Steering by Probability
Using AI to build the view leadership actually needed.
Every executive asks the same question, when will this be done, and most organizations answer with a date somebody made up. This session is what we built instead.
From the session
At an insurtech company I used the same AI agents that were writing the code to build what leadership actually needed. Here is the short version of what that took.
A probability, not a promise
A live picture of what was being delivered, and for every major initiative, the odds it lands by its forecasted date. Rebuilt daily out of the work teams were genuinely finishing, so nothing rested on anybody’s opinion of how it was going.
I didn’t write the code
I described what leadership needed and directed the agents that built it. The per-team views first, then the portfolio rollup, then the steering page itself. Restructuring how the company tracked its work is normally the project nobody wants to start, and it took a couple of hours.
The data had to be rebuilt first
A forecast is only as good as the system underneath it. The company’s own records of when work finished weren’t trustworthy, so we rebuilt the measurements out of the code itself and cross-checked them independently before anyone steered by them.
The agents made the practices stick
This is the part I didn’t expect. Limits on how much gets started at once, work sized against what the team actually delivers rather than a guess at hours, rules the agents can’t talk their way past. Getting teams to hold that discipline is usually the hardest part of my job, and here it was built into the tools they already worked in.
Monster Mail
Every month or so I send out a real case study, one small thing you can put to work that week, whatever I've written lately, and where I'll be teaching and speaking next. That's it, and you can leave whenever you want.
Thanks! The next one goes out when I've got something worth sending.
Email info@lilagilemonster.com and I'll add you by hand.
The forecasting behind this session is most of Applying Metrics for Predictability, which I teach as an accredited ProKanban.org course.
See upcoming classes →