Apps, MCP servers, and AI tools grown in the open. The first are live now — the AI Workflow Builder and the Agent Marketplace, gathered under the Foundry hub. Sign in with your Nyquest Pro account to use them; experiments still on the bench take early-access requests.
The front door to every live ZoidLab app. Sign in once with Nyquest and launch the Builder, the Marketplace, and whatever ships next.
Drag AI, logic, and integration nodes onto a canvas, wire them into a workflow, run it live, and deploy it as a webhook or schedule — all on the Nyquest runtime.
Turn documents and structured data into governed AI knowledge. Citation-first RAG with chunk inspection, retrieval testing, and evaluation — every answer shows its receipts.
The AI policy engine that answers one question: is this AI action allowed? Define, test, enforce, and audit AI usage policies with a deterministic decision engine — not an LLM's guess.
The AI cost optimizer. Break spend down by model, provider, and project, set budgets with live alerts, and run a savings simulator that recomputes your real usage on a cheaper model.
The model benchmark lab. Run your prompts across models on the live relay and measure speed, cost, and quality with real calls — real numbers, not marketing.
The AI evaluation lab. Score your AI's answers against your own criteria with an LLM judge and get a clear production-readiness verdict from real relay runs.
The control plane for how AI remembers, forgets, retrieves, expires, and protects information. Inspectable, governed, tenant-aware memory.
GitHub for enterprise AI prompts — write, test across models, version, govern, approve, and export prompts as reusable assets.
Browse, test, install, clone, and publish reusable AI agents — an app store where every listing shows what the agent does, what it can access, and what it may cost, up front.
See, extract, structure. Upload an image, define the fields you want, and get schema-shaped data back from a real vision model — with confidence, evidence, and PII/secret risk flags.
Design a voice agent — persona, goal, guardrails — then put it on a real simulated call against a caller scenario. Every turn is a live relay completion; every transcript is scored by an LLM judge.
A control plane for Model Context Protocol connectors. Register one, discover its tools with a real MCP handshake, gate which tools may run, test-call them for real, and freeze immutable versions.
Compose swarms of role-specialized agents connected by typed handoff edges, then run them for real on the live relay. The handoff contract is enforced, runtime is bounded, and every run is replayable.
Define an extraction schema of typed fields, point it at any document, and get back clean JSON with every field validated. Real relay extraction at temperature zero, billed to your own Nyquest wallet.
Spin up synthetic classification examples, QA pairs, and structured records from a short spec — then export them straight into a ModelBench dataset, an Eval test set, or a RAG QA set. Real generation, deterministically validated.
Upload a CSV, ask a question, get an answer and a chart. Safe by design: the AI only writes a validated analysis plan — the app runs it in pure Python, so the model never touches your data with code or SQL.
An always-on research agent that watches markets, reads filings, and writes up what it finds — thinking around the clock.
Drop multi-model AI into any WordPress site — content, chat, and automation wired to the Nyquest relay.
Your living room is the battlefield. A vertical slice is flying in the lab — not ready for outside hands yet.