Ashley McNamara woke up one morning and decided the chaos had to stop. As senior director of developer relations at GitHub, her days once dissolved into frantic context switching. Fifteen different systems held pieces of her work. Her brain served as the only glue. That stopped working.
So she built 40 custom automations inside the GitHub Copilot app. The result surprised her. The tools didn’t replace her judgment. They freed her to exercise it.
“Here’s the thing about senior leadership that nobody warns you about: the job isn’t hard because of any single task. It’s hard because your work lives in fifteen different places and your brain is the only system connecting them,” McNamara wrote in the post published yesterday on the GitHub Blog. “I was spending so much energy on context-switching that I had nothing left for the thinking, connecting, and creating that my role actually requires.”
Her mornings now start differently. Pre-run automations deliver summaries while she makes coffee. Meeting prep agents pull context from calendar invites, related documents, and recent discussions. No more pretending to have read the agenda while scanning twelve open tabs. The shift feels subtle at first. Then it compounds.
Some automations run on schedule. Others trigger on events. A few wait for her explicit prompt. Together they handle the invisible labor that once consumed her focus. Meeting prep. Daily triage. Follow-up tracking. Stale work detection. Travel logistics. The list covers categories from career architecture to team recognition to maintenance tasks.
One favorite, the Stale Work Finder, scans her GitHub activity using the gh CLI. It surfaces pull requests, issues, and comments that need attention. Results proved eye-opening. Another, the Commitments and Follow-Up Tracker, reviews her own messages for promises made. It flags anything not yet delivered. “No. And that distinction matters to me more than anything else in this post,” she notes about its role in protecting trust.
The Daily Wins Recap arrives in the evening. It counters the natural tendency to fixate on what went wrong. Instead it lists accomplishments with evidence. For someone with AuDHD, that consistency matters. Executive function varies. The automations don’t.
But automation carries risks. Austin Henley learned this the hard way. After automating 95 percent of his coding tasks, new glue work appeared. Context switching multiplied. Errors crept in. “It never gets easier, you just go faster,” he quoted cyclist Greg LeMond in his June 7 post on austinhenley.com. The lesson? Speed without reflection creates its own problems.
McNamara avoids that trap. Her agents augment human connection rather than remove it. They surface data so she spends meetings listening instead of scrambling for facts. They draft recognition notes based on actual contributions. They give her back headspace.
“Automations won’t fix organizational dysfunction or bad management or an unreasonable workload. But they can give you back enough headspace to actually do the work you’re here to do,” she cautions.
The New Reality for Technical Leaders
Recent coverage shows this experience isn’t isolated. Companies now create roles like AI business automation engineer specifically to help employees remove drudgery from their own jobs, The New York Times reported earlier this month. At Box, such positions sit inside IT yet focus inward on internal efficiency.
Job postings for AI automation specialists now routinely advertise salaries between $91,000 and $165,000, according to current listings aggregated by ZipRecruiter. Demand spans engineering, architecture, and leadership tracks. Forward-deployed engineers and AI platform leaders appear on nearly every major tech hiring list for 2026.
Yet the real story lies in how individual contributors and managers adopt these tools without waiting for top-down mandates. Developers already experiment with minimal prompts to Copilot CLI. They ask the model to watch session logs and suggest its own automations. Some explore self-improving agents inspired by Andrej Karpathy’s autoresearch project.
Henley advises a simple rule. Anything done more than twice deserves automation. Start small. Use the least detailed prompt possible. Let the model fail, then correct it. The approach produces reusable skills that run automatically rather than scripts that require manual triggering.
McNamara built hers iteratively. She started with one painful friction point — meeting preparation. Then audited every surface where her work lived. Calendar. Email. Messages. Repositories. Issues. Each revealed new opportunities. The Copilot desktop app made creation straightforward. Scheduled prompts, integrations with company systems, even MCP servers for advanced behaviors.
Critics once warned that widespread automation would hollow out roles. Evidence points the other way. Leaders who master these systems report stronger presence with their teams. Better decisions grounded in fresh data. Reduced burnout. The work doesn’t disappear. It elevates.
Still, not every task suits automation. Complex architecture choices. Difficult conversations. Creative strategy. These remain human domains. The agents handle the preparation and follow-through. People handle the nuance.
So what happens next? More leaders will follow this path. Some already do. They won’t announce they’ve automated half their workload. They’ll simply show up more prepared, less frazzled, and focused on what matters. Their teams will notice the difference in recognition, follow-through, and strategic clarity.
The tools exist today. The GitHub Copilot app. Similar capabilities from other providers. The question is no longer whether to automate. It’s which invisible tasks drain your attention first. Find one. Build the agent. Watch what changes.
McNamara’s experiment offers a blueprint. Start narrow. Iterate relentlessly. Measure the return in reclaimed focus rather than lines of code. And remember the goal isn’t fewer hours at work. It’s better work during those hours.
WebProNews is an iEntry Publication