Nhost gives developers Postgres, GraphQL, auth, storage and functions as one open-source stack. pylo gives your whole team one system: data model, permissions, flows, admin UI, forms and analytics, hosted in Germany.

TL;DR

Nhost is an open-source backend from Sweden: Postgres with a GraphQL API from Hasura, plus auth, storage, serverless functions and custom containers, in the cloud or self-hosted. Developers get a well-integrated stack and build the app logic, the permissions and every screen on top. pylo is a backend platform where data model, permissions, flows, admin UI, forms, events and AI access are one integrated system, for developers and the rest of the team. If you build a web or mobile app on Postgres and GraphQL and want open source with strong auth, Nhost is a great choice. If you would otherwise build automation and a back office around it, pylo gives you all of that from day one.

What is pylo?

pylo is a Backend-as-a-Service platform from Germany that you can use to build the operating system for your company. You model entities, fields and relations in the admin panel or through the API, and pylo generates a typed GraphQL API instantly. The same permission model governs that API, the admin UI, forms, flows and the MCP server. pylo flows run business processes and automations on top of your own data model and connect to the tools you already use. With pylo you build apps, automations and products on one coherent data base across your whole team.

What is Nhost?

Nhost is built by Nhost AB in Stockholm and is open source under the MIT license. Every project gets a Postgres database with a GraphQL API powered by Hasura v2; in 2026 Nhost introduced Constellation, its own GraphQL engine, which is still in alpha. Nhost Auth covers email and password, magic links, SMS and email codes, WebAuthn, OAuth providers and MFA. Storage comes with a CDN, image transformations and virus scanning. Functions run JavaScript and TypeScript, Nhost Run deploys custom containers in any language, and Graphite adds AI features on pgvector. The CLI runs the whole stack locally, and you can use Nhost Cloud or self-host it with Docker.

At a glance

Data modelling for the whole team

Nhost gives you a real Postgres database: you design tables in the dashboard or in SQL, track them in Hasura and keep migrations in your repository with the CLI. That is a clean setup for developers, and the data model stays in their hands.

In pylo you add an entity, a field or a relation in the admin panel and the typed GraphQL API updates at once, without any kind of migration or redeployment. Field types include relations, enums, JSON, public or private files, rich text, dates, booleans and number/text formatters, with validation built in. You filter across relations and group aggregations by any field, even across deeply nested data structures. The people who understand the data can change the model, not only the developers.

Permissions that follow the data everywhere

Nhost uses Hasura's permission model: per role and table you allow select, insert, update and delete, with row filters and column lists, and only the admin role has access by default. The dashboard has a visual editor for these rules. They protect the GraphQL API. Your functions, containers and any admin tool you build next to it need their own checks.

pylo separates tenants at the row level. Inside a tenant, any entity can act as the requester (ARO) and any entity as the requested object (ACO), with hierarchies and field-level rules. Every path into the data is GraphQL, so the same rules apply to the API, the admin panel, forms, flows and MCP.

pylo flows instead of hand-written functions

On Nhost, logic lives in functions that you write in JavaScript or TypeScript and deploy with a git push, or in custom containers on Nhost Run. Hasura event triggers call them when data changes and cron triggers call them on a schedule. Retries, rate limits, logging and connections to other systems are code you write.

In pylo you build the same logic in a visual workflow engine called pylo flows, where any event in pylo can start a flow, additionally you can use cron schedules and incoming webhooks. pylo flows include retries, custom rate limiting, error handling and full event logs and everything is statically typed. pylo ships integrations for email sending and a growing selection of third-party systems, but you can easily code your own flow actions in TypeScript.

An admin dashboard your operations team will actually use

The Nhost dashboard is a developer console with a table editor, logs and settings. There are no list views per role, no state-dependent fields and no screens made for the people who run the business, so teams build a back office on top.

pylo's admin UI is built for your whole team, no matter if they are developers or not: you compose list and detail views from predefined components, decide who sees what and show or hide fields depending on a record's state. The pylo form builder maps forms directly to your entities, including relations, with public or authenticated access, hidden fields, conditional rendering and submission limits per form or per user.

Events and dashboards built in with pylo analytics

Nhost shows metrics and logs for your project. Reporting on your business data means building it yourself or moving the data into an analytics tool.

In pylo, every create, update and delete emits an event with the full before and after state, which you can enrich with your own event data through our API. On top of your events and regular pylo data you can build customized dashboards directly next to your data and workflows.

