Pika

Create your Pika Agent, then create anything together. https://t.co/Rrakf4WV4F

Pika Agents are a persistent, AI-driven creative partner that users design with their own chosen face, voice, and personality, then talk to through text, voice, or live video to generate images, films, websites, presentations, and other media. The pitch is aimed at creators and operators who are tired of juggling separate prompt boxes and subscriptions, with Pika acting as the orchestration layer over a stack of underlying models including Seedance, Veo 3, Sora, Kling, MiniMax, Pika Video, ElevenLabs, and OpenAI Whisper. The launch matters because it marks a clear pivot for the company from a single text-to-video generator into a full multimodal agent platform. The agent is meant to live where you already work, with connections to Slack, Telegram, WhatsApp, Discord, iMessage, Instagram, LinkedIn, Notion, GitHub, Figma, and Zoom, among others. A new model called PikaStream 1.0 powers real-time video calls in which the agent joins a Google Meet (with Zoom and FaceTime planned), responds with expressive facial movements, and can complete tasks live during the call. Developers can also pull these capabilities into their own products through a Pika Agents API and a public Pika-Skills repo on GitHub. The company was started in 2023 by CEO Demi Guo and co-founder Chenlin Meng, who left Stanford's AI PhD program to build it , and is based in Palo Alto. For founders and investors, the interesting bet here is whether a personified, always-on agent that owns the model routing becomes a more durable interface for creative work than chat or prompt-driven tools.

Tags

Comments (8)