Blog

From the Serveka team.

Engineering posts, product updates, and technical guides.

CompanyMay 12, 2026

We built Serveka because meeting bot infrastructure is needlessly painful

Building a meeting bot from scratch means Playwright automation, WebRTC audio capture, Deepgram WebSocket management, VM orchestration, self-deleting infrastructure, and six months of your life. Or you pay Recall.ai and hope they don't change the pricing.

Read post
EngineeringMay 14, 2026

How we got bot join time from 2 minutes 45 seconds to under 60 seconds

Every cold start pulled the Docker image. 1m43s just for that. We fixed it with warm machine images, zone fallback, and pre-initialized browser instances. Here's the full engineering breakdown.

Read post
EngineeringMay 19, 2026

Why we built a plugin system instead of shipping features

Most meeting tools are monoliths. Every feature is always on, even if you don't want it. Our approach: core handles bots + audio. Everything else is a plugin.

Read post
EngineeringMay 22, 2026

Meeting transcription at scale: batching, deduplication, and delivery

Deepgram sends segments one at a time. If you forward each one as an HTTP request, you'll drown your API. Here's how we batch, deduplicate, and deliver at scale.

Read post
ComparisonMay 28, 2026

Serveka vs Recall.ai — an honest comparison from a competitor

We acknowledge where Recall is strong. This isn't a hit piece. It's a fair look at what each product does well and when you should use each.

Read post
ProductJune 2, 2026

I connected my calendar and never pasted a meeting link again

Connect Google Calendar, set rules, forget about it. Here's how the auto-join logic works and the edge cases that tripped us up.

Read post
TutorialJune 5, 2026

Building a Slack bot that posts meeting summaries — 80 lines of code

Set up a Slack app with incoming webhooks. Configure Serveka webhook for bot.completed. Post summaries to Slack automatically. Full code walkthrough.

Read post