AI infrastructure. Memory. Reason. Act.
AIVAX brings models, retrieval and tools into one OpenAI-compatible operating layer, so behavior can evolve without losing the trail of what happened.
Knowledge matters when its path into an answer remains visible.
Approximate cumulative activity since AIVAX launched.
- Token volume
- —
- Token volume (USD)
- —
- RAG searches
- —
- RAG indexes
- —
Loading AIVAX platform activity.
Latest from AIVAX
View all postsThe platform at a glance
Everything you need to put AI to work.
Connect models and tools. Give them knowledge. Understand and manage what happens next.
Act.
From a request to something done.
-
AI gateways
Models, instructions, skills and tools behind one endpoint.
-
Inference & model routing
An OpenAI-compatible API, model choice and your own provider keys.
-
Tools & MCP connections
Connect APIs, MCP servers and model-callable functions. MCP servers can also bring their published instructions and skills into your gateway.
-
Workers & hooks
Adjust context and tool results with your own request-time logic.
-
Web search & research
Find sources, fetch pages and bring fresh information into answers.
-
Media descriptions
Analyze images, audio, videos and files through dedicated media inference.
-
Realtime voice
Build live, two-way voice conversations with your assistant.
-
WhatsApp & Telegram
Connect AI gateways to conversations on your messaging channels.
-
Shell sandbox
Give agents an isolated environment to execute commands and work with files.
-
Image generation
Create or edit images from prompts and reference images.
-
Audio generation
Turn text into spoken audio with text-to-speech models.
-
Audio transcription
Convert recordings and spoken messages into text.
-
Internet research MCP
Bring AIVAX web search and page fetching to any compatible MCP client.
-
Inference MCP
Expose an AIVAX model or AI gateway as a tool for other agents.
-
Media Generation MCP
Discover models and prices, then generate images or MP3 speech from an MCP client.
Memory.
The right knowledge, in the right context.
-
Knowledge collections
Organize documents into searchable knowledge bases.
-
Import & automatic updates
Ingest documents in bulk and reindex content when it changes.
-
Media to knowledge
Turn videos, PDFs and other supported media into searchable knowledge.
-
Semantic search
Find relevant passages by meaning, not just matching words.
-
Reranking & Reflex
Prioritize relevant results, or rank documents without a collection.
-
Grounded answers & citations
Answer from your content and trace results back to their sources.
-
Agentic RAG MCP
Let MCP-compatible agents search your collections on demand.
-
Persistent memories
Keep user context available across conversations.
-
Reusable skills
Write instructions once, reuse them across gateways and load them on demand.
-
Skill learning
Turn a demonstration video into reusable instructions with Teach Skill.
-
RAG in AI gateways
Attach collections to a gateway and retrieve knowledge within the inference request.
-
RAG rules & strategies
Set result limits, score thresholds, rerankers, query rewriting and agent-driven search.
-
AIVAX-managed chat sessions
Keep conversation history on AIVAX for embedded chats and messaging integrations.
Reason.
See clearly. Decide. Stay in control.
-
Decisions & classification
Route requests, assign labels and score content against your criteria.
-
Structured JSON & tool calls
Call tools before returning schema-validated JSON, with repair attempts when needed.
-
Agentic tests
Simulate conversations and evaluate how your gateway behaves.
-
Logs & observability
Inspect messages, tool calls, errors and usage when logging is enabled.
-
Batch workflows
Process records asynchronously and track each result.
-
Usage & costs
Follow consumption, check model pricing and manage your balance.
-
Moderation & guardrails
Screen gateway input and apply shared policies before inference.
-
Account & resource control
Manage keys, limits and platform resources from the console or API.
-
OCR to text or JSON
Extract readable text or schema-shaped data from images and documents.
-
Multimodal pre-processing
Convert media into text for models that cannot natively read images, audio, videos or files.
-
Context truncation
Set context limits, trim older messages and control how much tool history stays in view.
-
Response rendering
Render reasoning, tool activity and answers together as blocks in your chat timeline.
-
Provider routing preferences
Prioritize price, speed or quality among eligible providers for the same model.
-
Text segmentation
Split long text into useful sections for indexing and downstream processing.
Customer stories
Real problems, solved with AIVAX.
How teams put AIVAX to work on the problems in front of them.
From idea to clients in days, not months.
“Every insurance brokerage we work with needs its own assistant on WhatsApp. With AIVAX we stopped building the foundation and went straight to the product. We try an idea, put it in front of a client the same week and see exactly how each assistant is talking to people. Anything outside the rules goes to a human. That gave us the confidence to grow.”
Rafael AltomareFounder, SeguroCor
One assistant for the clinic, another for the office.
“Our partner orthodontists never have time to spare. Now they ask our assistant, right inside our platform, to open a request, check their patients’ appointments or organize the clinic’s day. On our side, I ask for a sales or operations report and get the answer in seconds, instead of waiting for someone to pull the numbers by hand.”
Aline LyraOrthodontist and entrepreneur, Beyou Smile
A support queue that finally makes sense.
“Our queue was a mess: questions about tickets, transfers and refunds piling up on WhatsApp and on the website, with no order at all. Today the assistant answers right away, finds the order, resends the ticket and only calls our team when a person is really needed. It also talks to people interested in an event and helps close the sale.”
Luciano CarvalhoCTO and partner, Zé do Ingresso
Pay with Pix
The only model router that accepts Pix.
Top up your AIVAX balance with Pix, Brazil's instant payment system. No international credit card required, and the same credits cover every model, RAG search and tool call.
- 01Choose an amount in the console, from US$ 3.
- 02Scan the QR code and pay in reais.
- 03Credits land as soon as the payment confirms.
AIVAX Toys
Try a capability before you write any code.
Single-purpose demos that run in your browser and show the API call behind every step. Turn a PDF into Markdown and chunks ready for a RAG collection, or generate images with dozens of models and keep them in collections.
Sign in with your AIVAX account and pay only for what you use, at list prices. Each page shows the cost of every call, and the code is open source.
Read a PDF, then split it for retrieval.
-
01 / Upload
handbook.pdf Text or scanned pages, up to 10 MB.
-
02 / Read
# Leave policy ## Annual leave Employees accrue 1.5 days of paid leave per month. ## Sick leave A medical note is required after three days.POST /api/v1/web/fetchFetch & OCR return Markdown. -
03 / Split
docid: handbook.pdf#1# Leave policydocid: handbook.pdf#2## Annual leave — Employees accrue 1.5 days...docid: handbook.pdf#3## Sick leave — A medical note is required...POST /api/v1/generations/segmentChunks download as JSONL.
Generate images with dozens of models.
Pick a model by its sample image and price, choose a style, an aspect ratio and 1K to 4K resolution, and guide the result with up to four reference images on models that accept them. Collections stay in your browser with the cost of every image, and download as a ZIP.
Prompt A small red fox sitting on a mossy rock by the sea, a white lighthouse in the background, golden hour
-
Realism
-
Anime
-
Vintage Cartoon
-
Cute 3D
Everything the client doesn't have to carry.
Compose knowledge, model behavior and tools behind one OpenAI-compatible call. Each capability stays reusable, inspectable and shared.
RAG
02 / Reason Shape how context becomes a reliable answer.Gateway capabilities
03 / Act Connect models and tools to systems outside the prompt.Tools + providers
A shared path matters when every application can rely on it.
Field guide
Five ways into the system.
Start with the question in front of you: knowledge, execution, evaluation, model choice or cost.