Marketing videos in minutes, without a production team

What we had to solve

• 01

  Video rendering that doesn't make people wait

  Web-based video rendering is slow by default. Video processing is compute-intensive, and a traditional server means jobs queue up behind each other. We needed renders to complete in under 5 minutes regardless of simultaneous load. The architecture: render jobs enter an AWS SQS queue, which triggers an isolated Lambda function per job. That Lambda processes the video with FFmpeg, writes the output file to S3, and updates job status in Hasura. The frontend holds a live subscription on that Hasura table, so the progress indicator updates the instant a render completes — no polling interval, no manual refresh. The Lambda itself is split into three layers (a static library layer, an npm-modules layer, and the active renderer.js logic) so only the code that actually changes needs redeploying.

• 02

  A template library that feels useful, not overwhelming

  A library of 400+ templates is only useful if business owners can find what they need quickly. Too many options with poor categorization creates decision paralysis, the same friction the platform was built to remove. We structured the template library around use cases (promotional, explainer, seasonal, social, B2B) with preview thumbnails that made the right template obvious before clicking in.

• 03

  A live editor that responds before users expect it to

  The editor had to feel instant for someone who has never touched video software. We built real-time canvas-based editing on Next.js — users drag in images (auto-fitted to aspect ratio), swap fonts and brand colors, add audio, and see every change rendered immediately in the browser before committing to a full render. No round-trips to the server for each edit, no processing spinner on every text change. The live preview is the safety net that lets non-technical users experiment without fear.

• 04

  Team access that scales with an agency's client model

  A single-user account model wouldn't work for an agency managing dozens of clients under one platform. We built a multi-seat, org-level permission system where team-invite ability, monthly video limits, and template library access are each tied to the org's plan tier. Only the org admin can manage team members and cloud-storage credentials. That access hierarchy had to be enforced at the data layer — not just at the UI level — so no client could accidentally see another's assets or consume another's render quota.

• 05

  Email and workflow automation without custom maintenance

  Rather than building a custom notification system, we used Postmark for transactional email — render-complete and team-member-added alerts — with branded templates built and managed inside Postmark itself. Account creation and payment-gateway events flow through Zapier, so the agency can adjust onboarding workflows without touching the codebase. This kept the core application lean and moved operational logic into tools the client's team could maintain themselves.

1. 01Weeks 1–2

   Discovery and scoping

   We map the problem before writing code. Two weeks of technical audit, stakeholder interviews, and prototype — so both teams align on scope and risk before sprint one.
2. 02Ongoing

   Two-week Agile sprints

   Each sprint ends with working software, not a status update. You review a real build, request changes, and approve before we move forward. No surprises at handover.
3. 03Ongoing

   Daily async updates

   Slack for daily progress, Asana for task visibility, weekly video calls for decisions. You have full visibility without needing to attend every meeting.
4. 04Final

   Handover and warranty

   Full code handover with deployment runbooks and documentation. Thirty-day warranty period for production issues at no extra cost.

• 01

  AWS Lambda
  Video rendering is compute-intensive and unpredictable in volume. Lambda runs each render job in isolation so 100 simultaneous renders all complete in under 5 minutes, with no idle servers sitting between campaigns.
• 02

  NestJS
  The backend needed to handle concurrent render jobs, template management, and user accounts reliably. NestJS gave us a structured framework that could grow without becoming difficult to maintain.
• 03

  Hasura
  Template metadata, render job tracking, and user video history all needed fast, reliable API access. Hasura auto-generates GraphQL APIs from the database schema, which cut the time to build and iterate on data-heavy features.
• 04

  Firebase
  User authentication for a white-label SaaS needed to work reliably without becoming a maintenance burden. Firebase handles auth at scale with no infrastructure to manage.
• 05

  Next.js
  The template browser and editor needed fast load times and a good mobile experience. Next.js handles both with server-side rendering without additional infrastructure.
• 06

  AWS SQS
  Render jobs are queued in SQS before triggering Lambda. This decouples job submission from execution and gives the system natural back-pressure handling — jobs never overwhelm the Lambda pool, and the queue makes retries and dead-letter monitoring straightforward.
• 07

  AWS S3
  Finished video files are written to S3 by the Lambda renderer. S3's durability and direct-link generation meant we didn't need a separate media server to store and serve files that can be hundreds of megabytes each.