Captain’s Log #7: The Angela Algorithm - Build What's Yours Before They Automate It
Part 3 of 3: AI and the Productivity Paradox
This final piece is written for women.
Not because men don’t need this.
But because women are more exposed to automation, more penalized for using AI, more punished for claiming credit, and more likely to bear the cost when the system fails.
The Angela Algorithm applies to everyone.
The urgency doesn’t.
If you’re a breadwinner mother, if you’re in a role being automated first, if you’re in a workplace that punishes deviation … the stakes here aren’t theoretical. I see you.*
*Also, I see you trying to channel your circa-2008 attention span to read this whole thing. You can do it.
So….. What do we actually do about this?
The Case for Fierce Ambivalence
My incredible friend and former Harvard classmate Mara Bolis wrote a piece for Stanford Social Innovation Review that reframed something I’d been struggling to articulate.
Women use generative AI at significantly lower rates than men. The usual explanation is “risk aversion.” Mara calls that framing lazy. The real explanation: women are picking up on legitimate risks that the “just use AI more” crowd glosses over.
The research she cites is striking. Female engineers using AI-generated code get rated as less competent than male peers producing identical output. Women are nearly three times more likely to lose jobs to automation because of concentration in clerical roles. The tools hallucinate. The systems are opaque. The professional penalties for AI use fall unevenly.
These aren’t irrational fears. They’re pattern recognition.
Mara’s reframe: the response to dizzying hype shouldn’t be rejection. It should be fierce ambivalence—”passionately holding two seemingly contradictory truths at once: We should use generative AI to empower ourselves and others, AND we should demand exacting standards of transparency, fairness, and safety from those building and governing these tools.”
love that framing because it refuses the binary. Use the tools. Demand better from the people building them. And, since we all know systemic change moves at a glacial pace, what should we do in the meantime?
Use AI deliberately, skeptically, and for your own purposes.
This question—how to hold both skepticism and agency—is something Mara and I are working through in real time. We’re launching a podcast about it: Womansplaining AI. Curated research, honest talk, no consensus required. First episodes drop this month.
In the meantime, here's what I keep coming back to.
The Trap Has Two Doors
The risks Mara identifies are real. Women picking up on bias, opacity, and professional penalties aren’t being paranoid. They’re paying attention.
But here’s what’s also real:
79% of employed women work in jobs at high risk of automation (vs 58% of men). Education support. Healthcare admin. HR. Legal assistance. Customer service. Marketing coordination. The exact middle-income professional roles being automated first. For women in those roles—especially breadwinner mothers, especially women of color—you can’t opt out safely. But you also can’t opt in blindly.
That’s the trap.
Very fun. Love it here.
And yet… women aren’t frozen. They’re moving.
1,817 net new women-owned businesses are created every day in the US. Women-owned businesses grew 44% faster than men-owned from 2019-2025. 59% of female founders are solo entrepreneurs.
Women aren’t retreating from work.
They’re redirecting it.
Here’s my bet: the portfolio career—income and expertise from multiple sources, not all dependent on one employer—is becoming the more resilient path. Especially for women. Especially now.
These women starting businesses didn’t wake up one day as founders. They started by knowing what they were good at, being able to articulate it, and having something portable to build on.
That’s what the four moves below are actually for. Not “start a side hustle.” Start being legible to yourself … so that if you ever need to move, pivot, or build, you’re not starting from scratch.
Move 1: Name One Thing You Do
Not a framework. Not a methodology. Just one thing you do that works.
I kept having the same conversation about career resilience in the AI era. Different people, same questions. Eventually I just started calling my answer “the ADAPT framework” (Agility, Documentation, AI Fluency, Personal Brand, Ties). I needed shorthand so I could stop re-explaining from scratch every time.
That's all naming is. A reference point. A handle. Something that can travel without you having to be in the room explaining it for the nine hundredth time.
Why this feels harder for women: Research shows women now negotiate at similar rates to men… but they're more likely to be rejected when they do. The ask isn't the problem. The reception is. We’re already less likely to claim credit for our innovations. For women of color in predominantly white workplaces, the risk is sharper—visibility cuts both ways.
The reframe: You’re not naming it for ego. You’re naming it so when people reference it, there’s something to attribute. Otherwise your expertise gets absorbed into organizational knowledge with no trace of you. (See: Part 2, the whole thing about codified capabilities.)
The actual move:
Think of a problem you’ve solved multiple times
Write down the 3-5 steps you use
Give it a name (can be descriptive, doesn’t need to be clever)
Use that name when you talk about it
That’s it. You’re not trademarking anything. You’re creating a reference point that travels with you when you leave.
Move 2: Document One Process Publicly
Remember from Part 2: every process you optimize at work is training data for systems you don’t control. Your expertise is being extracted and made portable without your name on it.
Public documentation flips this. YOU become the source.
Pick ONE thing you know how to do that other people struggle with. Write 300-500 words about how you do it. Post it on LinkedIn or Substack or wherever your industry hangs out.
Examples of “mundane” expertise worth documenting:
How you onboard new team members
Your system for managing stakeholder communications
The way you structure client presentations
How you triage incoming requests
Your approach to running meetings that don’t make people want to die
This isn’t thought leadership. It’s creating a paper trail that says: I know how to do this. I’ve been doing it since before the AI could.
“But I don’t have time to write”
You’re reading a Substack. You have time.
Jk, jk. We are here the long form reading in 2026.
Use AI. Seriously.
Record yourself explaining the process (5 minutes). Have Claude transcribe and structure it. Edit for 10 minutes. Publish.
You just used AI to document your expertise before someone else’s AI learns to replicate it.
That’s the Angela Algorithm in practice.
Move 3: Start Building One Thing Outside the Walls
This doesn’t mean quit your job. It means: create one small income stream or expertise base that travels with you.
