Bidraft, the developer of the on-device AI application "Pocket," has reached 1 million downloads just seven weeks after the app's public release, according to a fact-check review that examined primary sources including four original documents preserved with SHA-256 cryptographic hashing. The milestone positions Pocket as one of the fastest-adopted on-device AI apps to emerge from Korea's startup scene.
Pocket is built around on-device AI, an approach that runs AI models directly on a user's hardware rather than in the cloud. This architecture allows an LLM to operate locally on a phone or PC, reducing reliance on network connectivity and keeping user data on the device. The category has drawn growing attention from both consumers and hardware makers as processors gain the capability to handle local inference.
The download figure was confirmed through a review of ten primary sources, with four original HTML documents hashed using SHA-256 to preserve the evidence trail. The verification covered the app's launch timeline and its cumulative download count, establishing that the 1 million threshold was crossed within seven weeks of release โ the seven-week window cited in the claim matches the documented record.
A seven-week run to 1 million downloads gives Bidraft an early foothold in a market where large platforms and device makers are competing to bring locally running AI to everyday users. For Korea's startup ecosystem, which has often ceded consumer AI mindshare to global players, Pocket's adoption curve offers a data point that homegrown on-device AI products can achieve mass-market traction.
Based on the primary sources reviewed, the claim that Bidraft's on-device AI app Pocket reached 1 million downloads seven weeks after its release is True.