how I got real AI product experience without changing jobs

how I designed my own AI project, and the 7 skills I talked about in senior AI product management interviews

Questions I’ll cover:

“What kinds of skills and experiences do I need to become an AI product manager?”

“How do I get hands-on AI experience if my company doesn’t work on AI, and there’s no one around me to learn from?”

The tough reality is, without hands-on AI experience — whether that’s through production-level personal projects or through launching AI products at a company — it’s hard to get past the resume round.

I created this guide because I know how it feels to be willing to put in the work, but not have the resources or not know where to start.

So here’s:

• how I designed my own project to get hands-on AI experience

• the skills I wanted to build and talk about in future interviews

• how I did gained those skills

Context:

I joined a creator economy startup as a senior product manager and was immediately asked to tackle a difficult challenge: salvage a product that had dragged on for 1 year without launching. The project had been stuck in engineering development, consuming resources without delivering value to users or the business.

The original intention was to build a video editor that helped users edit videos just like they edit text: it transcribed the video so that users could delete and rearrange sections of a video the same way they would edit a paragraph.

Some issues with the original product:

• Didn’t solve the most pressing problems for our users

• Wasn’t a strategic investment for our company; even if this product exceeded all our target metrics, it wouldn’t impact the company’s business significantly

What I pitched:

• A complete pivot to a content repurposing tool that analyzed the content of the user's long-form course videos, and repurposed that content into 40+ different types of content for their end-to-end marketing funnel, e.g., TikTok videos, blog posts, newsletter posts, LinkedIn posts, landing pages, discussion questions to engage their community, and more.

Skill #1: Prompt engineering

Each type of content required a different prompt engineering approach. I worked with my engineering team to build an internal tool that allowed me to tweak different dimensions and test prompts quickly. Also, to reduce API costs and handle scenarios where the original content was long and exceeded the model’s context window, I wrote a prompt to first summarize the original content in a way that preserved critical details for subsequent repurposing.

This gave me hands-on experience with:

• Creating base prompts for different content types (social media, long-form articles, video scripts)

• Designing objective evaluation criteria to measure prompt effectiveness

• Iterating based on real user feedback and output quality

• Building guardrails to ensure brand voice consistency across all generated content

Skill #2: AI models strategy

Models differ significantly in their strengths, price, and latency.

Through this process, I learned how to:

• Evaluate various models for different content types

• Balance cost, performance, and reliability when selecting models

• Implement a multi-model approach where certain tasks were routed to specialized models

• Create fallback systems when primary models failed to produce adequate results

Skill #3: AI partnerships / business development / corporate development

We were a startup and couldn't build everything ourselves, so I searched for companies we could partner with (or even acquire). I reached out to founders on LinkedIn and did due diligence on 10+ startups. I created a framework to compare them as objectively as possible, and we ended up choosing to partner with 2 companies.

I created myself opportunities to:

• Develop evaluation criteria for potential partners (technical capabilities, API reliability, pricing models, roadmap alignment)

• Negotiate partnership terms beyond product/eng benefits

• Integrate external services seamlessly into our product

• Create contingency plans for partner relationship management

Skill #4: Monetization / pricing & packaging

This product was a first for the company in many ways:

• 1st add-on in its 13-year history

• 1st AI product

• 1st product that relied on several external partnerships

Thus, we faced several challenges:

• Set new customer expectations that not everything would be included in the monthly subscription

• Lock the pricing and packaging with no historical data about usage patterns

• Build the underlying monetization technical infrastructure to support add-on billing

• Set the precedent for future add-on products

I built several financial models and ran a beta community to estimate projected costs. This approach helped me develop pricing tiers based on actual usage patterns from our beta users and create clear value propositions for each tier that justified the additional cost.

Ultimately, our pricing model was 2 tiers of add-ons:

• Tier 1: $19/month to repurpose any 4 videos into unlimited pieces of content

• Tier 2: $29/month to repurpose any 30 videos into unlimited pieces of content

I’ll create a separate post dedicated to how we arrived at this pricing model.

Skills I gained:

• Design pricing models that balanced profitability with customer value

• Run a beta community

• Implement usage monitoring and cost prediction systems

• Create transparent billing practices that built trust with users

Skill #5: Refining the UX of AI

I did extensive competitive analysis of AI products across different modalities — video, image, text, chat, etc.

Skill #6: Go-to-market

My product marketing colleague was busy, so I saw an opportunity to practice my product marketing skills. I drafted the majority of the marketing content needed for press releases, blog posts, social media announcements, website, and more. This included leveraging one of our partners for co-marketing opportunities.

I also pitched working with top content creators and influencers. I provided talking points, customized scripts, and sample b-roll to make it as easy as possible for them to film content.

Through this experience, I built skills like:

• Crafting compelling AI product narratives

• Developing educational resources to help users understand new AI capabilities

• Building excitement through targeted pre-launch campaigns

• Creating feedback loops to rapidly iterate on messaging based on user response

• Developing scalable systems for responding to user questions and feedback

Skill #7: Model-forward development

One of the most valuable skills I developed was what I call "model-forward development”: the ability to anticipate, evaluate, and seamlessly integrate new AI models as they emerged. This approach was critical in keeping our product competitive in the rapidly evolving AI landscape.

Anticipating and planning for model evolution

From an engineering perspective, I worked closely with my eng team to build our product architecture with model-agnosticism at its core:

• Designed a modular API layer that abstracted model-specific implementations, allowing us to swap in new models without disrupting the entire system

• Created standardized input/output formats that could accommodate different model response structures

• Implemented A/B testing infrastructure that allowed us to compare different models on the same tasks using real user inputs

From a financial/business model perspective, I built flexibility into our budgeting:

• Negotiated volume-based pricing with providers

• Maintained a financial cushion specifically for testing and implementing new models

• Developed cost projection models that accounted for potential increases in token usage or per-request costs