The Companion API is the highest-level entry point into Spike. You define who the companion is — name, personality, backstory, and visual character. We manage everything required to bring it to life across every modality: consistent visual generation, voice synthesis, real-time avatar rendering, and persistent per-user memory.
This is the right starting point if you want to ship a complete AI companion experience without assembling individual services.
What We Manage
| Capability | What that means |
|---|---|
| Visual identity | We generate and maintain a consistent character appearance used across images, avatar video, and media |
| Voice | End-to-end voice pipeline — speech detection, transcription, inference, synthesis |
| Avatar | Animated talking head, real-time or on-demand, visually consistent with the companion |
| Memory | Per-user context that persists across sessions — the companion remembers every user |
| Conversation | Session lifecycle, message history, and companion context applied to every response |
What You Control
- Who the companion is — name, backstory, personality
- The experience you build on top
- Which modalities you expose to users (text, voice, video, or any combination)
- How users interact with your app
Entry Points
The Companion API has two primary interaction modes:
Conversational — Text-based chat. Send messages, receive responses. The companion applies its personality and draws on its memory of this user on every turn.
Presence sessions — Real-time or on-demand voice and avatar. The companion speaks, listens, and appears with its own face and voice. Suitable for voice-first interfaces, video companions, or ambient always-on experiences.
Next Steps
- Creating a Companion — define identity and visual character
- Chat & Memory — conversations and persistent context
- Voice — end-to-end voice pipeline
- Avatar — talking head, real-time and on-demand