David Gu

conversations are the world's largest dataset.
ceo @recallai

[David Gu @davidgu]

We've raised our $38M Series B, led by @BessemerVP, with HubSpot Ventures and @SalesforceVC joining.

Over 2,000 companies (including @HubSpot, @clickup, @useapolloio) build on @recallai, the API for meeting recording.

[Sam Bhagwat @calcsam]

Building agents that's fast enough to help a human on a sales call respond in real time is no joke

But @thatguybg and @microhq figured out how to do it with @mastra + @recallai:

[Nik Samokhvalov @samokhvalov]

something new just happened here, at our live Postgres hacking session:

https://t.co/pfzgcz5fKk

@recallai was streaming our words right to @claudeai code, so we worked on new @PostgreSQL patches (support of logical decoding on physical standbys replaying WALs from archive) just talking

claude code became acting participant in this zoom call

going to explore this model further

look at it here: https://t.co/pfzgcz5fKk

[David Gu @davidgu]

@TrentonL we came across that in the docs afterwards, and we'll be doing that in the future. Appreciate your comment!

[David Gu @davidgu]

we had an incident because we migrated traffic to a brand new s3 bucket.

our millions of servers instantly crushed the new bucket’s partitions and started getting slammed with 5xx errors

[David Gu @davidgu]

@SJP1804 we did a slow rollout and rolled back as soon as we saw the errors

after tracking down the root cause, we restarted with a slower rollout, which worked

but it was definitely surprising to us that S3 buckets "learn" your load pattern, and new buckets need to be trained!

[David Gu @davidgu]

@QuinnyPig if one day you come across software that is NOT bad, please let me know 🥲🙏

[David Gu @davidgu]

@trycua congrats on the launch!!

[David Gu @davidgu]

@motatoeshq hundreds of thousands, and growing fast :)

[David Gu @davidgu]

@Mo69796244 auto partitioning works well for predictable workloads that scale up and down gradually.

too bad that real world workloads often don't fit that criteria!

[David Gu @davidgu]

The previous bucket had intelligently adapted its partitioning to our specific traffic pattern.

The new bucket had zero history and its default partitioning was a terrible fit for our workload.

Fun fact: when an S3 server is overloaded it will return a 503 Slow Down

[David Gu @davidgu]

@RAGNAR404_

[David Gu @davidgu]

@njpatel if only we could explicitly specify the partition structure we wanted...

[David Gu @davidgu]

@xav_db behind every magical infinitely-scaling service is a pile of regular servers and an algorithm that really dislikes bursty workloads 🙃