UX BY ASH

    AI for Design Leadership

    How design managers and directors can use Claude Cowork to automate the management tax.

    Most of a design leader's week is invisible coordination work. Claude Cowork is the first AI tool that can credibly take a real slice of it. Here is how I am using it, what it is genuinely good at, and where it still falls short.

    Author

    Ashwarya Subhluxmi

    Audience

    Design managers, Directors

    Format

    Practical guide

    Updated

    June 2026

    Context

    What Claude Cowork actually is, in one paragraph.

    Claude Cowork is Anthropic's desktop agent for knowledge workers. It launched in research preview on January 12, 2026 and went generally available on April 9, 2026 inside Claude Desktop on macOS and Windows. The short version: it is Claude Code's agentic loop, redirected at files, documents, browsers, and connected apps instead of repos. You give it a goal, it plans the steps, opens what it needs (Google Drive, Gmail, local files, browsers, plugins), and returns a finished artifact. It can also run on a schedule, so a task like "every Monday at 8am, summarize my team's design reviews from last week" becomes a recurring job, not a recurring meeting.

    I am writing this as a design leader who has been using Cowork daily since the preview. This is not a feature tour. It is the list of management tasks where it has actually saved me hours, and the ones where I stopped trusting it.

    The Real Problem

    Design leaders spend half their week on management overhead, not design judgment.

    If you have ever audited your own calendar, you already know the shape of it. Standups, design reviews, 1:1s, cross-functional syncs, hiring loops, performance docs, roadmap reviews, status updates for the leaders above you. Most of this is necessary. Almost none of it is the reason you were hired. The job you were hired to do is to set direction, raise the quality bar, and grow people. Everything else is the management tax.

    Claude Cowork does not eliminate that tax, but it does compress it. The pattern that has worked for me is to treat Cowork like a smart, very fast junior chief of staff. I do not ask it to make decisions. I ask it to do the reading, the drafting, the formatting, and the chasing that sits in front of every decision.

    The leadership unlock is not what Claude Cowork does for me. It is the meetings I no longer need to take because the artifact already exists.

    Use Case 01

    Weekly team digest, written before your Monday standup.

    Anthropic's help docs walk through scheduled tasks in Cowork, and this is the use case I built first. Every Sunday at 6pm, a scheduled task pulls the last seven days of activity from our shared Google Drive, our Figma weekly updates folder, and the design review notes in our Obsidian-synced vault. Cowork drafts a one page digest organized by pod, with a section for shipped work, work in progress, blockers, and themes worth raising in leadership.

    I edit it in fifteen minutes on Monday morning. Before this, the same digest took me two hours and I often skipped it. The compounding value is that my skip-level leader, my PM peers, and my own team all see the same picture of the week without me writing it three times.

    • Trigger: weekly recurring task, Sunday 6pm.
    • Inputs: Google Drive folder, Figma weekly update doc, design review notes.
    • Output: one page digest in a shared doc, with a clear shipped, in flight, blocked, and themes structure.

    Use Case 02

    Performance review prep that does not eat your weekends.

    Performance cycles are the single largest management tax of the year. For each direct, I keep a running Obsidian note of feedback moments, project wins, growth areas, and praise from cross-functional partners. At review time, I point Cowork at that note plus the person's last two quarterly self-reviews, plus their work artifacts in Drive, and ask it to draft a structured review against our company's competency rubric.

    It is not a final draft. It is a starting draft that hits eighty percent of the structure and citations, so I spend my time on the twenty percent that actually matters: the narrative arc, the candid growth feedback, and the leveling call. The judgment stays human. The synthesis becomes free.

    • Inputs: running 1:1 notes, prior self-reviews, work artifacts, competency rubric.
    • Output: a draft review with specific examples mapped to each competency.
    • Hard rule: never paste the draft into the HR system without a full human pass, and never let the model invent examples it cannot cite.

    Use Case 03

    Deep research that you would otherwise punt on for another quarter.

    This is where Cowork separates from chat-based Claude. Because it can open a browser, read PDFs, traverse a Drive folder, and run multi-step plans, it can do the kind of research project a director keeps on a wish list and never staffs. Recent examples from my own queue: a competitive teardown of every AI shopping agent that shipped in Q2, a synthesis of public posts on how design orgs are restructuring around AI, and a longitudinal read of our own customer feedback themes over six months.

    The right prompt shape is not "do a competitive analysis." It is a brief, the same way you would brief a senior researcher. Goal, audience, sources to prioritize, sources to ignore, deliverable format, and a clear instruction to cite every claim. Cowork will return a draft with footnotes you can click. The teams I trust most then have a person verify every load-bearing citation before the artifact ships.

    Treat Cowork like a research analyst, not an oracle. Brief it like a human, verify it like a human, ship it like a leader.

