<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Nathan Bilbao on Medium]]></title>
        <description><![CDATA[Stories by Nathan Bilbao on Medium]]></description>
        <link>https://medium.com/@nathanbilbao?source=rss-b341cc9ce7b1------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*bmcav-8GlcGIDbTajJ2cwA.jpeg</url>
            <title>Stories by Nathan Bilbao on Medium</title>
            <link>https://medium.com/@nathanbilbao?source=rss-b341cc9ce7b1------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 18:24:31 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@nathanbilbao/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[I Built a Bot to Track My Career So I Don’t Have To (Mostly)]]></title>
            <link>https://medium.com/@nathanbilbao/i-built-a-bot-to-track-my-career-so-i-dont-have-to-mostly-331591c065f1?source=rss-b341cc9ce7b1------2</link>
            <guid isPermaLink="false">https://medium.com/p/331591c065f1</guid>
            <dc:creator><![CDATA[Nathan Bilbao]]></dc:creator>
            <pubDate>Wed, 11 Mar 2026 23:39:47 GMT</pubDate>
            <atom:updated>2026-03-11T23:39:47.902Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Nathan Bilbao</em></p><p>I am inconsistent at documenting my wins.</p><p>Not at doing the work — at <em>recording</em> it. I’ll spend two weeks on a gnarly systems problem, land on something I’m genuinely proud of, share it with the team, and then completely fail to write it down anywhere. By review time I’ve forgotten half of it. My manager has to remind me of things I did. That’s embarrassing.</p><p>Every designer has a WIP folder, a shipped library, a graveyard of presentations and recorded walkthroughs, Figma files stacked three projects deep, prototypes nobody can find anymore, and design docs that went through final.doc, final-final.doc, and final-final-ACTUAL-final.doc before anyone read them. The artifacts exist. Connecting them to impact at review time is another thing entirely.</p><p>So I built a Claude skill that scrapes all of my company assets and activity — Google Meet transcripts, Slack, GitHub, Figma, and email — and gives me a monthly digest that maps my activity and impact to my business initiatives and competencies. The goal is to build evidence and confidence that I’m excelling at my job while surfacing blind spots before they become a problem at review time.</p><h3>What I Actually Get</h3><p>Every month, two things happen automatically.</p><p>First, I get an email digest. It reads like a mini performance review written about me, in third person, by someone who’s been quietly watching everything I’ve shipped, said, and built all month. It covers all 17 competencies in Flexport’s UX career matrix — things like <em>Interaction Design</em>, <em>Decision Making</em>, <em>Cross-Functional Influence</em>, <em>Mentorship</em> — and for each one it surfaces concrete signals: a Figma file I shipped, a Slack thread where someone said my work unblocked their team, a meeting transcript where I drove a key decision. Each competency gets a rating and real examples. Not vibes. Actual receipts.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BshaOI2VgGn9N-49g1KWXg.png" /></figure><p>Second, my bragsheet gets updated. This is a tab in my competency matrix Google Sheet — my running ledger of professional wins, skills demonstrated, and impact delivered. Instead of scrambling to fill it in at review season, it gets refreshed monthly with evidence the bot actually found. The document stays current without me touching it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CUs38HcHVRdX4gQ6-W7jaQ.png" /></figure><p>That’s the product. An email and an updated spreadsheet, delivered on the last day of every month, built entirely from work I was already doing.</p><h3>How It’s Built</h3><p>The whole thing lives in a SKILL.md file — a detailed procedure manual that a Claude agent reads before every run. Think of it as instructions specific enough that the output is consistent every time. Vague instructions produce vague results; the more precise the spec, the more reliable the agent.</p><p>When it runs, the agent works through each source: Drive for meeting transcripts, Slack for announcements and kudos and direct mentions, GitHub for commits and PRs, email for anything that didn’t surface elsewhere. It collects signals, maps them against the 17 competencies, and rates each one: Strong (2+ distinct examples from different sources), Partial (1 example or something indirect), or No Evidence — which doesn’t mean it didn’t happen, just that it left no visible trace.</p><h3>What It Changed</h3><p>When I built this I assumed the main benefit would be convenience. It is. But it also changed something subtler: what I pay attention to record.</p><p>The bot can only report on what it sees. If I do meaningful work in a private doc, it won’t know. If I help someone over DM instead of a public channel, it won’t know. If I make a call in an untranscribed meeting, it won’t know. That’s not a flaw — but it creates a real bias toward work that’s visible and documented. I’ve started being more intentional about that. Not performing for the bot, but making sure the work actually leaves a trace: posting the Figma link in Slack after a review, writing the reasoning in the doc instead of just explaining it verbally, by recording short form videos at the end of each day.</p><p>It will only report on what it sees. That’s both the feature and the caveat.</p><h3>The Tensions Worth Naming</h3><p>There are a few things this surfaces that I find genuinely interesting, not as criticism but as implications.</p><p><strong>It shifts the onus.</strong> Right now, tracking professional growth is a shared mess — the employee uses whatever system they happen to have, the manager observes what they observe, and at review time both parties reconstruct a narrative and hope it aligns. A system like this standardizes that, but the cost is that the onus falls more explicitly on the employee. If your work isn’t showing up in the places the bot looks, it won’t get credited. You’d need to work <em>in</em> the system, not just adjacent to it. That’s actually closer to how it should work — self-advocacy has always been the employee’s job — but this makes that expectation more visible and legible than it’s ever been.</p><p><strong>It’s surveillance, but it’s yours.</strong> The thing scrapes my meeting transcripts, Slack messages, GitHub activity, Figma files. It has a fairly complete picture of my working life. For now that’s useful because I control what it does with the output and who sees it. But a version of this run by an employer rather than the employee would feel very different. The same tool, with the same data, in different hands.</p><p><strong>It’s only as honest as you make it.</strong> Early versions of the agent would write things like “Nathan participated in a discussion about X” — technically true, not useful. I had to add explicit rules: every claim must trace to a specific source, ownership language must be precise (“decided” vs. “influenced” vs. “was credited for”), and anything unconfirmed gets flagged rather than stated as fact. A brag sheet that overstates your involvement is worse than no brag sheet. The bot needs to be honest about what it saw, which means being honest about what it didn’t.</p><h3>What I’d Actually Call This</h3><p>It’s not a performance review. It’s not a surveillance tool. It’s closer to a memory aid — a system that does the boring, consistent work of noticing and recording things so that I can focus on doing them.</p><p>And for now, it’s mine. I run it, I receive the output, and I decide what to do with it. That feels like the right direction for a tool like this — one that works for the person, not on them.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=331591c065f1" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>