---
type: "article"
title: "What’s Cookin’ in Hook-land? (Vol.5)"
summary: "Vol 5… pregnancy announcement?! Some fun/interesting news; Summer doldrums; Line up your QA!; and more"
newsletter: "Left Hook"
newsletter_handle: "left-hook"
newsletter_url: "https://usecommune.com/n/left-hook"
author: "Sean Matthews (@seanspeaks)"
published: "2025-07-30T03:03:03.000Z"
canonical_url: "https://usecommune.com/n/left-hook/a/what-s-cookin-in-hook-land-vol-5"
markdown_url: "https://usecommune.com/n/left-hook/a/what-s-cookin-in-hook-land-vol-5.md"
chat_url: "https://usecommune.com/n/left-hook/a/what-s-cookin-in-hook-land-vol-5/chat"
source_url: "https://lefthook.kit.com/posts/5"
body_source: "imported"
likes: 1
replies: 0
body_words: 2567
---

# What’s Cookin’ in Hook-land? (Vol.5)

***Newsletter iteration note:****After yet another Friday where I was making a decision on what thing I had to punt on... I've decided to move to Tuesdays for the drop. Combining that with taking notes as the week progresses should help make these more regular and reliable. As always, please reply and let's talk about whatever catches your attention.*

# On a personal note…

“I need you to cancel your lunch.”

“Okay. Can you give me a preview? Everything healthy?”

“Well… they are BOTH wiggly and healthy, and I’m 14 weeks in.”

And that’s how I found out we were having twins, about 3 weeks ago. I was at the YMCA with the three littles, taking advantage of their Kids Club offering to families.

Once we figured out the due and delivery days (January and December respectively), we also quickly realized- we are going to be a family of 5 under 5!

That day didn’t change everything. Finding out we were pregnant on the day of our youngest’s first birthday… that changed things a lot. Adding a second to the mix… that complicates things, but we knew it was a possibility from how the first trimester was going.

Still. We live in a tiny condo. We thankfully already bought the minivan.

It’s not like we can get any more tired. And every few months that go past, the other kiddos become more helpful and self reliant. Basically, praying for no complications, and then expecting a few months of intensity, and then this will be “the worst” it will be.

Downhill (or growth?) from here.

If it sounds like I just brain dumped all of the thoughts that ran through our heads one after another... it's because that's just what I did 😂

We've received excited responses, support, and lots of conversations around how "if anyone can do it, you guys can!". We're gonna be great, and I'm excited to live life in stereo for a while (thinking about walking into the nursery with dueling banjo cryers).

I’ve asked myself whether this belongs in a business newsletter. Do the internet stranger/friends need to know? Will it help anything?

I read a quote from Brené Brown recently (paraphrasing)- **your story may be someone else’s survival guide. Tell it.**

That’s how I’m justifying the sharing to myself. If you can follow along while I figure this out, it may resonate and help you out in some way. Or someone you know. Etc.

I’m figuring out if this actually fits for you. If it doesn’t, that’s okay. I won’t be talking about it too much, but I do think there’s something to learn while watching how someone can run their company while also making the best moves and decisions for their family.

Which brings me to why I keep doing this newsletter.

What am I doing here? Is this something just for me to get my name out there? Am I trying to establish credibility? Am I workshopping ideas out in public? Do I really care about this space?

These are the questions I keep asking myself, and I think the answer ultimately comes down to: **I need to be able to process what I’m doing in a way that’s both public and accessible to people.** This is a key component missing from our work, getting feedback that I’ve been needing for years on how my ideas are actually relevant to people building product in the trenches every day.

I ultimately view myself as someone who likes to create. There’s barely a day that goes by without a new problem or a thing that I think should exist in the world that doesn’t. Cursed with a creative impulse, basically.

And also, I’m just here for the grind.

So, I write to process all of that.

Maybe I have my why? I know that twins coming does not equate to “okay, write more newsletters.” But I think having discipline and doing something over and over again will be a steady practice for me, grounding while grinding (so to speak).

By the time I get to volume 700, the twins will be 14. They’ll be in high school. Ollie will be attending college or figuring out what he wants to do. Scout will be driving. What will I be? I’ll be just a dad that embarrasses his kids. But hopefully along the way I’ll have formed a tribe of people that have boosted me up and given me ideas and helped me move forward with how to contribute to the world in a positive way.

And that’s what you’re getting out of this. You’re seeing that exploration.

