Snailup is SolvoNova's own AI platform: one self-hosted REST + MCP surface for chat, retrieval, speech, vision, and media generation, running on our own AI Processor grid. Studio is Snailup's own browser-based front door, not a separate product, sign in there to use the platform directly.
Snailup owns AI endpoint discovery and health, response normalization, durable job state with retries and cancellation, and artifact storage. Every product keeps its own domain logic: legal RAG policy, chatroom moderation, tenant rules, and calls Snailup for everything AI, instead of standing up a parallel stack.
Our legal practice management product: case management, drafting, and a hybrid legal RAG pipeline for research, built directly on Snailup's inference layer.
Every future SolvoNova product reaches the same primitives through one REST + MCP surface, never a bespoke AI stack.
Every capability below is backed by a model actually deployed on the AIP compute grid today.
Resident large-language-model routing between a fast general model and larger code-specialist models, picked per request category.
Text embeddings for semantic search, plus a dedicated BGE reranking sidecar to reorder retrieval results by relevance.
Hybrid BM25 + vector search over an authority-classified legal corpus (constitution → statute → regulation → case law → secondary), fused and reranked before context assembly.
ColPali visual embeddings for document and image retrieval, no separate OCR pipeline.
Faster-Whisper transcription, resident on the compute grid, powering live transcription for partner products.
Fast neural TTS, plus expressive speech and singing across a set of configured voice presets.
A 4-step diffusion model tuned for a few seconds per image after warm-up.
Text/image-to-video, FP8-quantized for memory headroom, producing short clips in a few minutes.
Short instrumental clips plus a full composition/lyrics/mastering pipeline built around them.
A task/agent grid matching submitted work to registered machines and AI agents by declared capability.
Multi-agent collaboration over shared chatrooms: an orchestrator decomposes work and recruits specialists as needed.
One set of validated command implementations, exposed as a stable REST API for products and an MCP server for agents.
AIP, AI Processors, is our own compute fabric, not a rented one. Every capability above runs on hardware we operate directly, scheduled across the grid for the most efficient use of every unit as the fleet grows. No inference call leaves our own network unless a product explicitly opts into a fallback.
Separate companies that connect into the platform. Not Snailup products, real customers and integration partners.
A separate company and a Snailup customer. Jammin integrates our chat, moderation, and transcription services to power pro chatrooms and instant messaging between agents and humans, and call transcription for BuildMyCase.
Learn more about the company behind Snailup, or sign in to Snailup to start building.
