AI Operations: The Evidence Changed the Question
36 discovery sessions in, the question changed: what has to be true about the work before AI can improve it?
Product & AI OperationsKaty Simmons
I find the actual problem. Then I build what fixes it.
Sixteen years in messy systems: disconnected tools, buried knowledge, fuzzy ownership, and now AI dropped into the middle of all of it.
Flagship case studies
MAIA: AI-Generated Content Detection at Platform Scale
AI-generated music detection at 700K tracks a day. The model was one component. The decisions around it were the product.
Turning Consumer Support Into a Product
I stopped treating support as a queue and rebuilt it as a product, with confidence thresholds deciding what AI handles alone.
02The through-line
I learned at The Weather Company that an AI support system only works if the knowledge, workflow, analytics, trust, and escalation around it work.
At SoundCloud that expanded across customer operations, legal, marketing, platform integrity, monetization, and governance.
At Honeycomb I am a layer deeper: a repeatable way to understand work, decide where AI belongs, and create the conditions it needs to be reliable.
03Also available for
Fractional leadership and consulting.
Not every team needs a full-time executive. Sometimes you need someone to find the actual problem, design the operating model, and get the work moving. I take a small number of fractional and advisory engagements each year.
See how it works →04Where the work happened
The Weather Company
The Weather Channel
Weather Underground
Storm Radar
SoundCloud
honeycomb.io
Apptegy
The Venture Center
BBA CorpFive industries, one job: make it legible.
05How I work
My favorite discovery question is basically this: what keeps you up at night, and what drives you insane? I want all of it.
I listen for the problem underneath the problem.
Someone says they need an AI agent. Maybe. What are they trying to accomplish? What information do they need? Where does it live? What requires judgment? Who owns the decision?
By the time we get through those questions, the answer is usually starting to reveal itself. That is the work I love.
- 01Lead with the problem. The tool is the conclusion.Most organizations start with the technology and reverse-engineer a justification. That order produces expensive answers to questions nobody asked.
- 02AI is an amplifier, not a shortcut.It multiplies whatever is already there: clarity or confusion, judgment or guesswork. It does not supply the thing you are missing.
- 03Automate the mess and you get the mess at scale.Speed applied to a broken process is not efficiency. It is a faster way to be wrong.
- 04Treat operations as products.Internal systems deserve owners, users, research, roadmaps, and success metrics. Most get a spreadsheet and a hope.
06What people say
She consistently cut through the noise, evaluating opportunities and pressure-testing their potential.
Maria Von WatzdorfVP, Product at SoundCloudRead in full →She set the tone for the product team and laid the foundations for the majority of the products that are growing today in Apptegy.
Guillermo AraizaStaff Product Manager at CrunchyrollShe has strong convictions and ideas, and backs both up through methodical research and validation.
Kathleen ConleyProduct Manager at ApptegyShe finds the joy in solving problems for our clients and is really encouraging, the whole way.
Nicki HenningtonUI/UX Director at Apptegy
Contact
Let's get into it.
Ambiguous problem, AI adoption that stalled, an operating model that stopped scaling, a support function that should be a product. That is the conversation I want.