning summary. Each morning, I start with a single prompt that requests a synthesized brief: the most important unread emails that need my attention today, the day’s calendar with prep notes, any Notion action items that are overdue or due today, and anything from active proposals or projects that I should be aware of. The output is a one-page operational picture of my day before I’ve done anything reactive.

This is not about delegation. I still read every (important) email. I still engage with my calendar. But I start from a position of orientation rather than triage. The difference in morning cognitive load is significant.

End of day review. The bookend to the morning brief is an end-of-day summary: a review of what was scheduled versus what happened, any emails that came in during the day that still need responses, open action items from today’s meetings, and a forward look at tomorrow. This takes three to five minutes and replaces the scattered mental accounting most of us do while pretending to wind down.

The end-of-day review serves a second function: it surfaces things I missed. In a high-volume day, emails get buried. Action items slip. The AI review catches what the human missed and creates a clean handoff to the next day. 

This saves me nearly every day. 

Weekly synthesis. On Fridays, I run a broader synthesis: how did the week map against the stated priorities at the start of the week, what relationships need follow-up, what BD threads advanced and which stalled, and what does next week look like. This is strategy-level reflection enabled by operational data, a combination that’s historically been impossible without a dedicated chief of staff to prepare it.

The Technical Setup: Getting Started

For executives who haven’t yet built this stack, here’s the practical path.

The capabilities described in this article are available through Claude’s integration with Gmail, Google Calendar, and Notion, accessible through the Claude.ai interface. The setup requires connecting your accounts through Claude’s settings. This is a few minutes of work, not a technical project.

The more significant investment is in prompt development. The first few weeks of using any AI workflow system require iteration: refining what you ask for, how you ask for it, and what context you provide upfront. Maintain a running document of prompts that work well. Revisit and sharpen them. Share them with colleagues who are building similar workflows.

The upfront investment is real. The compounding returns are significant.

A few practical disciplines that accelerate the value:

Maintain a personal context document that you can drop into any prompt, covering your role, current priorities, key relationships, communication preferences, and standing instructions. This eliminates repetitive setup and dramatically improves first-draft quality on anything Claude produces for you.

Build your Notion structure with retrieval in mind, not just storage. Consistent naming conventions, clear tagging, and brief summary notes at the top of each page make AI retrieval far more reliable.

Develop specific prompts for recurring task types. Proposal drafting, congressional correspondence, meeting prep briefings, and debrief notes all have recurring structures. Templatizing your prompts for these task types is a one-time investment that pays dividends every time.

The Human Dimension: What This Actually Means

I want to be direct about something that often gets papered over in these discussions.

I have been actively reconsidering whether to hire an executive assistant.

That is not a casual statement. The EA role has historically been one of the highest-leverage investments a busy executive can make, a force multiplier that extends your capacity to communicate, coordinate, and operate at scale. For much of my career, it was simply assumed that at a certain level of responsibility, you needed that support. I’m there. Or I was there. Maybe.

Now I am no longer sure that’s true, at least not in the same form.

The tasks I would hire an EA to handle – inbox management, calendar coordination, meeting preparation, follow-up tracking, research synthesis, first-draft correspondence – are tasks that Claude handles with reasonable competence, improving as I invest in context and prompt quality. The remaining gap between AI output and EA output is real, but it’s narrowing, and the economics are dramatically different.

I want to be careful here, because this is territory where the conversation can become reductive quickly.

There is something an EA provides that Claude does not: the fully human quality of judgment, intuition, and relationship navigation that develops over time. A great EA understands not just what you need, but why. They carry institutional knowledge in a form that’s alive rather than archived. They can read a room and make a call in ways that no AI currently replicates. These are real and meaningful capabilities.

And yet.

For a startup executive in a resource-constrained environment, operating with a headcount budget that demands prioritization, the honest accounting is this: Claude covers enough of the EA functional envelope that the incremental value of a human hire, at current capability levels, is harder to justify than it was two years ago. Or even two months ago. 

I don’t say this with satisfaction. I say it because the people in this industry who are not doing this math are going to be at a disadvantage against those who are, and because intellectual honesty requires acknowledging that AI is not just automating tasks at the margins. It is moving into roles that we previously considered irreducibly human. This is real-time economic factors happening now. To me. Today. 

The question this raises for leaders is not whether AI will change the nature of knowledge work. That is already happening. The question is what we do with the capacity it creates. My answer is to reinvest it. The hours not spent on administrative overhead go into relationship development, strategic thinking, and the kind of high-judgment work that cannot be delegated to a machine. That is the appropriate response to a productivity expansion: not reduction, but redeployment toward higher-value activity, which will result in growth and even more hiring for tasks that AI cannot perform – relationship development, synthesis, strategy, and management. 

But we should be honest about the redeployment imperative. “AI won’t replace jobs, it will change them” is only true if the human invests in changing alongside it. For executives, that means developing the AI fluency to extract the full value of these tools and then doing the harder work of deciding what only a human can do, and doing it at a higher level.

The Compounding Effect

Here is what I’ve noticed over months of deliberate AI workflow integration: the value compounds.

In the early weeks, the gains are obvious but modest. You save time on email drafting. You get better meeting prep. You retrieve information faster. These are real improvements, but they feel incremental.

Over a few months, something different happens. The AI develops a richer model of your work and priorities because you’ve given it more context. Your prompts become more precise because you’ve learned what works. The workflows you’ve built start connecting: morning summaries inform calendar prioritization, Notion captures from meetings surface in proposal drafting, email context enriches briefing prep. The system starts operating at a level that feels qualitatively different from any individual capability.

This is why the executives who invest early will have a structural advantage over those who wait, which is likely due to the excuse “I’m too busy.”

The compounding doesn’t just accrue in productivity. It accrues in institutional knowledge management, in the quality of strategic communication, in the consistency of follow-through. These are the operational disciplines that separate high-performing organizations from mediocre ones, and AI-augmented workflow is becoming a genuine lever on each of them.

The window for disproportionate advantage is open. The only question is whether you’re building the fluency to use it.