Ep. #5, Maxed out on AWS, Still Falling Over with Scott Mitchell

Christine Spang: Hey everybody, welcome to the show. Today we have with us Scott Mitchell.

Scott was the CTO at Salesloft for seven plus years and since leaving Salesloft, he is taking on a bunch of fractional and consulting work, helping other people scale their businesses, especially with a focus on B2B SaaS.

And we're super excited to have Scott on to learn from and chat with today. Welcome Scott!

Scott Mitchell: Thanks for having me.

Christine: I'd love to just hear from you a bit, a little bit of the story of scaling Salesloft.

Like how big was the company and the team like when you joined and just take us through a little bit of the journey and some of the stories on, really this like multiple orders of magnitude of growing that business.

And maybe also kick off with like a little bit of just like, what is Salesloft? What does it do? I think some of our listeners might not be familiar.

Scott: Sure. Yeah, Salesloft, one of the things that was attracted to me about it in fact was one of the neat things with Salesloft was starting a new sales category and that became known as sales engagement.

And so what the heck is sales engagement? The way I like to think about it is if you know Salesforce as a CRM, think of sales engagement and Salesloft as being kind of the, the system that rides alongside.

If CRM is turned into this management reporting system that sales reps hate to use because the only thing they're there, yeah, they're just being nagged to log all their work there. Salesloft is the system where salespeople work, right? It is their workflow system.

And so it makes the sales rep's lives easy, managing all their contacts, all their prospects they're reaching out to, all the communication channels by which they do that. And then it logs all that data back to the CRM.

And so it makes both systems and both user groups happy, right? The sales reps are happy because they've got a system that's helping them do their job better.

The sales managers and leaders are getting better quality data in the CRM and have better fidelity and nobody's nagging each other anymore, right?

And so I joined Salesloft about right around 90 employees, about 7 million I think it was in revenue at that time.

And it was kind of weird that that actually had become a pattern for me. That was the third company I joined at that stage where we had product market fit and things were starting to heat up and time to figure out how to scale the business, right?

And so in terms of stories, Salesloft was a Ruby on Rail shop, it was my first time working with Rails.

I, obviously like everybody played with it, but it was really fun to kind of get in the trenches and understand where things are going here.

We were AWS based and then that kind of initial survey of, reaching out to everybody in the team, kind of getting to know everybody, what concerns do you have, a point came up that, oh, well we were already on the largest RDS instance, relational database instance that Amazon provided.

We had probably better figure out what was going on there. And so right away there was our first scaling challenge. And so that kicked off. Literally remember sitting out with our head of DevOps and asking, "All right Mike, well how long do we have?"

And he went and did some projections based on our current growth numbers and we had 12 months. If we didn't get it done within 12 months, we were going to have some real issues.

And so we very quickly did some tech evaluation, figure out what we wanted to do, decided to implement a horizontally scalable version of Postgres that we could live with for the long term.

And kicked off a project to migrate the entire application to that. The good news is Mike was right, we had 12 months. The bad news is it took us 13 months to execute the project.

So yeah, the month of September, 2017, will kind of always be in my memory. We were dealing with running right up against capacity, struggling with risking outages every day, making really quick changes to the system to actually allow it to survive that month while we finish the last little bits of testing and confirming, because this is one of those migrations, right?

We're changing the system all the time and always have a fallback plan. And this is one of those rare occasions where there's really not a great fallback plan.

Once you cut over, you're going to really need to stay there. And so we got it done, first few days of October, 2017 and largely smooth but yeah, that was quite a thing to walk into day one and say, okay, well we got our marching orders right away.

Hard to think of other really specific scaling stories per se, but--

Every year we used to joke it was a new company, right? Just 'cause we were going so fast that things were always needing to be reinvented, right?

Process, technology, hiring positions that we didn't have on the org chart in the previous year. It was definitely a place where people needed to be ready for change, right?

If you weren't adaptable to change, you really should have been somewhere else. It bred a a neat culture because of that.

Christine: Totally, there's a few things that you talked about that really kind of resonated with me.

One of the things that I think of that we're doing here on this show is like being like historical architects or investigators of where has like business software come from and then, where is it going?

