tips to accelerate delivery
Everyone is hyping up how artificial intelligence (AI) will replace business analysts, product owners, and product managers. spoiler alert - we're still very far away from that.
That being said, AI can help product teams speed up time to delivery, especially in the early discovery and analysis phase. In this post, I'll share three ways where I have personally used AI to speed up my product delivery.
tip 1: find & summarise information for product discovery
A product owner is always looking for ways to make a product better for customers. Many use IDEO's Desirability, Viability, and Feasibility (DVF) framework to analyse the feature idea. If there isn't an intersection between the three, then it probably isn't worth spending time on. The time taken to run through the framework can be long and complex, involving a lot of data, customer research, and market analysis.
AI can be the data analyst and business analyst to help speed up this process. Before diving in, keep in mind that AI cannot validate the data is accurate or even valid so it's important to focus on the quality of the data and sources it is using. Remember, garbage in, garbage out.
I've built a couple of agents using Claude and M365 copilot to help speed up the analysis and DVF process. Given a particular feature idea, I have Claude scour through Confluence using a MCP to find and summarise relevant information and source the response. I have a separate M365 copilot agent navigate through Sharepoint to review and visualise data.
Instead of spending time trying to search for the right keywords to find the right information, the AI does the heavy lifting. Time can then be spent on talking to the right stakeholders to collate and validate the information.
tip 2: automating tedious and repeatable activities
There are many times during the week where I will do similar activities again and again that I would love to automate. However, it will sometimes take much longer for me to articulate my thoughts into working code than it would take to do the task manually.
I spend a lot of my time data mapping and locking in OpenAPI contracts with other teams. I'd load the OpenAPI spec in Swagger Editor, copy the relevant information onto Confluence, then work with the team to lock in the data attributes and field mappings. Boring, Repeatable, and Tedious - but necessary to ensure both the consumer (my team) and provider (the other team) are on the same page.
Initially, I wrote a Python parser to extract the relevant information from the OpenAPI spec onto Confluence. I spent multiple hours writing and maintaining the code for different specs with varying success.
Once GitHub Copilot enterprise was greenlit, I uplifted my script to a much more maintainable and less error-prone solution. I wrote up basic requirements and test cases then let the AI handle the implementation, updating the requirements or prompt engineering as needed.
This meant that the data mapping exercise went from taking ~30 minutes per spec to ~5 minutes per spec. I was able to focus on getting the contract and field mapping right instead of spending time creating the Confluence page.
tip 3: transcribe, summarise, and capture tasks from meetings
My day is constantly juggling multiple responsibilities: attending (useful) meetings, analysing requirements, getting customer feedback, keeping the engineers well fed with work, and ideating valuable features. The role often requires context switching like crazy, balancing following up on meeting details and outcomes while making sure the product direction is clear and the team are well fed with work.
A little help goes a long way. AI powered recordings and transcriptions can help ease the cognitive load during and post meetings. I like using M365 Copilot to automatically capture meeting summaries and action items, though speaker identification doesn't work well in physical meeting rooms.
To make AI more #worth, implement a workflow to create and automatically assign people tasks in project management tools like Jira.
This approach eliminates manual note-taking, ensures accountability through automated action item assignment, and creates a searchable archive of decisions and discussions that can be referenced weeks or months later.