← Selected work

Case study | AI workflows · Process design · Operations

AI-assisted Marketing Ops

Marketing Ops work arrived from Slack, meetings and direct requests. I gave it one home in Asana, then built Claude workflows around it, and task completion rose 53% compared with the period before Claude.

Tools
Claude, Asana, Slack, Notion
My role
Process design, Claude skills and Asana setup
Worked with
Revenue Operations team and main stakeholders
+53%

task completion compared with the period before Claude

1

place for every request and initiative, with its full context

3

sources Claude sweeps for status updates: Asana, Slack and Notion

01

The problem

Marketing Ops work came in through several doors: requests started in Slack, and important initiatives came out of meetings. Each one needed context to be done well, and that context was easy to lose along the way.

Without a single place for the work, there was little to hand to an AI assistant, and a lot of manual effort went into logging information and keeping people updated.

02

What I found

Insight 1

Intake sets the quality

If a request arrives with everything needed, the work starts faster. The fix belonged at the front door, not in follow-up questions.

Insight 2

Context is what makes AI useful

Once every task carried its own context, each one became a ready starting point for Claude to evaluate and move forward.

Insight 3

A log makes work reusable

A task with its full timeline can be picked up later, and compared with similar ones, without asking anyone what happened.

03

Getting everyone on board

A new process only works if people use it, so I approached each group differently. With RevOps leadership, I laid out the benefits clearly so they backed the redesign from the start.

With teammates, I shared my own experience of working this way, rather than just describing it. And I asked the leaders of our main stakeholder teams to collaborate and encourage their people to adopt the new process.

04

What I built

Before
  • Requests and initiatives starting in different places
  • Information missing when a request arrived
  • Context kept by hand, if at all
  • Status updates put together manually
After
  • Every Marketing Ops task centralized in Asana
  • A custom Asana form that collects all required information from the start
  • Automations that log requests started in Slack in Asana too
  • Claude skills that turn meeting transcripts (Gemini, Granola) into structured tasks, with context, related links and next steps
  • Claude evaluating each task and suggesting next steps, or doing the work itself
  • Progress logged as comments over time, by Claude or by hand
  • Claude sweeping Asana, Slack and Notion to draft status updates for team meetings and leadership 1:1s

The key choice was making Asana the source of truth. Because requests, meeting follow-ups and other initiatives all landed there with the same structure, Claude could help at every level, from heavy-lifting on the work itself to the organizational chores of logging context and keeping stakeholders informed.

05

Results

Operational efficiency, measured by task completion, rose 53% compared with the period before Claude.

Beyond the number, every task now has its context and timeline, so work is easy to pick up later or compare with similar tasks. The same logs became the raw material for status updates, so sharing progress with the team and with leadership takes far less manual effort.

Want to talk about AI-assisted ops?