Photo by the editor
# Entry
Creation product requirements document (PRD) is a common process in product management and a common task in sectors such as software development and the entire technology industry. Some of the common difficulties and strenuous requirements when creating a PRD include ensuring transparency, preventing scope creep, and maintaining stakeholder alignment.
Fortunately, artificial intelligence tools have emerged to facilitate address these challenges more effectively, without completely delegating the strategic decision-making that underpins the PRD creation process – in other words, with a human still in the loop. One example is GoogleLM Notebookthat synthesizes grounded raw data or materials to answer questions, thereby accelerating the workflow in creating grounded, usable PRDs.
This article will guide you, based on a beginner-friendly operate case, through the process of using NotebookLM’s features to transform raw, sometimes disordered information into well-established PRD in minutes. Spoiler: It won’t just be a conversation with an AI assistant.
# From messy notes to structured PRD design
Let’s consider the following scenario. You are a newly hired product manager at a startup that wants to develop a recent mobile app called FloraFriend. The app aims to facilitate people stop accidentally killing houseplants.
Your team has assembled a set of three “messy” documents containing descriptions of a potential application:
interview_transcript_matt.txt: 30-minute interview with user Matt, who owns over 50 plants. In his interview notes, Matt claims that existing apps are “too complicated” and make it hard to remember aspects such as “which fertilizer to use.”competitor_research_notes.txt: a gritty list of points compiled after analyzing competing apps such as “PictureThis” and “Planta”, highlighting their firewalls and interface flaws.brainstorming_whiteboard.jpg: Random but somewhat “cool” ideas that the team mentioned during lunch breaks and other casual conversations, such as “Spotify plant playlists”, “watering reminders”, and so on.
Imagine full documents containing all the content described above. Manually converting them into a spotless PRD that ties it all together nicely may seem tedious, right? Enter NotebookLM!
Log in to NotebookLM with your Google account and click “Create a new notebook“. Give your recent notebook a name, for example “FloraFriend PRD“
After creating a recent notebook, you will be welcomed to the main interface of NotebookLM, which looks like this:
NotebookLM interface
One note: This newly created notebook itself is not astute. It’s not regular immense language model (LLM); does not know about plant care or other detailed topics. But we’re going to teach him an “express” master’s degree in this subject, using our messy – but informative for the tool – notes.
Let’s assume you have the three above-mentioned files with content related to a plant care app or other files of your own with raw information. You can upload them to the NotebookLM canvas using the upload button in the main middle section.
Once transferred, you can think of your notebook as something similar to a diminutive, toy-sized device search-assisted generation (RAG) that can start thinking and behaving like artificial intelligence based on the information it has access to. In fact, without asking you, by clicking on one of the uploaded files on the left, NotebookLM generates a concise, well-organized summary of the contents of that file: this is called Source Guide.
Now comes the crucial part. We could just ask in the chat box at the bottom something like “Write PRD” and that’s it. But we want to do it right and provide clear, detailed instructions, and that requires some quick engineering, namely getting the newborn AI to prioritize what we want our PRD to reflect: prioritizing user problems over random ideas generated by the team (without completely neglecting them). Here’s a well-made prompt that works:
I am a product manager at FloraFriend. Based solely on these sources, prepare a draft PRD.
Key Limitations:
1. Prioritize features that address the issues mentioned in interview_transcript_matt.txt.
2. Rule out any brainstorming ideas that don’t directly address the user’s problem.
3. Structure the output using the following headings: problem statement, core features, non-functional requirements (UI/UX), and success metrics.
Try adapting this prompt to your own business problem or operate case. Once submitted, chances are you’ll receive a nice and spotless PRD with key sections such as problem statement, core features, non-functional requirements (UI/UX), success metrics, and so on.
Interestingly, the PRD contains what look like numeric quotes that you can hover over. If you do this, you will see the source pop up (one of the source files):
Before you accept the first PRD as is, remember that the first version is rarely perfect. Continue the conversation to gradually refine it, e.g. if you notice the monetization section is missing, ask: “Based on competitor_research_notes.txt, what monetization models are our competitors using and what should we avoid?Then manually check the results, make sure they’re consistent with the rest of the first draft of the PRD, and incorporate your key monetization insights, either manually or by asking NotebookLM’s AI to do so – if you choose the latter, always check what you’ll get before blindly committing to it. Remember: artificial intelligence can make mistakes!
This is the icing on the cake Sound review section on the right panel (Studio). By clicking on it, you will generate an audio overview of the information contained in the source files. This is a great way to absorb information when reading may be less attractive, such as during your daily commute to work.
# Next steps
This article demonstrates NotebookLM’s ability to generate well-established PRD specifications from raw, messy documents in minutes using very basic steps. Therefore, a valuable next step can be taken Google antigravity to turn the PRD specification into a functional software prototype.
Ivan Palomares Carrascosa is a thought leader, writer, speaker and advisor in the fields of Artificial Intelligence, Machine Learning, Deep Learning and LLM. Trains and advises others on the operate of artificial intelligence in the real world.