Most LinkedIn series I've seen die after post five. The author starts with energy, posts a few times, then trails off. Readers who were following lose track. The whole effort becomes noise.
I want to avoid that.
So before I publish post two, I'm putting the entire roadmap out here. Every post I plan to write, in order, with what it's about and why it matters to the overall arc.
You can see the whole thing upfront and decide whether you want to follow.
The project, in one paragraph
A fictional mid-sized Canadian P&C (property and casualty) insurance company is drowning in claims intake. 500 FNOL (First Notice of Loss) calls per day. Each one takes a call centre agent 3 to 5 minutes to triage by hand. That's 25 to 40 hours of agent time disappearing every day, just on the front door of the claims process. Agents are burning out. Severity gets misclassified. Fraud slips through. The company wants to bring AI into the workflow.
I'm the BA on the project. My job is to make sense of all of it, talk to the right people, write the artifacts, build the prototype, test it, and ship it. Real BA work, end to end, in public.
Why all 32 posts
Most BA content online jumps straight to the templates. Here's a BRD template. Here's a user story template. Download and fill in the blanks.
That's not what BA work actually is. The templates are the easy part. The hard part is the thinking that goes into them. Why you pick this objective and not that one. How you spot the assumption nobody wants to name out loud. When to push back on a stakeholder. What to do when the metrics don't add up.
I want to show all of that. The 32 posts let me slow down on each piece of the work and walk through how I actually arrived at it, not just what the final version looks like.
The templates are the easy part. The hard part is the thinking that goes into them.
The eight stages
The series moves through eight stages, mirroring how a real BA project unfolds. Each stage is a chunk of related work. Posts inside a stage build on each other.
Before any artifact, a BA has to understand the world they're walking into. P&C insurance basics, what FNOL means, how the policy lifecycle works, the questions a BA asks in the first week.
The permission document. Problem statement, business objectives, success metrics, and the trap of measuring soft outcomes like agent experience. Five posts because this is where most projects get killed.
The sections most BAs rush through. Scope thinking as a real skill. Out-of-scope as the political protection layer. Assumptions stated honestly, not hidden. Stakeholder mapping that puts the actual users where they belong.
Current state and future state process maps. BRD vs FRD. User stories and acceptance criteria. Edge cases that nobody trains you to surface but everyone pays for when they're missed.
A working FNOL triage workflow on n8n connected to Google Cloud Vertex AI. Screenshots, JSON, prompt design, the surprises. Most BAs never build the thing. I think they should.
UAT for AI is different than UAT for normal software. Designing test cases that cover normal, edge, and adversarial inputs. The defect log. The full AI evaluation framework with Hawthorne effects, drift, and parallel runs in context.
Post-launch monitoring and dashboards. The change management work that decides whether the agents actually adopt the system. Jira and Confluence setup, which is boring but matters.
The honest retrospective on doing this in public. What worked, what didn't, what I'd do differently. And what's next for me.
Pace and timeline
Roughly two posts per week. At that rhythm, the full series runs about four months. Some weeks will be three posts, some weeks will be one. I'll skip rather than ship something half-cooked.
Each LinkedIn post will link back here. The LinkedIn version is the shorter cut. The full version, with the thinking and the artifacts and the parts that didn't make the social cut, lives on the site.
One last thing
This is a hypothetical project. I'll do my best to simulate the parts a text post can't fully capture, like in-person stakeholder meetings, phone calls, whiteboard sessions, the back-and-forth of a real workplace. I'll call out where I'm simulating so the demonstration stays honest.
If you're a BA who's done this kind of work before, or someone staring down an AI project at your company with no idea where to start, I'd love to hear from you as the series unfolds. The point of doing this in public is the conversation, not the lecture.
Everybody thinks the trick is the question list. Turns out the question list comes last. Who you sit across from comes first.
