Castingwords

@castingwords what's my current prepay balance and what SKUs are available for a 45-minute podcast episode?

Your Castingwords prepay balance is $127.50. For a 45-minute audio file, available SKUs include: Standard transcription ($1.25/min, $56.25 total, 4-day turnaround), Premium transcription ($2.00/min, $90.00 total, 2-day turnaround), and Verbatim transcription ($2.50/min, $112.50 total, 3-day turnaround). All SKUs include speaker identification. You have sufficient balance for any of these options.

This example shows the MCP's ability to check account status and pricing in a single query. Balance and SKU pricing are account-specific and change over time. The MCP requires API key authentication to access your Castingwords account data.

@castingwords order a standard transcription for https://example.com/interview-2024-12.mp3 and register https://myapp.com/webhooks/transcripts as the callback URL

Transcription order placed successfully. Order ID: CW-89234, Audio file ID: 45678. Estimated completion: December 18, 2024. Cost: $43.75 deducted from prepay balance (new balance: $83.75). Webhook registered at https://myapp.com/webhooks/transcripts — you'll receive a POST notification when transcription is complete. Run a test webhook call to verify your endpoint is ready to receive events.

This example demonstrates the MCP's core workflow: initiating transcription and setting up completion notifications. The order action deducts funds immediately from your prepay balance. If your webhook endpoint isn't configured to accept Castingwords' payload format, you won't receive completion alerts.

@castingwords I just received a webhook for file 45678 — retrieve the transcript and summarize the key discussion points in bullet format

The Castingwords MCP doesn't include a tool to retrieve completed transcript text directly. You'll need to fetch the transcript from the URL provided in the webhook payload (typically a .txt or .json file). Once you have the transcript content, I can summarize it. Alternatively, if you provide the transcript text here, I can extract key discussion points immediately.

This example reveals a scope limitation: the MCP handles order management and webhooks but doesn't fetch transcript content itself. This is intentional — Castingwords delivers transcripts via URLs in webhook payloads, which you retrieve separately. The AI can still process transcript text once you provide it.

When Castingwords fits a recurring podcast workflow

A two-person podcast team ships a 45-minute episode every Tuesday and needs transcripts live by Thursday for SEO and accessibility. Castingwords works here because the Order Transcript tool takes a direct media URL, the webhook registration means transcripts arrive without manual checking, and the prepay balance check prevents order failures mid-workflow. The API key auth is simple enough for a Zapier handoff or a lightweight script. The threshold: if you're doing more than one episode per week or need same-day turnaround, you'll hit the vendor's processing queue and should confirm SLA before committing. For a weekly cadence with 48-hour lead time, this MCP removes the upload-download-paste loop and keeps transcripts flowing into your CMS automatically.

Why research teams with 20+ interviews per month should look elsewhere

A product team runs 15-20 customer interviews monthly and wants transcripts in Notion for tagging and synthesis. Castingwords can handle the volume, but the MCP's toolset reveals a mismatch: there's no batch order endpoint, no transcript retrieval tool in the representative set, and no status polling beyond webhooks. You'd need to build a separate retrieval layer or rely entirely on webhook delivery to a staging endpoint. If your interviews are ad-hoc and you're okay with per-file ordering and manual download, it works. If you're running a structured research ops process with 50+ hours of audio per quarter, a transcription service with native batch APIs and in-app retrieval (like Otter or Descript) will save you integration overhead. Use Castingwords here only if you already have webhook infrastructure and want vendor-neutral transcription.

When async standup recordings need lightweight transcription

A five-person remote team records async standups in Loom three times a week and wants text summaries in Slack by end-of-day. Castingwords fits because the workflow is low-frequency, the Order Transcript tool accepts a URL (Loom's share link works), and the webhook can POST directly to a Slack incoming webhook or a thin Lambda that formats the transcript. The prepay balance check prevents silent failures when the account runs dry. The trade-off: Castingwords doesn't return structured speaker labels or timestamps by default—you'd need the Order Upgrades tool for timestamps, which adds cost and a second API call. If your standups are under 10 minutes and you just need a text block for searchability, this MCP is a set-it-and-forget-it win. If you need speaker diarization or real-time captions, look at a speech-to-text MCP instead.