2025
TwinSight
Winning entry in a NIST-funded national challenge on turning smart-city digital twin data into decisions people can act on.
- Status
- Proposal and methodology
- Evidence
- Independently documented
- Claims checked
The challenge asked how digital twin data could drive intelligent, real-time decisions in urban environments. Most answers to that question concentrate on the model — its fidelity, its coverage, the number of systems it ingests. TwinSight put the weight on the other end, where a person has to form a judgment and commit to an action.
The situation
Cities have invested heavily in digital twins that stream continuous sensor data, but most of that investment supports passive monitoring. The people who actually have to act — planners, city leaders, and residents — rarely have a way to ask what would happen if they changed something, or to judge how much confidence a projection deserves.
My role
I led Team Twinsight through the CivX Data to Decision Challenge, which ran from April to June 2025, was sponsored by the Griffiss Institute, and was funded by NIST. My contribution was to the research concept and the methodological approach. Alyssa Borders, Cristin Irwin, and Erik Ness contributed the remaining technical and domain work.
The initial assumption
We started from the framing the field usually starts from: that the limiting factor is data, and that integrating enough sensor streams into a sufficiently complete twin would produce better decisions.
What the evidence changed
Working the problem moved the emphasis from integration to interpretation. The design that resulted treats the decision interface as the real bottleneck. The same underlying model is presented differently depending on who is looking at it and how they reason, and the value of a simulation layer is that it lets someone test a scenario rather than watch a feed.
Record / 1
What was produced
- A competition entry describing TwinSight as a semantic simulation layer over existing smart-city digital twins
- A proposed architecture pairing a semantic data lake with agent-based models to support scenario testing
- A decision-support interface concept that adapts to a user's role rather than showing everyone the same dashboard
- A pitch delivered to the challenge judges at the closing virtual event
- A panel appearance at the 2025 Photogrammetry, 3D Visualization, and Lidar Community of Practice Conference at NGA Washington in Springfield, Virginia, in August 2025
Record / 2
Evidence available
- The organizer's public project gallery lists TwinSight as the challenge winner and names the full team
- The organizer's published highlight video states that Team Twinsight, led by Dr. Sazma, was named the winner and that the team received a $20,000 grand prize
- The same video documents the challenge's NIST funding and Griffiss Institute sponsorship, and the team's invitation to the 2026 Winner's Circle Showcase in Rome, New York
Credibility requires a boundary
Limits
TwinSight remains a proposal, a pitch, and a methodological exploration. It was not built into a deployed system, has not been piloted with any city, and has no users. The $20,000 prize was awarded to the team, not to me personally. Winning a design challenge shows that the concept persuaded expert judges under competition conditions; it is not evidence that the approach works in a live municipal setting. The August 2025 conference appearance is recorded in my own files rather than on a public page, as the 2025 agenda is no longer online.
Transferable value
If you hold a rich data asset that is not changing anyone's decisions, this is the work: reframing the question from what a system can display to what a particular person needs to decide, and specifying the test that would tell you whether it helped.