DeepBrain AI lets you video chat with deceased loved ones

What if you could jump on a video call with a deceased loved one and have an interactive, 30-minute conversation?

That chilling premise straight out of Black Mirror's Be Right Back is now a commercial reality. Seoul-based artificial intelligence company DeepBrain AI has rolled out a paid service called Re;memory, which builds interactive virtual humans of departed family members to let relatives talk with them again.

The concept even caught mainstream industry attention, earning DeepBrain AI an honoree nod at the CES 2023 Innovation Awards in the Virtual and Augmented Reality category.

A 400-inch screen and a final unfinished conversation

Most grieving families rely on old photos and home videos to remember lost relatives. While meaningful, those media formats are strictly one-way.

Re;memory attempts to close that loop. The company gathers a family's digital archive of pictures and video recordings, pairs it with a structured pre-interview process, and spends several hours processing the data.

Under the hood, DeepBrain AI combines image synthesis with conversational technology. The system mimics non-verbal cues like lip-syncing, natural facial expressions, and physical gestures, while running speech recognition and natural language processing to handle real-time back-and-forth dialogue.

Clients step into a private memorial showroom equipped with a massive 400-inch display and a dedicated sound system for a 30-minute reunion session.

Beyond real-time conversations, the platform can generate scheduled video messages for milestones like birthdays and anniversaries, craft personalized biographical videos, and even allow the AI persona to attend an actual memorial service.

DeepBrain AI did not start in the bereavement market. The firm previously built AI news anchors for a national South Korean broadcaster and generated an AI avatar for Yoon Seok-yeol during his presidential campaign.

Why it matters

Turning deep learning toward grief tech makes commercial sense because the emotional pull is undeniable. For families carrying unfinished business or sudden losses, a personalized 30-minute session offers emotional closure that static photo albums simply cannot provide.

Yet the technical barrier remains high. The service inherently demands a substantial treasure trove of existing photos and footage alongside family interviews. If a loved one left behind sparse digital footprints, the output risks landing squarely in the uncanny valley.

The deeper question is psychological. Immersing grieving relatives inside a private theater with a 400-inch screen might offer temporary relief, but it could just as easily prolong denial or create unhealthy attachments to a synthetic avatar. DeepBrain AI has proven that recreating human likeness in real time is technically viable, but the ethical guardrails around digital resurrection are still completely unpaved.