Automotus conducted a pilot scoping exercise with the Tunnels, Bridges & Terminals (TB&T) team, identifying a potential deployment at the Midtown Bus Terminal (MBT). Using its camera-based technology, Automotus proposed a pilot that would automatically track which buses use the terminal and when—identifying each bus by carrier as it enters and exits, counting trips and measuring how long buses are within the MBT, comparing that against carriers' official schedules to understand discrepancies, how long buses stay in the terminal for arrivals and departures, and how many times buses recirculate in various roadway conditions.

MADISON was evaluated on a Global M4 street sweeper at MTA's Randall's Island fleet facility. Technicians used the platform to ask questions about maintenance history, equipment specifications, and fault codes. MADISON retrieved answers directly from technical manuals, service records, and other maintenance documentation, achieving over 98% accuracy in identifying the correct information while citing the specific source for every response. This provided technicians with complete transparency and traceability back to the original documentation.

Hazel AI deployed their Solicitation Builder module with LIRR to support drafting technical scopes of work, a key process in the procurement lifecycle that takes hours to complete. In the three weeks of user testing, Hazel AI trained seven LIRR technical project managers, who used the tool to draft 15 technical scopes of work. On sample scopes, Hazel cut drafting time by 40%—exceeding its 10% target—and included relevant requirements, with applicable standards and specifications, which helped develop a more complete Scope of Work for the user to confirm or expand. 100% of testers affirmed they would like to continue using it.

In A Blink ran two live demonstrations, including at the Staten Island Ferry Bus Loop and at MTA offices at 2 Broadway, transferring 95 GB consisting of 144 hours of original video footage from 12 high-resolution cameras. Data was transferred within 5 minutes and 20 seconds with zero errors or data lost, reaching a net effective bandwidth of 2.45 Gb per second.

Ironloop deployed its read-only platform on MTA-provisioned hardware inside MTA's isolated OT cybersecurity test lab, fully air-gapped (meaning no connection) from the internet and from MTA's production network. Over the eight-week pilot, MTA's cyber team tested it against configurations from nine vendor families, including Rockwell, Siemens, and Cisco devices, and ran OT-safe network discovery across the lab's switches, firewall, and PLCs. All four POC goals were met. The platform ingested and analyzed configurations from legacy PLCs (Programmable Logic Controllers, the industrial controllers that operate physical equipment like signals and switches) and network devices, mapped findings to NIST CSF 2.0 (the National Institute of Standards and Technology's Cybersecurity Framework, a widely used federal standard for identifying and managing cybersecurity risk), and produced a repeatable configuration baseline

Ontra conducted a proof-of-concept with both the MTA LIRR and NYCT: For LIRR, Ontra generated weekend timetable options for the Huntington and Ronkonkoma branches. Ontra ingested LIRR GTFS data (standardized files containing route, stop, and schedule information used across transit systems), along with timetable data, station-to-station demand, runtime/dwell assumptions, stop patterns, and service constraints to minimize passenger travel time, improve schedule regularity, and identify implementable schedule changes, including lower-disruption express-trip options. For NYCT, Ontra modeled and redesigned the Brooklyn Bus Network. Ontra ingested NYCT bus ridership/origin-destination/path data, GTFS data, regional demand inputs, road network data, and operating constraints to improve ridership, convenience, equity, and efficiency.

VIATechnik deployed Voyager for the Port Authority of New York and New Jersey's JFK Bergen Substation and the Bayonne Bridge, integrating seven and eight data sets, respectively. The deployment covered 838 assets at JFK and 1,985 elements at Bayonne Bridge, and was tested across 12 the Port Authority of New York and New Jersey users. Data ingested included design and model data (e.g., revit models, CAD files), asset and element registries, surveys, scans, and more. Voyager supported facility teams to find information and make more effective maintenance and operations decisions. 100% of surveyed users found Voyager can support challenging tasks like answering detailed questions about a facility, asset, space, or physical element, without needing to track down separate drawings, records, or subject-matter experts.

Voicd tested its platform against 29 MTA and 9 the Port Authority of New York and New Jerseyaudio samples spanning customer service, public safety, and multilingual scenarios. MTA and the Port Authority of New York and New Jersey users focused on use cases of deepfake and synthetic speech detection, speaker authentication and voice biometrics, lexicon and content analysis, and stress and emotional state detection. The platform correctly classified 89.7% of the 29 MTA samples and matched enrolled speakers with 100% accuracy, while PANYNJ testing achieved 100% accuracy across all four evaluated capabilities based on 9 audio samples, exceeding the 80% target.