And you talked a little bit about how like the concept of CRM, it sort of started out as being like, well we have this super clunky system of record and it's very horizontal and everyone kind of uses the same system and everybody hates it and then there's a lot of manual data entry and it's not like quite right for anybody.

And I think Salesloft was kind of one of the pioneers of kind of exploding this categorization of CRMs of creating a bunch of new types of applications that are sort of like CRMs but are also more.

They have this sort of custom workflow engine for some specific vertical of businesses.

In Salesloft's case, it was for sales engagement for people who were, were selling things and engaging with sales workflows.

So, maybe lots of different types of businesses but you're selling to sales teams specifically.

And I think that like, you guys pioneered a bunch of different features around just really creating a richer user interface experience for that group of people.

And obviously also like went through a bunch of the like 2010s era scaling challenges of, you're not the first Ruby on Rail shop that I know of that has gone through a big order of magnitude change.

And that's the thing that every company that has gone through like a strong period of growth. You're always doing migrations, you're always rewriting things like every sort of 10X growth, there's something else that becomes a bottleneck and falls over.

Scott: We always joke about we're always moving the bottleneck.

Christine: Yeah.

Scott: It really doesn't matter your tech stack, right? You're just always optimizing, optimizing out and just moving that bottleneck from one place to the next, right?

Christine: I think your background, you really focused a lot on kind of the backend and the technology that was powering all this functionality.

But I'm wondering if you had any sort of insights to share around what were kind of like features that were kind of new in the sort of 2017 to 2019 era that Salesloft kind of pioneered and essentially like raised the bar for these types of applications.

Scott: Sure, I think one was just introducing a phone dialer the way we did.

I come from a background in email, right? And so we managed a ton of email communication and such and this became, the answer was no, no, no, it's not just about email, it's about all the communication channel, right?

And so integrating a dialer that was purely software based, purely browser based, that was tightly integrated into the workflow, I think was not unique.

But that was a time where that was becoming something that could be done really well.

And then the second piece of that was then introducing a Chrome extension that rode along with the sales rep no matter where they were so that we could be there in their email client, that we could be there in their CRM.

We could be on a random website where they looked up a prospect they'd want to reach out to and right mouse click on them and called them, right?

So having that flexibility that chrome extensions provided was really, really powerful. At the same time, years later, Chrome and Google recognized that also there could be bad things about having Chrome extensions that were that present and that had that much access.

So there was an interesting kind of push and pull from a security perspective around making sure that Chrome extensions in general and of course ours specifically played by all the right rules so that we are honoring privacy concerns in that regard.

But I think those were kind of bringing the software to where the user was, was a pretty unique aspect of our software.

Isaac Nassimi: Scott, I recall reading or hearing somewhere that a surprising percentage of the usage of Salesloft users occurred in like the Gmail interface, right?

Where they're actually using the Chrome extension is kind of this, I guess to say it poorly, it's native extension to the way that they normally like browse their email or do these things.

Was that surprising to you? And then also how did that make you guys kind of pivot the way you were looking at the product or the technology or even the way you were staffing this stuff?

Scott: Yeah, it is true and yeah so for others to understand what we were doing there, the largest part of that Gmail extension was the ability to loft an email.

You're sending from your Gmail account, but you're also sending it through Salesloft so that it gets automatically logged and sent back to the CRM and becomes a part of that prospect's record forever.

Also, we had additional context we could show in the inbox as you're reading emails from prospects, you had the side panel that showed context about that prospect so you could know all the communication you've already had with that prospect, opportunities that are open, all that sort of context.

Christine: So you were kind of adding in these layers of communication context.

First you had the emails that, maybe people were sending them from Salesloft, capturing them from Chrome as well, making them send through Salesloft and you said you added this dialer.

I'm assuming that means like just like making a phone call through the browser?

Scott: Yep, through the browser. And you could do that while you were sitting in Gmail as well.

So wherever and you know, I think the point there was, you're right, it is popular and so what does it mean when there's the identity crisis a little bit, right?

If Gmail is the primary user interface all my users are seeing, what is Salesloft? And so it did guide us towards some philosophies about when we sent people back to the platform, we always wanted them to have connection to the platform itself.

And so there were links out of Gmail that would take you into Salesloft to perform other functions, right?

You only had an access to a small sliver of the overall platform within Gmail. And other companies could approach that differently.