Best,

~Sean

# Quick Note on Format

Before we dive into the meat of the newsletter, I wanted to take a moment to review how I’m thinking about this format. You’ll see here we have a diagram (thanks to ChatGPT, who my kids decided to call Derek):

​

![](https://embed.filekitcdn.com/e/gMtzGmZ1nP1uogiQWhDnPA/7mRg26Y5c3MXajWWPdWgKx/email)

*Newsletter framework diagram showing three panels: What We’re Seeing (person observing market/news), What We’re Hearing (people in conversation), What We’re Doing (person working at desk)*

The intent for each section:

**What we’re seeing** should be something akin to looking out over the marketplace and seeing news items that come through. Think of it like how a bike shop owner will see product announcements - a new building material for bikes or a new brake that came out. They’ll be aware of the nuances to a very deep point of view that the casual bike owner probably can’t get to.

We’re doing the same thing for integrations. As we see events and news things that roll out, you might see the same thing and not be able to see the same insight that we see.

**What we’re hearing** is more like a doctor or therapist for people in the trenches trying to do the work. We’re pulling out broadly applicable lessons or topics from client conversations - things that we think shouldn’t be the case, so let’s publicly talk about them and see if we can help someone else.

This commonly comes down to blockers, delays, frustrations that people are having, or how to set up things correctly.

**What we’re doing** is effectively just telling you about the things we’ve been working on, both for clients and internally. Our clients and people we talk to don’t often know what we actually can do or have been doing for other people, so this acts as inspiration - “oh there’s some stuff I haven’t thought about doing yet.”

***

# What we’re seeing

**Technology partnerships coming together for integration use cases**

This week I saw an interesting case of three companies coming together to talk about integration use cases as a[technology partnership webinar](https://www.linkedin.com/posts/openphone_icymi-we-are-hosting-a-live-event-with-hubspot-activity-7355625408597237761-1GW8?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAebMZ8BQwmRCusyaW979DrWyU07mcA4Vxo). I think it’s interesting to focus on both the type of companies involved and why it works - I’d love to see more of this kind of thing across the industry.

The concept: instead of each company trying to explain their integration story in isolation, they’re collaborating to show a real-world use case that spans multiple platforms. It’s like saying “here’s how our three technologies work together to solve an actual business problem” rather than “here’s our API documentation, good luck figuring out how to use it with other tools.”

It's also interesting to note that it's HubSpot, OpenPhone, and Make.com. Why interesting? OpenPhone HAS a HubSpot integration/app. And HubSpot HAS a workflow tool! Why would they include Make into the mix?

Because of something else that's true about integrations that I posted about recently- you need a healthy mix, and each of these partners has something they excel at, without competing. It will be fascinating to tune in and hear how they lay the case out to their users.

This is smart. Integration stories are more compelling when they’re concrete and cross-platform.

**The trough of disillusionment for AI coding and integrations**

There’s been tension lately around “vibe coding” and I’ve seen people talking about hitting the trough of disillusionment. I haven’t seen anyone directly talk about integrations in that mix, but one thing I’ve experienced in my own AI coding adventures is that **integrations can be a source of big frustration**.

I’ve also seen this directly in client work. It would be interesting to analyze everyone’s AI coding frustrations and see how often the issues came down to integrating with external technologies.

There are different ways to solve for that - being more intelligent about how you’re prompting - but at that point it’s more context and prompt engineering. The thing is, when you’re working with external APIs, you’re dealing with: - Documentation that may be incomplete or outdated - Rate limits and authentication quirks
 \- Different response formats and error handling - APIs that change without notice.

AI is great at generating code, but it struggles with the nuanced, real-world messiness of API integrations. And "testing locally" isn't exactly easy... you're constantly running through integration tests, and I can imagine that by the time a user is wrestling with its AI buddy in a spiraling loop they may want to chuck their computer out a window.

![](https://embed.filekitcdn.com/e/gMtzGmZ1nP1uogiQWhDnPA/4dy5GRFxn4FvF81abm546g/email)

*ChatGPT (Aka "Derek")'s amusing attempt at a stick figure throwing a computer out of a window. Figured the challenge of directionality was maybe a match? Turns out, it couldn't figure it out. 😂*

**MCP-UI: Making AI tool interactions more visual**

Less news, more just a thing we saw released that’s really compelling. Someone put out essentially an extension to the MCP (Model Context Protocol... for a refresher, take a peek at [our first issue](https://lefthook.kit.com/posts/1)) called [MPC UI](https://mcpui.dev/).

The concept: instead of MCP servers just returning text responses, they could return UI components that could be rendered. So when an AI tool calls your API, instead of just getting back data, it could get back an interactive interface.

