Deyta Weekly Digest

A Hey, welcome back to the Deyta weekly — it's Friday, February seventh. I'm Alex, joined as always by Sam. And honestly? This was one of those weeks where it felt like a lot landed at once.

S Yeah, it really was. We shipped the new pipeline engine, had a gnarly incident mid-week, closed our biggest deal of Q1 so far, and we've got three new people starting Monday. So — lots to cover.

A Let's get into it.

A Alright, biggest thing first. The pipeline rewrite is live. The team flipped it on Wednesday for about 10% of workspaces, and so far the numbers look really promising. P95 latency dropped from around 1200 milliseconds down to about 340.

S That's a massive improvement. And honestly hats off to Priya's team — that was a three-sprint effort that kept scope tight. There's already a follow-up epic for expanding to 50% of workspaces by end of month.

A On the platform side, the webhook retry overhaul also shipped. Miguel's been working through that one pretty much solo — the new exponential backoff with jitter is already reducing failed deliveries. Early data suggests about a 60% drop in permanent failures.

S Yeah, and there's a nice RFC in Notion from Miguel on the retry architecture if anyone wants the deep dive. Worth a read.

A Okay, so — the not-so-fun part. We had a Sev 2 on Tuesday.

S Right. The ingest API started returning five hundreds around 2 PM UTC — basically a Kafka pipeline clog. Turned out to be a connection pool exhaustion issue on the primary Postgres cluster. Time to detection was about four minutes, which is solid, but resolution took closer to forty-five.

A Elena ran point on that one and the postmortem is already up. The root cause was a missing connection timeout on a new query path that got introduced in last week's deploy. The fix is in, and there's a follow-up ticket to add connection pool alerting thresholds so we'd catch it faster next time.

S Honestly, the response was clean. Four minutes to detect, clear incident commander, good comms in the Slack channel. The forty-five minute resolution is the piece to improve, but the action items look solid.

A Alright, shifting gears to the revenue side. Big week in the pipeline.

S Yeah, so the Meridian Health deal closed. That's the one that's been sitting as a late-stage opportunity for about six weeks. $120K ARR, 12-month contract, and they're planning to roll out to their entire platform engineering org.

A That's our second-largest deal this quarter. And on the pipeline side, we've got two new enterprise opportunities — Northvolt Energy came in through an inbound demo request, and Cascade Logistics was sourced by Dana on the outbound team. Both sitting in early discovery right now.

S Total active pipeline is sitting at about $1.4M, which is up about 15% from where we were at the start of the month. Conversion from discovery to proposal has been running around 30%, so the math is starting to look healthy for Q1 targets.

A Speaking of customers — a few things worth flagging from the customer channels.

S Yeah, Acme Corp — one of our bigger accounts — raised a thread in their shared Slack channel about sync latency on large datasets. It's not a support ticket yet but the CS team is already looped in. Something to watch.

A On the flip side, the team at Beacon AI posted in their channel that they've fully migrated off their legacy pipeline and are now running 100% on Deyta. That's a nice proof point, especially since they were one of the early beta users of the new engine.

A Alright, people updates. We've got 3 accepted offers through Ashby this week.

S Yeah — a senior backend engineer joining Priya's platform team, a product designer on the growth squad, and an SDR on the sales team. All three start Monday, so if you see new faces in onboarding, say hi.

A And just a heads-up — the engineering manager role for the integrations team is still open. Interviews are moving into final rounds next week, so if anyone has referrals, now's the time.

A Alright, that's the week. What are you watching for next week, Sam?

S I'm keeping an eye on the pipeline engine rollout — going from 10% to 25% of workspaces is the next milestone. And the Acme latency issue — I'd like to see that get ahead of before it becomes a formal escalation.

A For me, it's the Northvolt discovery call on Tuesday. If that progresses, Q1 is looking really strong. Alright, that's a wrap — have a good weekend everyone, and we'll catch you next Friday.

S See you then.