One gateway to AI models. Your keys, European control.
An OpenAI-compatible endpoint for every project: Claude Opus, Sonnet and Haiku via Anthropic, GPT, Mistral or your own server — on your own keys (BYOK). Plus rate limits, budgets, caching, PII redaction (national ID, e-mail, phone), logs and automatic embeddings into pgvector.
import OpenAI from "openai";
const ai = new OpenAI({
baseURL: "https://host.ulewicz.space/api/ai/v1",
apiKey: process.env.HOSTBASE_AI_KEY, // hbai_…
});
const res = await ai.chat.completions.create({
model: "claude-sonnet-5-5",
messages: [{ role: "user", content: "Summarise this support ticket" }],
});Capabilities
What you get with AI Gateway
OpenAI-compatible
Change only the base URL and key in any SDK. SSE streaming, tools and images are translated to/from the Anthropic Messages API.
BYOK and encryption
Anthropic, OpenAI, Mistral or custom-endpoint keys are encrypted at rest and never written to logs.
Keys, limits and budgets
Gateway keys with scopes, per-minute rate limits and monthly budgets; notifications at 80% and 100%.
Cache and fallbacks
Exact-match response cache with TTL and automatic fallback to a backup model on provider errors.
PII redaction
Polish PESEL numbers (checksum-validated), e-mails and phone numbers are masked before the prompt leaves your project.
Vector pipeline
Pick a table and column — we generate the SQL: vector column, trigger, pgmq queue and an embedding worker.
Use cases
What teams use it for
Assistants and chatbots
Streaming answers, prompt templates and cost control per key.
RAG on your own data
Automatic embeddings of Postgres tables and semantic search without a separate database.
Sensitive data
PII redaction and logs without prompt bodies — for apps handling personal data.
The rest of the platform
Every service shares the same database, authentication and permissions.
Build in a weekend. Scale to millions.
Spin up a free project in 2 minutes. No credit card, no pausing after a week.