Multi-tenant platform for stateful AI characters. Persona, memory, emotions, life events, proactivity — all managed server-side. Plug in your own data sources so the bot knows your world. You bring your users, we bring the soul.
For chat apps, games, robots, and any digital being that needs to feel alive.
Get started Read the docsTalk to a friendly AI built on the very platform documented here. Persona, memory, emotions — all live. He'll remember you across sessions.
Drop in a paragraph about anyone — friend, late grandparent, fictional character. The engine extracts a calibrated Big Five, seed memories, an emotional baseline, and the social bond to you. Free, ~30 seconds, demo clones expire in 24 h.
A complete personality engine — not a thin wrapper around chat completion.
Big Five, custom traits, persona, intimate-persona (consent-gated). Auto-generated or hand-crafted.
Drop in a Slack export, Gmail mbox, or a free-form description — the engine produces a fully initialised character: calibrated Big Five, 8-14 seed memories with concrete specifics, an emotional baseline, and the right social bond to the user from turn one ("my husband", "old colleague", "late grandmother"). Not a static profile — the full memory decay, emotion wheel and relationship arc kick in the moment chat starts. Read more →
Category-driven TTL, importance bumping on reinforcement, periodic LLM consolidation.
Trust, friendship, conflict, milestones, secrets gated by trust. Configurable progression speed per tenant.
Full 6-emotion wheel, time-decayed needs (tiredness, hunger, loneliness), optional biological cycles.
Bot has its own day — events, mood, schedule. Mentions them naturally when chatting.
Server decides when to reach out and what to say. You decide whether to deliver. Quiet hours, daily caps.
Three-layer safety: pre-prompt guard, post-filter, per-character consent + age gate.
API keys, per-tenant feature flags, full token-level usage tracking and cost dashboards.
Attach RSS, CSV, URLs, plain text at three scopes (tenant / user / character). Bot speaks from sources you approved — not random web search. Read more →
HMAC-signed webhooks for proactive messages, billing changes, milestones. Subscribe once, get pushed forever. Read more →
Server-Sent Events for typewriter rendering. Same memory and state semantics as the regular endpoint. Read more →
Run inference on your stack — OpenAI, Anthropic, fine-tuned, on-prem Llama. Your key never reaches us; Vilow keeps tracking personality and memory in the background. Read more →
No prompts to write, no memory layer to build, no LLM provider to manage.
# 1. Create a user (your own ID — opaque to us) curl -X POST https://api.vilow.dev/v1/users \ -H "X-API-Key: ck_live_..." \ -d '{"external_id":"alice"}' # 2. Auto-generate a character curl -X POST https://api.vilow.dev/v1/users/alice/characters \ -H "X-API-Key: ck_live_..." \ -d '{"generation_mode":"auto","gender":"female","locale":"en"}' # 3. Chat — reply, facts, emotions, relationship deltas, all in one call curl -X POST https://api.vilow.dev/v1/chat/alice/42/send \ -H "X-API-Key: ck_live_..." \ -d '{"message":"hi, how was your day?"}'
Guides for the integration patterns most clients reach for.
Five steps from sign-up to your first stateful chat. Big Five test, character creation, first message.
Read → GuideAttach RSS, CSV, URL, or plain text. Three scopes (tenant / user / character). Bot speaks from your data.
Read → GuidePush instead of poll. HMAC-signed events for proactive messages, billing, milestones.
Read → GuideServer-Sent Events for typewriter rendering. Browser and Node examples.
Read → GuideKeep control of your LLM stack. Vilow returns the styling prompt; you run inference on your own provider or on-prem.
Read → Reference14 tenant-level toggles — relationship pace, intimate mode, life intensity, AI disclosure, more.
Read → ReferenceAuto-generated Swagger UI with every endpoint, schema, and "Try it out".
Open →Start free, scale to millions of messages. Pay-as-you-go overage with pre-paid balance.
Try it without a card.
For indie projects and side companions.
For real products and serious traffic.
Studios, robots, on-prem deployments.