Qelos is the company building mandatory application tools and services for future-ready apps.
We started with the Qelos application-layer gateway — and grew a focused toolkit around it: AIDev, Prici, TestAI, and Better MCP. Open source at the core, managed where it matters.
Hi, I'm David Meir-Levy, the creator of Qelos. After more than a decade as a Tech Lead and Senior Developer, I kept seeing the same shape every time a team started a new product: months spent building authentication, multi-tenant data, role permissions, audit logs, and an admin panel — before anyone could touch the actual product.
With AI in the mix, that gap got worse. Teams would bolt LLM features on top of an already-fragile backend, where the AI didn't share auth, didn't respect workspace boundaries, and couldn't be audited. The "AI part" lived in a separate world from the rest of the app.
Qelos is our answer: an AI-first application layer gateway. One managed instance exposes a RESTful API, AI agents, authentication, multi-tenant workspaces, permissions, and an events log — all reachable through a unified SDK and per-framework middleware. You stay in your own framework. Qelos plugs in as middleware: the @qelos/integrator-* package for your stack reads the request, attaches the authenticated user, workspace, and SDK client, and lets your handlers focus on application logic.
Your app talks to the same Qelos instance that hosts your admin panel — so data, auth, and AI all stay in sync. You define your data model with blueprints in the admin panel, and CRUD endpoints, permission enforcement, and audit logging are generated for you.
We sell that gateway as a managed, hosted service. You get a tenant; we run it. You ship features.
As we built Qelos, we kept hitting other mandatory pieces of modern app infrastructure — and turned each one into a focused, standalone tool. Today Qelos (the company) maintains the Qelos gateway alongside four sibling projects: AIDev (and its hosted counterpart Auto Qelos) for automated development, Prici for entitlements and pricing-limit enforcement, TestAI for testing AI skills and MCPs, and Better MCP for managing MCP servers safely. The full catalog lives at /products.
Started in web development, building a strong foundation in JavaScript and Node.js — and noticing the same application-layer plumbing appear in project after project: auth, tenancy, permissions, audit, admin panels.
As a Tech Lead across multiple startups and enterprises, I watched teams rebuild the same backends from scratch — and then bolt AI on top as an afterthought, untethered from auth or tenancy.
AI-first products needed something different: a managed application layer where data, auth, permissions, and AI agents share one context — reachable through middleware, not a rewrite.
Founded Qelos as an AI-first application layer gateway: one managed instance exposing a RESTful API, AI agents, auth, multi-tenant workspaces, permissions, and an events log through a unified SDK and per-framework middleware.
Shipping hosted Qelos tenants and @qelos/integrator-* packages so teams can stay in their own framework — Express, Nest, Next, Fastify — and still get a fully managed backend in place by the end of the day.
Open-sourcing the auxiliary tools we had been using internally to ship Qelos itself: AIDev, our executor for developer tasks, and Prici, our entitlements and pricing-limit enforcement service. Both released as standalone OSS projects under the Qelos umbrella.
The current chapter: public launch of TestAI (a framework for testing AI skills, MCPs, and subagents) and Better MCP (an MCP aggregator with auto token-shrinking and PII guardrails), plus Auto Qelos as the hosted automated-developer product on top of AIDev. Qelos.io becomes the catalog of mandatory tools for future-ready apps.
Every tool drops into your existing stack as middleware, a library, or a service — no rewrite, no proprietary runtime. Applies to the gateway and to every OSS sibling.
AI agents share the same auth, workspace, and permission context as the rest of your app — not a sidecar. The same principle drives AIDev, TestAI, and Better MCP.
Your app, admin panel, AI agents, and entitlements all talk to the same managed backend, so data, auth, and AI never drift.
Unified SDKs, typed clients, clean middleware, and OSS by default. Your handlers focus on application logic — nothing else.
Every product in our toolkit is open source. Where it makes sense, we offer a managed version on top.
Skip the backend rebuild. Spin up a hosted Qelos tenant, drop the integrator into your framework, and ship features by tonight.