​

Hi Reader,

Have you been asked to “add some AI” to your internal product yet?
If not, it’s only a matter of time.

And if, like me, you’re not an AI expert but want a quick, practical overview of what’s possible – this one’s for you.

​

​

Today in 10 minutes you will:

• Learn 3 levels of AI you can apply to internal products
• Understand when to use each level with my AI Decision Matrix Tool
• See real examples (chatbots, agents, and custom models)
• Know where to start without hiring AI engineers

​

​

Why This Matters

I’m genuinely excited about what AI can bring to our work.
​
Yes, it’s a bit scary, but it also has the potential to elevate internal product management and free our users from repetitive, low-value tasks.

According to Gartner, by 2028 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024, enabling 15% of daily work decisions to be made autonomously.

That’s our world.
We can either get on the AI train or run after it.

Here’s how I think about applying AI to internal products.

​

Step 0. Start from genuine user needs

Users don’t wake up wanting AI.
They wake up wanting less friction in their day.

→ They don’t want to reconcile budgets across three tools.
→ They don’t want to read every single email.
→ They don’t want to fill out forms twice because one field was wrong.

AI should solve these pain points, not exist for the sake of it.

Before you start, evaluate your user needs and assess where AI can actually help.

​

Level 1. AI for user interfaces

Goal: Help users self-serve faster

Examples:​
→ Chatbots that answer questions from your knowledge base
→ AI writing assistants for reports or release notes
→ Smart form filling and data entry helpers

Time to value: 2–8 weeks
​Complexity: Low – use existing APIs (OpenAI, Anthropic, Gemini)

​

Level 2. Intelligent agents

Goal: Scale knowledge work and shorten decision loops

Examples:​
→ Data analysis agents: “Explain trends across our Power BI reports”
→ Incident triage assistants: categorize and summarize logs
→ Product update agents: summarize sprint data from Notion or Azure DevOps

How it helps:​
→ Scales knowledge work
→ Enhances visibility into delivery
→ Reduces repetitive analysis

Time to value: 2–4 months
​Complexity: Medium – integration work + prompt engineering

​

Level 3. Build your own AI capabilities

Goal: Create custom AI for specialized use cases

Examples:​
→ Fine-tuned models for document classification or ticket routing
→ AI layers embedded into platforms (e.g., internal recommendation engines)
→ Domain-specific assistants (legal review, code analysis, process optimization)

Time to value: 4–12 months
​Complexity: High – needs ML expertise, infrastructure, and maintenance

Tip:​
Start here only if you have a strong use case or data advantage.
Otherwise, fine-tuning existing models will get you 90% there.

​

Try the Decision Matrix Tool

If you’re still not sure which level of AI makes sense for your internal product,
I created a quick Decision Matrix Tool to help you decide.

Click below to access it and explore your best next step:
👉 Open the AI Decision Matrix Tool

Getting started: 3 practical steps

1. Run a quick survey: “What repetitive tasks waste your time most?”
2. Pick one high-impact, low-complexity use case.
3. Prototype in two weeks using a GPT or Claude API.
   Then measure → time saved + user satisfaction → iterate or scale.

​

​

Behind the Scenes

Did you know Agatha Christie is the best-selling novelist of all time?
I’ve been winding down with her books lately, two in two weeks.
If you’re looking for something light but engaging before bed, she’s a classic.
Highly recommend.

​

​

What do you think?

Did this newsletter help you see how AI fits into internal products?

Hit reply and tell me – what kind of AI use case would you want to build for your users?

See you next week,
​Maria

​