bitdrift, the mobile observability platform spun out of Lyft and installed over a billion times across hundreds of millions of devices around the globe, today announced bitdrift AI: a real-time, full-fidelity observability system that lets AI agents query mobile user behavior and act on it autonomously. Purpose-built for mobile, bitdrift AI delivers unsampled data on the edge that enables agents to triage, investigate, and resolve issues without having to wait for slow release cycles or rely on limited, sampled data.

“Agentic AI has changed software engineering, but agents can only be as smart as the data they see,” said Peter Morelli, CEO of bitdrift. “Traditional observability solutions simply weren’t built for the unique and often less-than-ideal conditions involved with mobile, whether it’s a bad network, rainy weather, or a misplaced button within the UI. bitdrift AI is the first solution that gives AI agents direct access to full-resolution mobile data, without sampling, e.g. live sessions, targeted cohorts, and on-demand telemetry. That means 100x faster investigations, fewer false leads, and fixes before users notice a problem.”

Key Highlights of bitdrift AI

• Full-resolution, unsampled data: bitdrift captures 100% of on-device telemetry in real time, avoiding the blind spots caused by sampling and delayed SDK syncs.
• Autonomous agents: bitdrift AI exposes workflows, charts, issues, and captured sessions via a CLI, a public API, and customized Skills so that AI agents and engineers alike can query millions of devices in real time, create instrumentation on the fly, and iterate in tight feedback loops without waiting for app releases.
• Precise targeting: local capture buffers and server controlled targeting enable agents to gather only the context they need, preserving model context windows and reducing irrelevant data.
• Built by experts: the founding team’s experience managing mobile applications at scale for both Twitter and Lyft informs an architecture designed for billions of installs and real-world device variability.

bitdrift AI works by capturing telemetry on-device (logs, traces, and session context) into an unsampled ring buffer, streaming it in real time to the control plane, and exposing programmatic access via CLI and public API. AI agents can create and manage workflows in code, retrieve full-fidelity sessions on demand, and integrate bitdrift into deploy pipelines all without releasing a new app version.

“With bitdrift, we’ve brought infinite scale to problem resolution," said Valerii Kuznietsov, senior staff software engineer and mobile engineering lead at ThredUp. "Because we have enough data and AI to help us, we can help 90-95% of customers experiencing minor problems instead of the 20-30% that we could support before.” The bitdrift platform has already been installed across billions of devices and can support hundreds of millions of concurrent streams. Beta users of bitdrift AI report a 10X improvement in MTTR.

bitdrift has raised a total of $15M in funding from investors such as Amplify Partners, 01 Advisors, Primeset, and Lyft. bitdrift AI is available today – developers and engineering teams can learn more at https://bitdrift.ai/.

About bitdrift

Query reality with bitdrift: mobile observability built for the real world. Traditional crash reporting and RUM tools sample away the signal, leaving teams with tunnel vision on the 1% of issues they can see. bitdrift captures everything — unsampled, on-device, across 1B+ installs — and surfaces what matters. bitdrift AI enables AI agents to query that reality directly, so teams fix the right things before users feel them. Learn more at https://bitdrift.io/.

View source version on businesswire.com: https://www.businesswire.com/news/home/20260819980197/en/

“Traditional observability solutions simply weren’t built for the unique and often less-than-ideal conditions involved with mobile."