AI and MCP with guardrails

Nhost has an MCP server in its CLI that can run GraphQL queries and mutations, create migrations and manage permissions, with allowlists for which operations are allowed. Since 2026 there is also an MCP service that logs in through Nhost Auth and applies the user's permissions. In both, allowed writes apply directly.

pylo's MCP server is made for both, developers who want to use agentic coding and team members who want to analyze their data. The AI gets exactly the connected user's read permissions, nothing else. pylo never applies an AI's writes directly, instead they arrive as proposals that a human accepts or declines, so you can benefit from the speed of AI without kicking the human out of the loop.

File handling

Nhost offers more media functionality out of the box: its storage comes with a CDN, on-the-fly image transformations and virus scanning. pylo stores files in EU-hosted object storage with public or private links, no file size limit and for any file type. You can build any kind of file-based functionality in your custom pylo flows, of course.

Agencies and multi-tenant work

With pylo, agencies can run many client workspaces from one user account; each workspace is its own tenant separated by RLS and one login switches between them. On Nhost, each client usually gets its own project, from $25 per month on Pro. Inside each client's app, you build tenant separation, roles and the back office yourself.

Pricing

For a pure app backend, Nhost Pro can be cheaper to start with, and self-hosting the open-source stack costs only your servers and the time to run them. Compute beyond the included credit, extra storage and egress are billed on top.

pylo's price already includes pylo flows, the admin UI, forms, analytics and personal onboarding, which on Nhost you would build or buy separately. Every pylo plan has every feature and unlimited users, plans differ only in included usage.

Hosting, compliance and reliability

All pylo content data, data models and databases live in data centres in Germany, our subprocessor list is public and names every non-EU parent company openly. A GDPR data processing agreement is part of the terms and we never use customer data to train AI models. Paid plans come with a 99.5% availability SLA, hourly encrypted backups kept for 30 days at a separate German location and support from the team that builds the product.

Nhost AB is a Swedish company, which is a plus for European customers, and its cloud runs on AWS with regions including Frankfurt and London. SOC 2 Type II is included from the Team plan, and uptime SLAs are part of Enterprise. If you self-host, all of that is up to you.

Where Nhost is the better choice

We would rather tell you than have you find out later:

• Nhost is open source under MIT and self-hostable

• Nhost offers a greater variety of login providers

• Real Postgres with SQL access, migrations in your repository and a CLI that runs the whole stack locally

• Custom containers in any language with Nhost Run

If you build web or mobile apps on Postgres and GraphQL and want open source with strong auth, Nhost is an excellent choice. pylo is for teams that would rather not build automation and a back office on top of their backend.

Where pylo is a great Nhost alternative

We think these are the main reasons to pick pylo over Nhost:

• The business layer is already built. API, permissions, flows, admin UI, forms, events and MCP come with your data model. You don't write them in functions and frontends yourself.

• Schema changes without a deploy. Add an entity or field and the typed GraphQL API is live immediately. You skip the migration, the metadata change and the redeploy.

• Permissions that hold everywhere. API, admin panel, forms, flows and MCP all go through the same permission layer, with hierarchies and field-level rules.

• Business logic without deployments. The flow builder covers event, cron and webhook triggers, with retries, rate limiting, typed actions and full logs. Custom actions are plain TypeScript.

• Non-developers can work without you. Configurable admin views, state-dependent fields and a full form builder let operations teams handle their own changes. Not every change has to become a developer ticket.

• Analytics you don't have to instrument. pylo records every change as an event with before and after state, and dashboards sit right next to the data.

• AI access that's safe for production. MCP runs with the connected user's exact read permissions, and every write goes through human approval as a proposal.

• Predictable pricing. pylo doesn't bill compute, egress or database size. You get unlimited users and API requests and all features on every plan. Only three meters are billed: flow actions, records, and file storage/traffic.

• Built for agencies. You can run many client workspaces with separate tenants and permissions, and switch between them with one login.

• German hosting from a German company. pylo hosts user data in Germany and the DPA is part of the terms. You get a 99.5% SLA from the first paid plan, hourly backups with 30-day retention, and personal support from the people who build the product.

FAQs - Questions and answers about pylo

• Does pylo give me direct SQL access like Nhost? No, you work with your data through the typed GraphQL API, the admin panel, flows and MCP, all with the same permissions. This is an active choice, so we handle security and performance and you can focus on your product.

• Is pylo open source? No, but you can export all data at any time as CSV or JSON or through the API. The data models, flow definitions and generated SDK code are yours to keep, even after you leave.

• Can I self-host pylo? No, but if you need specific hosting options, we can discuss them as part of the Enterprise plan.

• Does pylo scale? Yes, production workspaces of our customers already run with millions of records today.

• Where exactly is my data? Content data is in data centres in Germany and backups are kept at a different German location.

• Can I get access now? pylo is currently in private beta. Request access and we will set up your first entities with you on a call and onboard you personally for your first project.