Isaac: I think a lot of product teams would look at, hey, our customers are leaving or coming out of our web app and going into Gmail to do these things and treat it as a problem that they need to solve and decide to replicate a 100% of the features of something like Gmail inside of their platform.

But there's sort of this like nice, like Aikido to taking that momentum of what the users are naturally doing and rolling with it. How do those kind of discussions shake down internally because it's not an obvious solution to land at?

Scott: Right, I think the general philosophy is being available where our users are, right?

And so making sure that we weren't forcing people to go do their basic workflows inside our platform and yet having attractive reasons for them to want to be in the platform, if that makes sense, right?

And to be clear, we're talking Gmail, but also it was Outlook as well. And so we had separate extensions, right?

Because they were different ball games to be kind of technical constraints on how we showed up, right?

Microsoft was much more strict and much more restrictive in terms of what our capabilities were.

So it did mean there were some architectural decisions that had to be made there too like what does code reuse look like in that kind of environment.

How do you also make sure you're providing the right experience for each user in each platform beyond just those kind of software design concerns?

So, it was an interesting kind of yin and yang sort of environment, right, where it's just like you want to make sure that you don't lose the identity of Salesloft, but also make sure you're really meeting the user where they are and with the need that they have.

Isaac: Just thinking about Salesloft, there's a huge service area now that's kind of coming together in my head when I think about it, right?

You guys have this email sequencer, you have the connection to Salesforce, you've got a reporting suite, you've got your dialer, you have all of these extensions and you know, following people around the web so that they can do their BDR work or their operating work no matter where they are in the browser.

How did you manage this whole surface area?

Scott: Yeah and not to mention, to add on to that, conversation intelligence that records all your Webexes and analyzes those and our opportunity management software as well.

And your question meant really is how do we deal with the breadth of the offering and kind of work with that efficiently?

Isaac: Yeah because Salesloft seems to do all those things quite well, which is a unique achievement.

Scott: That's good to hear. I'm glad that's the impression. Okay, so start with a little bit of a weird story.

When you say that what pops to mind is, so my children aren't children anymore. And when we were taking our oldest on college tours, they had a bias towards small schools.

And we made a trip over to UGA here in Georgia, which is a massive school and it shut down pretty quickly on whether that was a school that was appropriate for them.

And the tour guide said something interesting and it resonated with me, which was, you can make a big school small, but you can't make a small school big, right? And by that he meant you could go find your community here and kind of narrow your scope, right? And have a nice intimate environment even though it's a massive school.

By the way, that didn't persuade them at all. They wound up going to Agnes Scott, which is like a 900 person school. But I say that because it sort of resonates with how I felt about managing a really broad product service, right?

The answer is you have many, many small teams that have a narrow focus, have complete ownership of that functional area.

And then architecturally how we facilitated that was through microservices, right? And so you had now individual code bases that in and of themselves were quite focused.

And then the question is then how do you successfully bring that to the user in a way that feels like one coherent, cohesive unit?

And so for us, I think that worked quite well. And I think that the real kind of kicker there culturally is making sure everybody feels vested in their functionality, right? That they know that they are the ones that are responsible.

There's nobody else kind of riding in to save the day, it's going to work or fail based on their skills, talents, efforts, right, their passion.

Which made everybody really vested in, "Well if I'm a conversation intelligence person, well darn it. That's what we're doing and we're going to knock it out of the park.

So then it became a UI game of how do you compose that UI in such a way that user is not concerned or not aware that that's what's going on behind the scenes.

Christine: How did you think about the product with regard to sort of its space and kind of the data space?

Like, you were ingesting all this data from all these different places. Did Salesloft become like a system of record for your customers?

Did they use it alongside a system of record? How did it sort of fit inside the buyer's stack and did that change over time?

Scott: Right, it's sort of a "yes, and" answer, right? The CRM is the system of record or the customer record, right?

And a lot of the activity around it. It's a really interesting question. Philosophically, were we ever really a system of record?

We like to say, yeah, we were the system of record for all the engagement your sellers are having with the customers.

Is that enough to really hang your hat on? I don't know.

But what we were was a really interesting place to go understand that engagement with your customers and what was resonating, what wasn't, what you needed to look out for.