![](https://embed.filekitcdn.com/e/gMtzGmZ1nP1uogiQWhDnPA/aQUDQVaVLcFHD93pXVnMPQ/email)

*Example interactive UI component rendered in a chat stream, from a tool call (i.e. that's live data from some service, and that's live interaction with that service via buttons!)*

This is interesting for three different use cases:

1. **Inside your own app** - dynamic UI rendering based on AI interactions
2. **Within chat widgets** - something like Claude or ChatGPT where they could tell their technology partners “give us your MCP server and we’ll make integrative use cases for your users, but let’s make this more interactive”
3. **Standardized app interfaces** - there’s no standardized way to determine how you want your app to show up in other places, and this feels like a push in that direction

Chat is great for getting into a thought process, but when you want more structured information and interaction, it’s really helpful to have deterministic UI. Getting that generated and being able to define it once to effectively cater your users’ experience in different chat environments is really compelling.

I’ve been experimenting with this specific project. Happy to talk about what it looks like to build your own.

# What we’re hearing

**QA and testing still gets blocked, no matter where you are in the process**

This week has been all about following up with previously delivered work and making sure we’re getting unblocked on testing and QA. It’s pretty interesting to see that blockers can show up even in the most well-staffed organizations. It's usually details or permissions or configuration that takes some special knowledge. (In this case- turning on specific permissions for a test account)

The pattern keeps repeating: integration is “done” from a development standpoint, but then it sits in a queue waiting for: - Test accounts to be set up with the right permissions - Someone to actually test the integration end-to-end - Feedback cycles that take weeks instead of days - Sign-offs from people who are stretched thin

The bottleneck isn’t the code anymore. It’s the human coordination around making sure the thing actually works as intended.

​

**Things people wanted months ago are still waiting to happen**

We’re finding that things people thought were priorities a few months ago are still sitting in backlogs. The summer doldrums are real, but I think there’s something deeper here.

When development speed increases (thanks, AI), but organizational decision-making and prioritization processes stay the same speed, you get this weird backlog buildup. It’s easier than ever to build the thing, but just as hard as ever to decide what to build and when to ship it.

# What we’re doing

**Personal automation experiments**

***Pro Tip- even if you're not a developer, asking for a script you can run many times is often better than asking an LLM-based AI to do some task over and over again***

I’ve been finding that for personal cleanup tasks, instead of asking Claude to do something over and over, I can ask it for a script to do the thing. That script becomes repeatable, deterministic, and debuggable.

This came up in a few client conversations too. There’s a moment where instead of asking an AI to solve your problem each time, you ask it to create a tool that solves your category of problems. The benefit is that the tool is predictable, every time. And because it's a "throwaway" script (i.e. just for you, on your machine or in your Google Apps Script etc), you don't need to worry about "breaking things in production".

It’s the difference between “can you help me clean up this CSV?” and “can you write me a script that cleans up CSVs like this?” The second approach gives you something you can use again and again, modify, and understand. With predictable results. Faster. Costing way, way less.

Give it a shot this week. Next time you ask an LLM for something, ask it instead for a script to help do that thing.

The other thing I did was add more functionality to my personal MCP server that I’ve plugged into Claude. It’s been fun to experiment with that - nothing that’s modifying my whole workflow, but it’s getting closer to where I wanted it to be.

**Organizational process cleanup**

We’re doing cleanup on how we track conversations with clients and the things we’re providing for them. Over time, especially when you’re deluged with different things, it’s harder to focus on what matters.

That’s where I think getting a trusted and useful recurring email is helpful - it gives you something to act on instead of everything feeling equally urgent.

**Project work: HubSpot builds and popup limitations**

We’re kicking off another HubSpot build and looking at the different intricacies of their popup models and custom UI components. They’ve made some cool improvements to the platform and UI extension concept, but there are still some limitations to what you can do within their ecosystem from a UX perspective (I can audibly hear the frustration from designers everywhere 😂)

It’s a good reminder that even as platforms become more extensible, you’re still working within their constraints. The art is figuring out how to create great user experiences within those boundaries.

Anywho, we’re still looking for more companies to take us up on our [5 company unlimited integrations offer](https://learn.lefthook.com/unlimited-integrations). We have a few interested, hoping to close out the cohort by the end of July.

Reply to let me know your thoughts. What do you want to know more of?

Smash that unsubscribe button below if you don't like it.

Forward to a friend if you do!

# Until next time!

Sean Matthews
Co-Founder
​Left Hook​

***

## Discussion

No replies yet.