M Mermaduckle Self-hosted core, managed option
Read docs Open hosted eval
Self-hosted first. Hosted by us when needed.

Deploy governed AI workflows in your stack, or use our hosted control plane to move faster.

Mermaduckle is software you can run in your own environment. The /app route on this site is our hosted evaluation instance so teams can inspect the control plane before deploying it themselves. Commercially, we make money on managed pilots, hosted operations, and annual self-hosted support.

Open hosted eval See deployment model
Self-hosted customer deployments Hosted evaluation at /app Managed rollout available One Rust service from local dev to Fly
Operator console

Build, review, and ship from one surface.

workflow import draft architect run + debug
Webhook Event enters the system
Agent Model and tools do the work
Approval Human checkpoint on risky actions
Action External side effects only after review
Current strength Compose fast without losing the operator view.

The shipped app already supports canvas editing, workflow import/export, an AI architect endpoint, debugging, and scheduled execution.

Rust + Actix-web
Vanilla JS control plane
Hosted evaluation on this site
Self-hosted deployment artifacts in repo
Approvals, audit, reporting, and recovery
Webhooks, schedules, agents, and secrets
Deployment model

The product is self-hosted. The instance on this site is just one operating mode.

This should be obvious when you land here: Mermaduckle is built to run in the customer's environment, but we also host a live evaluation instance on this domain and offer a managed commercial path for teams that want help getting to production.

Self-hosted core

Run it in your environment

The repo, docs, Dockerfile, and Fly configuration are there so teams can deploy their own instance and keep operational ownership.

Hosted evaluation

Inspect the product before you deploy

The /app route on this site is our hosted control plane so buyers can see the workflow, approval, and audit surfaces immediately.

Managed option

Let us run the first rollout with you

If a team wants outcomes before they want infrastructure ownership, we can host, integrate, and support the production rollout as a paid engagement.

Positioning

Other products help you build flows. This one helps you operate them.

Public positioning from Langflow and Flowise centers on visual AI building. n8n and Pipedream emphasize broad automation and integrations. Mermaduckle is strongest in the space between those two camps: the governed operating layer for AI workflows in a system your team can actually host, inspect, and ship.

Builder-first

Great at composing graphs

Tools like Langflow and Flowise are optimized for designing chains, prompts, and agent logic quickly.

  • Strong visual assembly of AI flows
  • Fast experimentation with prompts and models
  • Less emphasis on review queues and operator controls
Automation-first

Great at moving data between systems

Tools like n8n and Pipedream are optimized for connector breadth and event-driven automations across many apps.

  • Big integration surface area
  • Fast automation for SaaS-heavy stacks
  • AI governance is usually not the product center
Mermaduckle

Built for the operator moment

When a workflow needs approval, auditability, secrets, recovery, and a production deploy path, Mermaduckle keeps those concerns in the same control plane.

  • Approval checkpoints and audit events built into the app surface
  • Agent, workflow, and run visibility in one place
  • Self-hosted Rust service with a direct Fly deployment path
Commercial model

We make money around production ownership, not cheap builder seats.

The goal is not to race n8n, Flowise, or Langflow to the lowest self-serve plan. The money is in getting important workflows live, keeping them governed, and backing teams that either want a hosted operator or need support for self-hosted rollout.

Managed pilot

The first paid engagement

Use Mermaduckle to launch the first production workflow with hands-on delivery.

  • We implement the first real workflow with your team
  • We can host or co-run the control plane during rollout
  • Commercial shape: setup fee plus ongoing support
Hosted operations

The recurring revenue layer

Run the control plane for customers that want outcomes without owning infra from day one.

  • We operate releases, monitoring, and production support
  • Priced as a platform fee plus usage and support expectations
  • The public /app instance is the clearest proof this mode exists
Enterprise self-hosted

The annual contract path

Sell commercial terms to teams that want control over data, deployment, and operations.

  • Customer runs the service in their environment
  • We sell onboarding, support, upgrades, and commercial license terms
  • Best for buyers who care about control more than a generic SaaS seat model
Governance in the product

Human review, traceability, and production safety are not bolted on later.

The app already exposes a pending approvals queue, audit events, reporting endpoints, retry and recovery flows, and the settings surface needed to manage keys, secrets, and integrations around the workflow itself.

01

Approval checkpoints

Pause at pending_approval, hold risk, and resume only with an explicit action.

02

Audit depth

Track workflow, approval, and agent activity with event types, actors, and timing.

03

Recovery paths

Use logs and context to patch a run and retry without discarding the operating record.

1
Webhook or schedule starts the run

Enter from external events or timed execution.

2
Agent logic executes under policy

Use the workflow engine while secrets and integrations stay centralized.

3
Approval and audit close the loop

Keep the decision record, the run data, and the report endpoint available.

What the app actually does

The current build already covers the hard middle between idea and production.

This is not speculative roadmap copy. These are the product surfaces already present in the app and server: workflow building, agent operations, governance controls, runtime settings, reporting, and deployment infrastructure.

01

Workflow studio

Create, edit, import, export, debug, and run flows on a visual canvas.

02

AI architect

Turn plain-language workflow requirements into a generated starting graph.

03

Agent management

Configure agents, inspect recent runs, and chat directly inside the control plane.

04

Runtime settings

Manage API keys, environment secrets, integrations, notifications, and webhook endpoints.

05

Scheduling and webhooks

Run flows on cadence or trigger them from external systems in production.

06

Reporting and visibility

Expose dashboard metrics, audit timelines, workflow reports, and system status in one app.

Production path

Run your own instance when customer-managed deployment is the point.

The repo already ships with a Fly Dockerfile, a working fly.toml, GitHub Actions deployment workflow, and a PowerShell helper script. The same Rust service can power the hosted eval here or a customer-owned deployment elsewhere.

Ship to Fly Read the full deploy guide 
flyctl deploy -a mermaduckle --remote-only --config fly.toml

Push to main for GitHub-driven deploys, or run the included helper script from your machine when you want to cut a release directly.

Included in this repo

Everything needed to run the service in production

  • deploy/Dockerfile.fly for multi-stage Rust builds
  • fly.toml with HTTP/TLS ports and health checks
  • .github/workflows/fly-deploy.yml for push-to-main deployment
  • tools/deploy_fly.ps1 for local production releases
  • /api/health for verification after deploy
Pick the path

Open the hosted eval, or read the self-hosted runbook and deploy your own copy.

Open hosted eval Open self-hosted docs