Since we started building GloriaMundo, one question has come up more than almost any other: “Can I call my workflow from my own app?”

The answer today is: not yet. But it’s coming — and we’re designing it properly rather than rushing out a webhook and calling it done. This post lays out what we’re planning, how it’ll work, and why we think it changes what GloriaMundo is.

The idea in one sentence

Any workflow you build in GloriaMundo will be publishable as a secure, authenticated API endpoint that your own applications can call — with custom input parameters, real-time status tracking, and full cost transparency.

Why this matters

Right now, GloriaMundo is a tool you use. You open the platform, describe what you want, build a workflow, run it. It’s powerful, but it’s interactive — you’re the one pressing buttons.

Workflow-as-API turns GloriaMundo into infrastructure you build on. The difference is significant. Imagine you’ve built a workflow that takes a company name, researches it across multiple sources, and produces a formatted investment brief. Today, you can schedule that on a timer or trigger it via webhook — but you can’t embed it as a living component inside your own application. With Workflow-as-API, your CRM could call that workflow automatically every time a new lead enters the pipeline. Your internal tool could offer a “Generate Brief” button that fires a request to GloriaMundo and displays the result inline. A scheduled script could process a list of companies overnight and have the briefs waiting in your inbox by morning.

You built the workflow once, in plain English, with our Glass Box preview to verify it does what you expect. Now it runs wherever you need it, as many times as you need it, without you being there.

How it’ll work

Building and parameterising

The workflow building experience stays exactly the same. You describe what you want, GloriaMundo generates the workflow, you preview it with Virtual Run, you refine it until it’s right.

The new step is parameterisation. When you’re happy with a workflow, you identify which values should be provided by the caller rather than hardcoded. If your workflow sends an email, maybe the recipient and subject should be API inputs. If it researches a company, the company name should be an input. You’ll be able to click on any step parameter and mark it as an API input — give it a name, a type, and a description.

Behind the scenes, this replaces the hardcoded value with a template variable. When an API call comes in with {"company_name": "Anthropic"}, that value flows through to every step that references it.

Publishing

Once your workflow is parameterised, you publish it as an API endpoint. Publishing does two important things. First, it freezes the workflow definition — so edits you make later don’t silently change the behaviour of a live integration. You’ll need to explicitly re-publish for changes to take effect. Second, it generates your API endpoint and an input schema that documents exactly what callers need to provide.

This is the Glass Box philosophy applied to developer workflows. You see the frozen definition, you see the input schema, you see what will happen when someone calls it. No surprises.

Calling the API

The API is deliberately simple. Here’s what a call will look like:

POST https://api.gloriamundo.com/v1/workflows/{workflow_id}/execute
Authorization: Bearer <YOUR_GM_API_KEY>

{
  "input": {
    "company_name": "Anthropic",
    "report_depth": "detailed",
    "recipient_email": "analyst@fund.com"
  }
}

The response comes back immediately with a run ID and status:

{
  "run_id": "run_8f3a...",
  "status": "running",
  "urls": {
    "get": "https://api.gloriamundo.com/v1/runs/run_8f3a...",
    "cancel": "https://api.gloriamundo.com/v1/runs/run_8f3a.../cancel"
  }
}

AI workflows take time — they’re calling LLMs, fetching data from APIs, generating content. Rather than pretending this is instant, we give you three ways to get results.

Polling: Hit the status URL periodically. You’ll see progress (step 3 of 7, generating content…) and the full result when it’s done.

Webhook callback: Pass a callback_url in your request, and we’ll POST the completed result to your server when the workflow finishes. Signed with HMAC-SHA256 so you can verify it’s genuinely from us.

Synchronous wait: Send a Prefer: wait header, and we’ll hold the connection open. If the workflow completes within the timeout, you get the result inline as a normal response. If it takes longer, you get a 202 with the polling URL. This is ideal for shorter workflows where you want the simplicity of a synchronous call.

API keys

You’ll manage API keys from your GloriaMundo dashboard. Keys use the format gm_live_ (or gm_test_ for development), following the same conventions as Stripe and other platforms developers already know. When you create a key, it’s shown once — after that, it’s masked. You can name keys, set rate limits, configure spend caps, and revoke them instantly.

