Why We Built QuickCast — From Webhook Testing Headache to One-Tap Voice Automation

Every automation builder has a dirty secret: testing webhooks is tedious. You open a tool like Postman or webhook.site, craft a JSON payload, remember the right headers, hit Send, check the logs, tweak something, do it again. And again. If the payload includes an audio file — say, for a voice-to-text pipeline or an AI agent trigger — the friction doubles because now you’re recording something, exporting it, uploading it manually, then running your test.

We hit that wall repeatedly while building n8n workflows for clients. The workflows themselves were elegant. The testing loop was not. And the moment we wanted to test end-to-end — real audio, real transcription, real downstream actions — we had to leave our flow editor, juggle three apps, and hope we remembered which endpoint we were targeting.

That frustration became QuickCast.

The original idea: the shortest path from voice to webhook

The first commit was brutally simple: record audio on an iPhone, POST it to a URL. That’s it. No accounts, no cloud, no backend. Your voice goes from the microphone to your webhook in a single tap. The file lands as multipart form-data with the right Content-Type, ready for n8n, Make, Zapier, or any custom API to pick up.

We used it ourselves for a week and immediately found the next set of problems.

What v1.0 taught us

The first version worked — technically. But using it daily surfaced the real friction:

“Which webhook am I sending to?” We kept forgetting which endpoint was selected. So we added persistent webhook selection with status indicators — green dot if the last upload succeeded, yellow if there were issues, gray if untested.

“The upload failed and I have no idea why.” HTTP errors, timeouts, wrong Content-Type — all silent. We built a proper error handling layer with retry logic, detailed error messages, and a recording history that shows exactly what happened to each upload.

“I’m on the train and there’s no signal.” Recordings disappeared into the void. So we built an offline queue: record now, the app detects when connectivity returns and uploads automatically. No lost recordings.

“I want to test with real transcription.” Many automation pipelines care about the text in the voice note, not just the audio file. We integrated Apple’s Speech framework for on-device, real-time transcription. The transcript ships alongside the audio in the same multipart upload — no external transcription service, no API key, no round-trip latency.

Each problem we solved for ourselves turned out to be a problem other automation builders had too.

From tool to product: the 1.0 → 1.5 journey

After shipping 1.0 to the App Store, we kept using QuickCast as our daily driver for client work. The roadmap wrote itself:

Audio format flexibility

Voice notes are fine in AAC, but some pipelines expect WAV, others want Opus for compression. We added four formats (AAC, WAV, Opus, FLAC) with configurable sample rate, channels, and quality presets. Now you match the format to the pipeline, not the other way around.

Design system and dark theme

The first UI was functional but rough. We rebuilt it around a proper design token system — a dark, professional theme with glassmorphic cards and a glossy record button that feels good to press 50 times a day.

Widgets and Siri shortcuts

“Open the app, find the record button, tap it” is already three steps too many. We added Siri Shortcuts (“Hey Siri, record a QuickCast”), home screen widgets, and lock screen widgets so the recording is always one gesture away.

Action Button (1.5)

The iPhone 15 Pro’s Action Button was the perfect physical anchor. Press it, record, press it again, the audio ships to your webhook. No screen, no unlock, no app. This is the fastest possible path from thought to webhook payload.

Multi-recipient fan-out (1.5)

One voice note, three destinations. Send the same recording to Slack (for the team), n8n (for the pipeline), and a personal backup endpoint — in parallel, with per-webhook delivery tracking and individual retry buttons for any that fail. Webhook Groups let you save these multi-target sets so it’s one tap next time.

Searchable tagged history (1.5)

After a few hundred recordings, “scroll and squint” doesn’t scale. Full-text search across transcripts, webhook names, file names, and error messages. Custom tags, favorites, sort by date/duration/size/status, and an advanced filter sheet. Auto-tags classify every recording by webhook, duration bucket, and whether it has a transcript.

Onboarding and analytics (1.5)

We wanted to understand which features people actually use, so we added opt-in anonymous analytics via Aptabase — a privacy-focused, EU-hosted service. No audio, no transcripts, no webhook data is ever transmitted. The first-launch onboarding flow asks explicitly, and you can toggle it off any time in Settings.

What we learned building it

The webhook is the universal API. Every low-code platform, every automation tool, every AI agent framework can receive a webhook. By targeting that single abstraction, QuickCast works with everything without integrating with anything specific.

On-device processing is underrated. Speech-to-text on-device means no API keys, no latency, no privacy concerns. It also means transcription works offline and in airplane mode. Apple’s Speech framework isn’t perfect, but for voice notes and quick captures it’s surprisingly good.

Physical gestures beat screen taps. The Action Button changed how we use QuickCast. There’s something powerful about a hardware button that triggers a full automation pipeline. You press it walking down the street, talk for 10 seconds, and by the time you put the phone back in your pocket the audio + transcript is already in your n8n workflow.

Privacy is a feature, not a constraint. No accounts, no cloud, no tracking by default. Your recordings go where you tell them and nowhere else. This isn’t just an ethical position — it’s a competitive advantage. People building sensitive automations (client calls, medical notes, legal dictation) need to trust their tools.

Where QuickCast is going

We’re not done. The roadmap for the rest of 2026 includes:

• Apple Watch app — record from your wrist, sync to your phone, upload to your webhook
• On-device AI with prompt templates — per-webhook prompt templates that summarize, extract action items, or restructure your voice note before it ships, all processed on-device using Apple’s Foundation Models framework
• Live Activity and Dynamic Island — see your recording timer on the lock screen with a one-tap stop button (this one is 90% done but we need to fix a type-sharing issue between the app and widget targets before it ships)

Try it

QuickCast is a one-time purchase on the App Store. No subscription, no ads, no account. If you build automations and want the fastest way to fire a real voice payload at your webhooks, give it a shot:

Download QuickCast on the App Store

We built it because we needed it. Turns out a lot of other people did too.