The Angela Algorithm isn't named after Angela Martin from “The Office” —but honestly, she got there first. By the series finale, Angela had quietly built a farm while everyone else was still optimizing for Dunder Mifflin promotions. Meanwhile Andy Bernard, who spent his seasons chasing visibility, ended up doing... beer commercials.
What “outside the walls” actually looks like:
Tier 1: No extra time required. Turn internal presentations into LinkedIn posts. Repurpose work analyses into short essays. Document processes you’d do anyway. This is just redirecting output you’re already creating.
Tier 2: 2-5 hours/week. Weekly newsletter. Monthly workshop. A few consulting calls. (A friend charges $300/hr for two client calls a month. That’s $600 for four hours of work she’d basically do for free as “networking.”)
Tier 3: 10+ hours/week. Course. Template library. Physical product.
Start at Tier 1. The point isn’t hustle culture. The point is making sure some of your productivity gains compound for you, not just your employer.
Move 4: Use AI To Buy Back Time, Not To Do More Work
This is where most people screw it up.
They use AI to get through email 30% faster... so they can take on 30% more projects... at the same salary.
That’s not productivity. That’s extraction. You’re 3x-ing yourself for someone else’s benefit.
Remember the line from Work Without Jobs that I quoted in Part 2:
“If the human is now 50% more productive doing 80% of the tasks of their job, what do you pay them?”
Nobody has answered that question. And until they do, the default is: you do more work for the same money.
The redirect:
If AI saves you 5 hours/week, spend those 5 hours on:
Building the thing from Move 3
Deepening relationships that matter
Learning something that compounds
Literally anything that builds YOUR leverage
How this actually works for me:
I use Claude to set up a marketing automation system (~2 hrs/week saved), analyze interview transcripts from my user research (~4 hrs/week), generate first collateral drafts (~3 hrs/week).
I don’t use that time to do more “work” work. I use it to:
Write this Substack
Develop the Womansplaining AI podcast
Work on research that matters to me
Same hours. Different output. One builds equity I own. The other builds productivity someone else captures.
The Part That’s Actually Hard
I’m not going to pretend this playbook lands the same way for everyone.
Some people are experimenting with AI because they’re curious.
Others are doing it because the math is brutal.
If you’re a breadwinner mother, you’re not optimizing for “career optionality.” You’re optimizing for stability. For making sure one job loss doesn’t collapse everything.
If you’re in one of the roles being automated first (education support, healthcare admin, HR, legal assistance) waiting to “see how this plays out” isn’t neutral. It’s risky.
And if you’re in a workplace where questioning tools gets you labeled difficult, or refusing tools gets you penalized? That tension is real.
I’m naming this because pretending this is equally hard for everyone would be dishonest.
But I’m also naming it because the alternative — hoping your employer will protect you from what’s coming — isn’t working either.
Between February and April 2025, roughly 300,000 Black women left or were pushed out of the labor force—even as the overall economy added jobs.
So the question isn’t “Is this fair?” (It’s not.)
The question is “What are you building anyway?”
Pick One Move This Week
Don’t try to do all four. Pick one:
30 minutes: Name one thing you do. Write it down. Use that name in your next conversation.
2 hours: Document one process publicly. LinkedIn post. Substack. Wherever your people are.
5 hours: Start building one thing outside the walls. Newsletter. Consulting. Product.
Already using AI: Redirect 1 hour of saved time per week toward YOUR thing, not more work.
That’s it. Not a rebrand. Not a side hustle empire. Just one move that starts compounding for you.
And hey, if that move eventually becomes the foundation for something bigger (a consulting practice, a product, a business) you'll already own it.
The Pattern I Keep Seeing
I talk to people all day about AI and work. Here’s what I notice:
The ones who feel like AI is extractive are using it to optimize someone else’s workflow. The ones who feel like it’s leverage are using it to build something that’s theirs.
And here’s the thing: you don’t have to start with work.
The women I know are all relentless problem solvers. Guarantee there’s some annoying thing in your daily life you could solve with AI - and honestly, that’s often a better place to start than trying to optimize your job.
Real examples:
The group chat problem: I kept scrolling up endlessly in my college group message to find restaurant recommendations. Built a simple tracker that pulls recs from the chat.
The parent coordination nightmare: Multiple friends have built calendar sync tools for kids’ activities, community coordination platforms for parent groups, shared resource libraries for their neighborhoods.
My mom’s retirement move: She’s starting a travel advisory business (classic post-retirement pivot). We built a trip builder tool in Replit. It’s actually pretty badass.
None of these started as “businesses.” They started as “this is annoying and I bet I could fix it.”
Some stayed personal tools. Some turned into things other people wanted. The point is: you get to decide what happens next.
When you build something - even something small - for yourself first, you learn:
How the tools actually work
What you’re capable of making
Whether this thing has legs
And if it turns out other people want it? You already own it. Your name is already on it. You built it outside the walls.
That’s the Angela Algorithm.
Not “use AI to do more.”
Use AI to build something that’s yours.
This is Part 3 of a 3-part series.
P.S. What’s your version of the Angela Algorithm? What have you built (or what are you building) that deserves your name on it? Hit reply.
P.P.S. Any takes on what I should write about next? I had one TikTok about AI and hiring hit 300,000 views... seems like the people might have already spoken.
See you next week.*
*or... sometime in January.
Logan
This resonates because it names the asymmetry most productivity narratives avoid: AI increases output, but ownership doesn’t automatically follow.
What you describe isn’t hustle — it’s boundary-setting. Using AI to externalize and document what you already know is a way of reclaiming authorship before systems extract it by default.
The core insight here isn’t “use AI more,” but “decide who captures the upside of your increased capacity.” That question is quietly becoming existential.