one page draft

Roadmap Decision Brief

An AI tool that helps product managers make better product decisions using their company’s data.

Team: Sophia, Sathmi, Romir, Raza

Description: What Is It?

Roadmap Decision Brief is an AI tool that helps product managers decide which features to build next. Users upload customer feedback, engineering estimates, and business goals, and the tool turns that information into a short decision brief explaining the options and evidence behind them.

Problem: What Problem Are We Solving?

Product managers have to make decisions based on information spread across customer interviews, internal documents, and engineering reports. Going through everything takes time, and it can be hard to know which information is reliable.

Our goal is to make this process faster and easier while helping PMs make decisions they can confidently explain to their teams.

Why: How Do We Know This Is a Real Problem?

We created five fictional B2B companies and tested our prototype using sample customer feedback, business goals, and engineering estimates.

Our testing showed that the workflow was easy to follow, but we found issues with how the tool handled evidence. For example, attaching an unrelated document incorrectly marked a claim as supported. The prototype also sometimes confused technical restrictions with conflicting information.

Of the four checks we completed, only one passed. This showed us that making the tool trustworthy is just as important as making it easy to use.

We still need to test with actual product managers to understand how useful it would be in their daily work.

Audience: Who Are We Building For?

Our main customers are product managers at B2B software companies who regularly decide which features their teams should prioritize.

We are especially focused on PMs who deal with lots of customer feedback and need to explain their decisions to engineers, designers, and leadership.

What: How Does It Work?

The PM starts by entering a question, like whether their company should build a Slack integration or a mobile app.

They then upload relevant documents, and the AI reviews the information to create a brief comparing the options. The brief highlights customer demand, business goals, technical constraints, and missing evidence.

Users can check the sources behind each claim before sharing the brief with their team.

Our current prototype uses a pre-written example, so building the actual AI document analysis is our next major step.

Business Model: Who Pays and Why?

We plan to charge $20 per user per month through a subscription model.

Our customers would be individual PMs or product teams at software companies. They would pay because the tool could save time preparing roadmap decisions and help them make stronger recommendations.

We haven’t validated this price yet, so we plan to ask working PMs whether they would actually pay for it. We also need to update our landing page, which currently shows $15.

Success: How Will We Measure It?

In the short term, we want at least 4 out of 5 new testers to complete the workflow in under five minutes without help. We also want 4 out of 5 testers to recognize weak evidence and understand where claims come from.

We want all five of our evaluation checks to pass and each fictional company to receive a brief based on its own documents.

Longer term, we will measure how much time PMs save, whether they find the briefs useful, and whether they are willing to pay.

How: What Is Our Testing Plan?

First, we will fix how the prototype handles evidence. Claims should not be marked as supported unless the attached information actually supports them. We will also improve how the tool handles technical restrictions and older documents.

Next, we will connect the AI model and test it with our five fictional companies to see whether it generates accurate, different briefs.

Finally, we want to test with real product managers to get feedback on usefulness, trust, and pricing.

When: What Are Our Next Steps?

Our first step is fixing the problems we found in testing. After that, we will improve the workflow, add AI document analysis, and test the complete product.

Once we know the tool works reliably, we can explore launching it to a small group of real users.

Non-Goals and Risks

We are not trying to replace product managers or build an entire project management platform. We just want to make roadmap decisions easier.

Our biggest risks are inaccurate AI outputs, privacy concerns around company documents, and whether customers will actually pay for the product.

Connection to This Week’s Reading

We used Lenny’s one-pager template to keep our focus on the problem, who we’re building for, and how we’ll measure success. We also added a business model section to explain how we plan to make money and what we still need to test.

Avatar

About the author

Leave a Reply