We’re planning per-key rate limiting and spend caps so you stay in control. If you set a cap of $50/day on a key, we’ll reject requests once that limit is hit rather than letting costs spiral. You’ll see usage breakdowns per key in the dashboard.

Billing

API calls run through the same transparent billing system as everything else in GloriaMundo. Each call uses credits based on the actual resources consumed — the LLM calls, the API actions, the compute time. The 2× markup on underlying costs applies identically. No hidden per-call fees, no separate API pricing tier that makes simple calls expensive.

Every run tracks its costs, and they appear in your usage dashboard just like interactive runs do. The Glass Box applies to your bill too.

What we learned from studying the competition

We spent considerable time analysing how existing platforms handle this problem — Zapier, Make.com, n8n, Pipedream, and others — plus best practices from Stripe, Replicate, and Twilio for API design.

Most workflow platforms don’t actually solve this problem well. Zapier’s webhook triggers are fire-and-forget with no built-in authentication — security relies on nobody guessing your URL. Make.com gets closest with authenticated webhooks and synchronous response capabilities, but has a 180-second timeout and no schema validation. n8n offers flexible auth options but no input validation. None of them offer workflow versioning for API stability.

We think we can do meaningfully better in three areas.

Input validation. When you publish a workflow as an API, the input schema isn’t just documentation — it’s enforced. Send a request missing a required field, or with the wrong type, and you get a clear 400 error explaining exactly what’s wrong. This catches bugs at the API boundary rather than letting bad data flow through to a confusing mid-workflow failure.

Versioning. Publishing freezes your workflow. Edit it freely in the UI — your live API doesn’t change until you explicitly re-publish. Every other platform we studied changes the live endpoint the moment you edit the workflow. For developers building production integrations, that’s unacceptable.

Transparency. Every API run produces the same detailed execution log as an interactive run. You can see what each step did, what data flowed between steps, what it cost, and how long it took. If something fails, you see exactly where and why — not just a 500 error.

The Developer Tier

Workflow-as-API will be the centrepiece of a new Developer Tier, priced at $79/month with $50 in included credits.

This isn’t about restricting a feature to extract more money. It’s about recognising that developers using GloriaMundo as infrastructure have different needs and create different value than users interacting through the UI. The developer tier reflects that with higher credit allocations, higher concurrency limits, webhook callbacks, API key management, usage analytics, and priority execution.

If you’re using GloriaMundo to automate your own tasks, the Pro tier at $20/month remains the right choice and isn’t losing any features. If you’re building products and integrations on top of GloriaMundo workflows, the developer tier gives you the tools and the headroom to do that reliably.

Here’s what the developer tier includes beyond the current Pro tier:

API access with key management, per-key rate limits, and spend caps
Workflow publishing with versioning and input schema enforcement
Webhook callbacks for asynchronous result delivery
$50/month in included credits (vs $20 on Pro)
Higher concurrency — 5 simultaneous workflow executions instead of 2
Priority execution in the workflow queue
Usage analytics dashboard — call volume, error rates, latency, and cost breakdowns per API key and per workflow

When

We’re focused right now on the core platform launch — making sure workflow generation is rock-solid, the integration ecosystem is reliable, and the Glass Box experience delivers on its promise. Workflow-as-API is designed and specified, but we’re building it second, not first.

We expect to begin rolling out the developer tier within two months of launch, starting with a small group of beta partners who’ll help us refine the API design and developer experience before general availability. If you’re interested in early access, drop us a note — we’d love to co-design this with people who’ll actually use it.

What this means for GloriaMundo

We started GloriaMundo because we believed that AI automation should be transparent. The Glass Box philosophy — see what your agents will do before they do it — is the foundation of everything we build. Workflow-as-API extends that philosophy into a new context: your agents still operate transparently, with full audit trails and cost visibility, even when they’re called programmatically from code you wrote.

Other platforms treat API access as an afterthought — a webhook URL and a prayer. We think developers deserve the same transparency and control that our interactive users get. That’s what we’re building.

GloriaMundo is a workflow automation platform that converts natural language into executable, transparent AI workflows. Learn more at gloriamundo.com.