And so it was a very significant data footprint, lots of it unstructured data, right? We had so many call recordings, transcripts, emails, text messages.

And so the real challenge was how do you bring value out of that body of data, right? And that's where thankfully, you know, technology was evolving rapidly in that space to help us then kind of bring meaning from all that unstructured conversational data.

So system of record, I don't know, you can squint at it and try to make that case.

Either way, highly valuable data could be brought together all in one place and be able to be analyzed.

Isaac: You guys seem to have a very, like a almost Skunk Worksy kind of approach to these things, right?

And the only reason I know that is because you guys released all of these tools, you know, for a while and put them on the web, right?

I remember using like subject line grader or email grader, things like that.

That pre GPTs, I mean, I remember using these in like 2019 or something like that, were really able to take an email that you wrote and tell you how well this was going to convert.

Absolutely astounding. It's something that I hadn't seen at that time. And even by today's standards is still considered pretty awesome even with all these advances in AI.

What gave you guys the idea to run it that way and even put this stuff out to the general public in sort of this, you almost used it as like a lead engine, it seemed like.

What was your strategy there?

Scott: Yeah, I think back it up a step, the first question is, well how do we have the knowledge or the awareness of what would work?

We started a data science team in late 2017, I guess it was, knowing that we had this mountain of data that we needed to start getting real insights from, right?

And so we started with not the unstructured data. We had a ton of action data that we knew we could start getting an interesting information out of.

And so, the first real big thing we picked up was what does a good cadence look like, right? There were all these--

I laugh a little bit about it, but when I first got to Salesloft and started learning the sales industry, there were all these studies saying what worked and what didn't.

And what the study meant was the researcher called a bunch of sales leaders and said, "What are you doing and what's working?"

And then collated that information and produced a report. And that was the state of science in the sales industry at that point, right?

And so we were able to say, "No, no, no, no, let's not do that. Let's go look at the real data and see what is actually working."

And so we were fortunate enough to be able to build an engine that actually said, derive a cadence from all this activity data.

If you want something that looks like X, Y, or Z, well how should that be structured in a way that it would be most optimal?

And yet, the very next question we got was great. So we now know the structure of our cadence what should we say in our email.

Oh crap! Yeah, there's the hard question. And so we had lots and lots of experimentation. You're right, it was pre GPT launching.

We built some interesting models that helped analyze that text based on open source tooling that did exist, right? BERT and such.

And then structurally, we had a couple of different research arms. We had that data science team that was super focused on the hardcore data research.

And then we had an arm that was, to your point, very much in the idea of well let's create some free tool that can provide value that to your point, we're lead generation engines, right?

And so they could leverage some of the research that we were doing but not have to go through the rigor producing a SaaS application that was going to be used by a hundred thousand users, right?

And so it became this really neat, you hit it, nail the head, right, Skunk Works sort of deal where we could go experiment rapidly, try some things, shut down some things that didn't do what we wanted to do and let those others just keep riding.

And then some of those then migrated into the full platform over time if they got so much traction that they really needed to be formalized.

Isaac: Yeah, I see that stuff has been rolled into like the Salesloft Innovation Center now and things like that. I think it's really cool.

And my takeaways from using things like the cadence creator and whatnot was that everything was completely counterintuitive to what might be like standard operating procedures for cadencing, right?

Emails with fewer than three sentences seem to convert better than the emails with two paragraphs telling you about a product or service that's being offered.

All kinds of things like that. Everything was counterintuitive. Was that shocking to you guys?

Scott: I mean, I think it depends on who you ask, right? Shocking to us geeks back in the corner who'd never made a sales call. Yeah, probably.

I remember one of the findings was double and triple tap phone calls and the success of calling multiple times in succession and how well that worked.

Like, gosh that seems sort of stalkery, but okay but it worked. And so I think here's an interesting question.

You're digging through all this data and you're deriving cadences. How do you know you've hit something that's finally working, right?

The answer was we didn't have, there was no formal test framework, right? And so we would look over these cadences and go, are these good? I don't know.

And we finally dragged in one of our top sellers and said, help us, over time, look at these cadences and see does this resonate with you? Does this make sense?

And I can very distinctly remember her feedback on the early ones and saying, no, no, no, we weren't close yet, right?

And that over time, kind of between the data science hard data and the great experience of a successful seller kind of working it into something that actually did really resonate was high value.

