These pieces are about automating the work people actually do, not the clean version from a demo. They cover status reports nobody has to write, meeting transcripts that become logged decisions and tasks, internal tools that used to take too long to build, an email assistant with its limits written in code, and what a company keeps when someone leaves.
Each one works because someone involved knows the workflow well enough to know what breaks if you get it wrong.
Companies that stop hiring junior developers are taking on a business risk. Seven reasons I'd still hire them, and the work juniors can do now.
Before leaving Verizon, I built my team a searchable site from seven years of my emails, Slack messages and documents, and left it behind. I explain why that record belongs to the company, and the pay question its lasting value raises.
Most meetings now end with a transcript that sits unread in your inbox. An AI tool can check those transcripts on a schedule against rules you set, draft the next step, and bring you anything that needs your judgment.
I built an email assistant in a few hours that answers mail sent to assistant@boobridge.com, from my salad recipe to article and music picks. Its limits run in plain code, not AI judgment, and anyone can email it to try it out.
Companies like Atlassian and Glean are building knowledge graphs that connect a company's email, docs and chat so AI can read all of it. A graph shows what you can see, not what you should use, and experienced people have to write that judgment down.
One ten-minute decision about personalizing a busy web page touched three teams, and dozens like it happen every day. I describe how AI and connected systems could shrink that legwork, and why it only works when you know the workflow deeply first.
Company wikis, workflow tools and weekly reports have long been the roughest software at work, because custom tools were never an option. AI now lets the people who do the work build them, and I cover when a tool needs a full development team.
Fred, a merchandiser on my team, shared a project workspace with an AI agent built in, and his manager and I got fully caught up without a status meeting. His meeting transcript updated the project brief, and anyone can ask it questions anytime.
Asking for a server at work set off a flood of registration, security and ticket emails. My AI assistant read them all, wrote a PDF summary of what was done and still due, and sent it to my team, so every form got done properly.