AI product launch
Taking a flagship AI product live after an abrupt pivot
Sequencing dependencies across teams, one meeting and one source of truth, and a staged roll-out that started at a 2,200+ attendee conference.
Written by
Bianca Lomeli
Technical Program Manager
Some projects arrive on schedule. This one arrived all at once.
In early 2025, Workiva was shifting to become AI-native, and Intelligent Finance, one of its three flagship AI products, landed on my team abruptly, just as we were about to start another project. It had a conference date to hit: Amplify 2025, Workiva's customer conference, with 2,200+ attendees.
It was also a new way of working for the whole company. Other teams had to build their parts before mine could move, and senior leaders were watching every decision. I led delivery for the 9-person squad that owned it, working alongside ML specialists and industry experts.
Pivot fast, then protect the plan
We set aside the project we were about to start and re-planned around one date that couldn't move. From there, the job was to keep the plan realistic, so the team could hit the conference without cutting corners.
Sequence the dependencies
Intelligent Finance was a new way of working for the whole company, so I sequenced dependencies across the teams building its parts. Releases were often multi-step: one team moved, then the next, then mine again. Each step had to land before the next could start.
One meeting, one source of truth
Senior leaders were close to the work, questions came from several directions, and different versions of the status were circulating. I set up a weekly meeting for everyone involved. My team gave its update, then each team reported in turn. That kept every team accountable for its part, without line authority over them, and surfaced problems before they grew. It was recorded for anyone who couldn't join.
Alongside it, I built a Confluence hub with RAID logs, decision logs and dependency tracking. It gave 50+ stakeholders, including the CTO and VPs, one place to check status and decisions.
Roll out in stages
We didn't launch to everyone at once. We planned three progressive roll-outs, controlled with LaunchDarkly feature flags and a dual sign-off workflow for each release, so access widened step by step. Amplify was the first stage.
The result
Intelligent Finance went live in about 8 months, in time for Amplify 2025 and its 2,200+ attendees, as the first of three progressive roll-outs. I then took it to general availability, and took its second phase through approval for launch. My team also built its AI peer benchmarking, which Workiva later launched as its Benchmarking Agent.
The hub and the weekly meeting cut the questions coming to the team, saved me about 20 hours a month finding information, and gave management one place to refer to, which helped speed up decisions.
What I'd tell anyone launching under pressure
Fix the date, then protect the plan.
Map the dependencies before they block you.
Have teams report in turn, so problems surface early.
Keep one source of truth, and point every question to it.
Widen access in stages, so you can stop if something goes wrong.