And you're right, there's some counterintuitive stuff in there.

Isaac: Yeah and I think that's so interesting because a lot of people assume, " hey, we're going to get enough data at scale and we're going to just throw it at some sort of machine learning thing or statistical analysis and we'll just have the answers we're looking for."

And it's never that easy. You always have to compare it against real life and bring in actual experts.

Scott: Yeah, I mean even just algorithmically what in the data are you optimizing for, right? That was not entirely clear.

And so we had to play with a couple of different optimization techniques to figure out, "Oh gotcha!"

So what we settled on ultimately was if you look at it, there's a curve of engagement with a prospect during the lifecycle of a cadence.

And what you're looking for is that inflection point. And once we could find that inflection point, we could then work backwards to see what cadence got them there.

So we had three or four different things we tried before we nailed algorithmically what was delivering the right result.

Isaac: Yeah.

Christine: I'm really curious about sort of looking forward a little bit.

I think we've talked a bit about kind of the evolution of sort of what happened with Salesloft in particular and how this horizontal CRMs have turned into basically workflow specific tools for an audience that just make everybody's lives easier in that specific context.

Where do you kind of think this space is going? Like, we're making a lot of advances right now in terms of AI capabilities.

I think everybody's looking at like, well I need to bring in data from lots of different sources because data is kind of a foundation to AI and I need to do stuff with data and add that to my product.

But what should people be thinking about in terms of like how to drive value with the changing landscape that we have today?

Scott: It is really, really interesting what's happening and so how do they solve--

What's the role of the sales engagement platform in this era, right? And I think it's probably going to wind up being agent based, right?

We used to joke that our philosophy around data science and AI was "Iron Man, not Terminator," right? We wanted to make the humans better, we didn't want to replace them. And I still think that's right, right?

I still think ultimately the selling process is a human to human process. And so I think how these systems evolve, much like the the AI coding tools now, right, is that you now ultimately are looking to have an assistant for the seller who is there to help in every way possible.

Now I think the really interesting question is what are all possible ways, right? And how much insight can these AI agents provide reps that help them make them more efficient.

Now, I guess what I wonder is where in the spectrum of jobs to be done that a seller participates in, right, can the AI help with, especially a large language model based AI, right?

I find people falling in strap of thinking that's the only, the only AI/machine learning that matters anymore.

I still think there's a role for predictive type of AI that we've all used for years. And so the first job to be done is I need to figure out who to sell to.

And is Claude or a OpenAI model or something, is that going to be the thing that helps in AI?

I'm not convinced, maybe, but it could very well be the interface to other models that then help you get that done.

And so I do think it's going to be this very personal interaction with this assistant that is the way the software will manifest going forward. It'll be a really interesting transformation for sure.

Isaac: I'm curious what you think the future of, you know, SDRs and BDRs will be in terms of outcomes.

Because the way I've seen it looking at, SDRs, the really good ones get 20X the results of the mediocre ones.

And I'm wondering if you think that, let's say you have these AI agents that are helping out or even crafting the outreach or even proactively improving things, do you think it will improve the mediocre SDRs and get the results to be closer to that of the high performers?

Or do you think it will widen the gap between the non-performers and the mid-level performers?

Scott: I think even absent this transformation, the SDR, BDR function was dying.

Isaac: Wow!

Scott: I think it's going away. I don't think selling is going away at all, right?

But if you think about an SDR's role is to really be effectively an assistant to a full cycle AE, right? And feed them the leads so that they can then go close them.

I'm not sure that is going to be a distinct role going forward. Now, happy to be wrong, but also cold outreach is also a much different game in 2025 than it was in 2015, right?

All the regulation around what's allowed, all the privacy regulations, independent of regulation, people's insensitivity to being reached out to, right?

I will offer that while I was in seat, you basically couldn't reach me unless I wanted to be engaged, right?

You could email me all you want, you could call me all you want. I wasn't answering. I'm not at all unique in that, right?

And so maybe not the answer you're looking for, but I just don't really think that function necessarily in a sales team exists years from now. I think you've got sellers.

Isaac: I don't think you're wrong, but my question is what do you think it looks like then?

If it gets to the point where you're unable to reach these people who don't want to be reached, how do you sell your software, how do you sell your service?

