Doctor Droid bets on-call runbooks become executable notebooks after YC W23 pivot

Dipesh and Siddarth came from an Indian food-delivery startup with a busy on-call rotation and a small tech team, where business-impacting issues such as orders dropping more than 5% in 15 minutes escalated to Dipesh as the long-tenured lead dev. To avoid being the bottleneck, he pre-wrote scripts that fetched custom metrics and order logs every five minutes during peak traffic, so engineers checked the usual suspects against fresh output before diving into generic exploration.

They entered YC's W23 batch building a data lakehouse with support for dynamic log schemas, realized it was a product in search of a market, stopped, and pivoted to PlayBooks, an executable-notebook idea they had prototyped before YC. PlayBooks turns runbooks into notebook-style documents connected to every part of the stack: fetch logs from AWS CloudWatch or Azure Log Analytics, pull PromQL/Datadog/New Relic metrics, query PostgreSQL or ClickHouse, run remote bash, hit custom APIs or check EKS/GKE events, with each step displaying its output next to the instruction.

The bet is that on-call investigations can be scripted and automated so alerts trigger the usual-suspect checks before a human is even awake, with results shared across the team. The 2024-06-04 Show HN drew 147 points and 35 comments; the founders said they were piloting with a large enterprise of 100+ SREs that had struggled to make Jupyter notebooks self-serve for its product teams, while commenters questioned whether teams would adopt a Jupyter-style workflow. By the 2026-09-04 crawl the repo showed 460 stars, 33 forks and 2,960 commits, with docs, sandbox and demo alerts still live.