Home/Product/Infrastructure

Everything between the user's browser and your bucket.

Ingestion in two regions with two zones each, recovery when connections drop, a transcoding pipeline that turns whatever the browser produced into an mp4 that plays everywhere, and delivery to the storage you choose. Here is how it fits together.

2 × 2Regions (US, EU), two zones each
200k+Recordings a day per region, in production
1 TBStorage and bandwidth included
100%Of inputs delivered as one mp4 profile

Built for uploads from bad connections, and for updates without downtime.

Recording produces large files from phones on mobile data. Each part of the ingestion path assumes something will go wrong.

US and EU regions

Ingestion and processing run in both. End users are routed to the closest region by default, or you can pin an environment to EU or US for latency and data-protection reasons.

Two zones per region

us1 and us2, eu2 and eu3. Each can carry the whole region's recording and processing load on its own, so we take one offline for updates while the other serves.

Automatic zone fallback

If the recording client's first secure connection attempt fails, it retries against the other zone in the same region before the user notices.

Recording recovery

Media is streamed as it's recorded, not uploaded at the end. If the connection dies mid-recording, we take what was received and try to process it anyway.

Processing queue

During rare spikes, recordings queue instead of failing, and every one is processed in order. Concurrency per region is a plan setting; see pricing.

EU silo for EU companies

With the EU silo, users' personal data and recordings stay inside the EU and never touch a non-EU company: EU ingestion, EU processing, Scaleway storage, and pipe.js served from our own EU delivery servers instead of CloudFront.

Client delivery by CDN

Load pipe.js from our Amazon CloudFront distribution for worldwide edge delivery, or directly from the EU servers when the silo matters more than the last few milliseconds.

Status you can watch

Every region is on the public status page, and the changelog lists what shipped and when.

Everything in, one mp4 out.

Browsers and phones produce a dozen container and codec combinations, most of them unfit for web playback. The pipeline normalizes all of them to a single, tuned mp4 profile.

.webm VP8.webm VP9.mp4 H.264.mov HEVC.3gp.flvOpusfragmented mp4rotated mobile
.mp4H.264 + AAC, non-fragmented, plays in every browser, player and device
ftyp
moov
mdat
moov atom first. Playback starts before the file finishes downloading.
Keyframe at 0:00. No black frame while the decoder waits for the first I-frame.

Mobile rotation handled

Device orientation metadata is read and applied, so portrait phone recordings and their snapshots come out the right way up.

Snapshot and filmstrip

A thumbnail for lists and embeds, plus a 25-frame filmstrip for scrubbing previews, extracted from every video.

Watermarks

On the PRO plan, place your watermark in any corner of every video, per environment.

Audio Transcription

Automatically detect the language of the audio and transcribe it.

Resize, scale, crop

The transcoding engine can change aspect ratio and resolution on the way through, and keep the original file next to the output if you need it.

Audio-only stays light

Audio recordings are processed to a compact audio file, and audio can be extracted from video as a separate output.

Stored with us, or delivered to you. Your call per environment.

1 TB of storage and bandwidth is included. When you want the files on your own infrastructure, AddPipe pushes them there automatically after processing.

Pipe storage Default

Long-term complimentary storage on Amazon S3 (US, behind CloudFront) and Scaleway (EU), encrypted at rest, with playback and download URLs ready to use.

Amazon S3

Any bucket, any region. Supports multipart upload for large files, canned ACLs and storage classes.

S3-compatible services

Custom endpoints for Google Cloud Storage, DigitalOcean Spaces, Cloudflare R2 and anything else that speaks the S3 API.

FTP, FTPS, SFTP

Push the mp4 and its snapshot and filmstrip to your own server over the protocol your infrastructure already accepts.

Dropbox and Box

New recordings appear on every device and trigger notifications, with no manual downloading.

YouTube

Publish individual recordings to a channel straight from the dashboard.

Do not store

Turn it on and recordings are pushed only to your storage, never to ours. Combined with the EU silo and the PII controls, AddPipe becomes a pass-through processor rather than a data holder.

When something goes wrong, the answer is in the logs.

The dashboard and REST API expose the same data: every recording, its metadata, and what happened to it at each step.

Every recording, immediately

Search and open recordings and their metadata in the dashboard the moment processing completes, or list them as JSON over the REST API.

Rich metadata

Resolution, duration, IP, user agent, codecs, camera and microphone names, and any payload you attached, stored with each file.

Three kinds of logs

Webhook events with responses, storage push attempts, and getUserMedia calls with their errors, so a failed upload or a blocked camera is diagnosable.

Environments

Separate settings, storage targets, region and recordings per website, project or client, with their own webhooks.

Your time zone

Dates and times display in your local time everywhere in the dashboard.

Export

Recording data for any environment exports to .csv, and monthly usage can be exported for billing reconciliation.

Put a recording through it.

Record from a phone on the trial, watch it land in your bucket as an mp4, read the logs. Free for 14 days, no card required.