    Use Case 04

    Hiring loop logistics, finally off your plate.

    Hiring is the management task that breaks first when you get busy, and it is the one where breakage costs the most. Cowork is unusually good at the boring half: drafting structured interview guides from a job description, synthesizing debrief notes into a single recommendation doc, comparing candidate portfolios against a scorecard rubric, and writing the first pass of a recruiter update.

    It is not good at the judgment half. Do not let it score candidates, do not let it write rejection emails without a human pass, and do not feed it sensitive candidate information unless your workspace is on an enterprise plan with the right data controls in place.

    • Draft a structured interview guide from a JD and rubric.
    • Synthesize debrief notes from four interviewers into a single recommendation.
    • Compare a finalist's portfolio against the leveling rubric and flag gaps.

    Use Case 05

    Leadership updates and board-style narratives in a quarter of the time.

    Every director I know writes some version of a monthly or quarterly update for the leader above them. It is high-leverage writing and it is almost always done at 11pm the night before. Cowork shortens this loop in two steps. First, it pulls the raw inputs: shipped work, metrics from a connected dashboard or exported CSV, themes from the team digest, and any leadership feedback captured in 1:1 notes. Second, it drafts the update against a template I have already approved.

    The output is never the final word. But the difference between writing from a blank page and editing a structured first draft is the difference between a four hour night and a forty five minute pass.

    The Setup

    A minimum viable Cowork setup for a design leader, in under an hour.

    If you want to copy this exactly, here is the setup that has worked for me. It takes less than an hour to wire up and pays back in the first week.

    • Install Claude Desktop and enable Cowork. You will need a Claude Pro, Max, Team, or Enterprise plan.
    • Connect Google Workspace so Cowork can read Drive folders, Docs, and Gmail threads. Scope access narrowly to the folders that hold team artifacts.
    • Point Cowork at your notes vault (Obsidian, Notion, or a synced Drive folder) so it has memory of past decisions and 1:1 context.
    • Install only the plugins you will actually use. Anthropic's open knowledge-work-plugins repository on GitHub is a good starting point.
    • Create three scheduled tasks: weekly team digest, monthly leadership update draft, and a Friday afternoon "what slipped this week" review.
    • Write one shared "how I work with Cowork" doc for your team, so everyone knows what is automated and what is still a human decision.

    Limits and Risks

    Where Cowork is still wrong, and how I keep it from biting the team.

    Cowork is a tool, not a teammate. It hallucinates citations under pressure, it occasionally edits the wrong file, and it does not understand the politics of your org. Three rules have kept it useful and safe on my team.

    First, never let it write anything that goes to a person without a human pass. That includes performance reviews, candidate rejections, and 1:1 talking points. Second, never let it touch sensitive data unless you are on a plan with the right data controls and your security team has signed off. Third, log what it does. The scheduled tasks panel inside Cowork is the audit trail, and you should treat it like one.

    Use the tool to compress the management tax. Keep the judgment, the relationships, and the calls in your hands. That is still the job.

    Sources and further reading

    Everything cited above, in one place.

    Newsletter

    New writing, frameworks, and lessons from AI-native product work, straight to your inbox.

    By subscribing you agree to receive occasional writing from UX by Ash. Unsubscribe anytime. See the Privacy Policy.

    Advisory

    Startup advisory & fractional leadership.

    Partnering with founders and leadership teams to build AI-first products, scale design organizations, and drive measurable business impact.

    Request Availability Book A Discovery Call
    AI-Native Product Strategy

    From idea to roadmap, I help define the right problems and scalable solutions.

    Design Leadership On-Demand

    Embed executive design leadership without the full-time overhead.

    Team & Process Building

    Hire, mentor, and align design teams to build high-performing culture and processes.

    Growth & Impact

    Align design with business goals to drive adoption, retention, and revenue growth.

    Ashwarya Subhluxmi

    Design leader for Credit Karma's $3B marketplace. Five business lines, a team of staff-level designers, 140M+ members.

    Now

    Building AI agents at Credit Karma. Coaching design leaders on the side.

    San Francisco, available for advisory
    © 2026 Ashwarya Subhluxmi

    Legal notice. All content on uxbyash.com, including written work, case studies, frameworks, visual design, code, illustrations, and page layouts, is the original work of Ashwarya Subhluxmi and is protected by copyright and other intellectual property laws. No part of this site may be copied, reproduced, republished, mirrored, scraped, fine-tuned into a model, or used to train AI systems without prior written permission. Short quotations for editorial, educational, or commentary purposes are welcome with clear attribution and a link back to the source page. Unauthorized use, including verbatim reuse of copy or design, will be treated as infringement and pursued under applicable law.