block2

# Transcript: 2026-05-06

> 13 time blocks from 9:57 AM to 12:10 PM

---

### Planning remark fragment
**10:37 AM - 10:37 AM PDT** | *work*

**Microphone:**
And then we'll have our plan.

### Work task discussion with dog
**10:53 AM - 11:13 AM PDT** | *work*

**Microphone:**
Start, stop. Get over here. Poe! You're right. Get over here. Maybe it's a man.

Hey, how are you doing, buddy? Hey! Poe, get over here.

Oh, oh, oh. No, just a hug. I'm sorry. Or a lot. I have one. Why don't you come over here?

There's a little under cover on it. There's another cover on the grass.

One, two, three. I'm sorry.

**System Audio:**
Mm-hmm. Above the wallpaper, hello in this room. Very nice, good taste, so I can't take credit for it or anything.

All right, okay. You have a task to work on just for like an hour or two? Cool.

Works for both companies and people. But just take that job and move it into its own queue, like batch indexing or something. I manage the queues now. We should always have one worker running that. Okay, yep, that makes sense.

User notifications and default queue onto their own worker. Those people jobs should be on a queue separate from indexing. That way we can pause the morning refreshes or things like that if we have to.

I just talked to David. I spent the past two hours—well, two hours since standup. That makes sense. It's simple enough. Here's where I don't want to turn the SFDC stuff on anymore. I want to delete it and start fresh.

I don't think I know so much about it. Here's what we had turned on with Prashant at Postman: the user sync. So we're getting every user, thanks, we get their IDs.

And so I did try to explain, because he was like, "David, you're not—" If you get the list, we can update the list at 4 p.m., but at 11 p.m. every night. And he's like, "Well, it shouldn't work." I'm like, "Syncs have a definition." Thinking something definitively means I have. I post in Postman what we're blowing at all the target companies—130 or whatever some odd partition files. Why it's taking hours to run, I don't know, but we're taking it.

On the companies in the file, I'm skipping the account object, but it's other stuff too. I have TCs before—now they're only going to have 4,500. They are sending 30-some-odd thousand, and 15,000 of those are owned by reps for SmallWorld users.

Okay, it's like I'm going to have to explain to them why they're not seeing 15,000 records. And I said, "Well, if you want, we can just create all those companies and we have 10,000, you know, credits. So we can also write some real quick scripts to enrich that data a little bit." Just worried on the companies themselves—10,000 are...

So here's to research—worthless in the system by name. It's maybe not perfect, but it was a pretty aggressive search. Yeah, I still worry. We're going to display nothing for. Yeah, that's true, but that's okay because if their domains, I might be a thing we do where we also then, as part of the enrichment over the next couple days, have to run the employee search through for those 10,000 companies.

Like I hear you. Well, that all sounds good to me. Have any concerns? Two, um, everything. I could use a start from scratch sort of vibe, and most reconciliation because Claude is so confident. But I know there's not a whole lot we can do unless we want to like go it. Yeah, which just doesn't make sense. But yeah, feasibility-wise, no concerns there.

I'm going to start working on this and then I'll share what David's going to share with Postman.

And I'm blaming this on our user sync having a bug where we weren't updating affiliation IDs, and that's why Owen's work wasn't being updated and Bruno's work wasn't being updated. There are other things at play—issues with the queuing, issues with looking at account, not target company. It's easier to point to like, "This is a definitive bug." It was where things could be awkwardly wrong.

I read at least what you shared with Postman. It made sense. Cool. I just want to make sure you're on the same page. I'm going to get on that. I'll talk to you. Don't worry. Sounds good. Talk soon. Bye.

### Brief personal exchange leaving
**11:34 AM - 11:44 AM PDT** | *personal*

**Microphone:**
I'm sorry. You... Let's go. I'll be back.

### Casual chat about work stress
**12:10 PM - 12:10 PM PDT** | *casual*

**Microphone:**
Yeah. Let's go. What's bugging you at work? I'm just getting a bunch of messages. I'm making sure.

Fuck those guys.

Huh?

It's warm in here. Yeah, the space heater's off. I know, but should we close the window? Is it warmer outside? I don't know. I mean, it is pretty nice and warm outside.

---

<details>
<summary>Background Noise (9 blocks)</summary>

### Silent or near-silent capture
**9:57 AM - 9:57 AM PDT** | *background-noise*

**Microphone:**
.

### Brief audio filler fragment
**10:02 AM - 10:04 AM PDT** | *background-noise*

**Microphone:**
Thank you. so

### Single thank you utterance
**10:10 AM - 10:12 AM PDT** | *background-noise*

**Microphone:**
Thank you.

### Repeated filler phrases
**10:17 AM - 10:18 AM PDT** | *background-noise*

**Microphone:**
Thank you. so so

### Single thank you utterance
**10:40 AM - 10:40 AM PDT** | *background-noise*

**Microphone:**
Thank you.

### Departing with background audio
**10:44 AM - 10:48 AM PDT** | *background-noise*

**Microphone:**
I'm going to go. Thank you.

**System Audio:**
Thank you. Woo-hoo-hoo-hoo! Thank you. Bye. Woo-hoo-hoo-hoo-hoo. Thank you. Mm-hmm. Thank you.

### Filler sounds and thank yous
**11:18 AM - 11:21 AM PDT** | *background-noise*

**Microphone:**
. .

**System Audio:**
Thank you. Mm-hmm. Thank you.

### Moving to next video
**11:47 AM - 11:48 AM PDT** | *background-noise*

**Microphone:**
I'm going to go to the next video.

Thank you.

### Unintelligible night fragment
**12:03 PM - 12:03 PM PDT** | *background-noise*

**Microphone:**
Or in the night. Lines of that.

</details>

The morning of May 6th was dominated by a work discussion (likely with a colleague at SmallWorld) about Salesforce (SFDC) data sync issues and a plan to start fresh. A critical bug was identified in the user sync where affiliation IDs weren't being updated, causing downstream failures for Owen's and Bruno's work. The conversation covered queue architecture changes — separating people jobs from indexing into dedicated queues — and a strategy for handling ~15,000 company records from Postman/SFDC, including enrichment of ~10,000 companies using credits and employee search. The decision was made to scrap the existing SFDC integration and rebuild, with David preparing to share details with Postman.