Scott: I don't have that answer. I think I'm fortunate that I'm not in a job where I need to have that answer, which is nice.

But I mean we joke about, right, my bot will talk to your bot. I mean who knows what's going to happen then, right?

Because you already have bots that are going to look at your email inbox and help you decide just independent of sales just as a person who uses email, what you should or shouldn't be worried about and what you need to interact with.

So your your own software gatekeeper and/or knows you're going to have software bots sending you those emails.

So I don't know, sounds like chaos, but we'll see. That yet again is another opportunity for renovation, right? Somebody's going to step in and try to figure out how to help people connect because ultimately they need to, they want to, just not an unsolicited way.

Christine: Scott, you mentioned that you stepped away from Salesloft about a year ago.

You've been doing some, I think some fractional CTO work and some advising to other companies. What sort of new challenges are you seeing people have today?

And I don't know if you started thinking about whether you want to have like another full-time thing, join another company, start a company but what would you do different based off of all your experience now that you've got a bunch more battle scars?

Scott: All the battle scars, yeah. What my own business is shaped up into is primarily coaching first time CTOs. And just to kind of tee that up a little bit, right?

You've got a bunch of co-founders or engineer number one in the company and companies have found success and so all of a sudden they're realizing their job is a lot more than just building software, right?

And so helping them understand that. In terms of the challenges I am seeing in their landscapes, right, one is very directly related to AI conversation of from two angles.

One, what's their expectation of how their engineers write software, these things, right? And therefore what's the corresponding like velocity expectation?

Do these tools directly increase the velocity of their teams? And if so, what's the best way to use them to optimize that?

And then the other is everybody's got all this pressure, all these existing products to add AI to them, right? And so we have some discussions about how to think about that, right?

It's not, I know your board wants you to add AI, but if you just go to and I'm sure you guys have the exact same conversation, right?

If you just go say I'm going to go add AI to the product, you kind of aren't keeping the customer front and center. And so instead it's really about, you know, old school product development. It all comes down to customer value and AI and these large language models are yet another piece of technology we can use to facilitate that.

So kind of helping them understand how to think about that has been a lot of fun.

I think your other question was basically what would I do differently if I jumped back into seek right now with a startup?

Gosh, I don't know, it's going to be such a radically different landscape in the coming years.

Christine: Change just keeps speeding out.

Scott: Yeah. Seriously, right? Because of so much tech that has been prebuilt for you and you get to continue to leverage more and more of it, yet the acceleration rate is ever faster.

So I don't have an answer there. I think it's going to be super interesting to see how it plays out in the coming years.

What I wonder, I guess to say it out loud is for new startups it seems pretty clear that you're going to start with kind of, we used to joke about, not joke but philosophically you were web first back in the day and you became mobile first.

And now I think you're going to be AI first, right or agent first. And so for a new startup, that's great, that's easy.

The question is what are all these startups with 5 to 10 years of history and a large customer base and a innovator's dilemma, right?

How can they make a pivot that helps them stay relevant and valuable to their customers and not lose them to a startup that just opened their doors a year ago?

Christine: So Scott, if I'm a new CTO just taking on this team and I'm really good at technology but I'm sort of having trouble with the people stuff, how can I get in touch with you?

Are you taking new clients, you looking for folks to work with?

Scott: Absolutely. Obviously they can find me on LinkedIn or you can email me directly, Scott@atlcpo.co. And yeah, obviously always happy to help.

And I think the good news is I like to tell people who are first time managers or first time leaders in this who've got a strong technology background.

Oddly, some of our software design skills can be applied to people too. So yeah, they aren't completely starting over. But it's fun work, right?

It's a fun challenge to help to be in seek and be having these scaling problems in terms of software which is familiar and then people, which is not, right, but they're fun problems to solve either way.

Christine: Awesome. I've definitely had a lot of help from various different coaches throughout my journey in building Nylas and I can definitely vouch for the fact that it's incredibly helpful to have someone to help you navigate that, especially if your business is really taking off.

So, get in touch with Scott, if that's something that would be helpful for you.

Scott: And before we close, I at least want to say out loud that thanks to the Nylas team along the way with the Salesloft journey too. We were obviously partners throughout.

And the email component, obviously, one of our challenges was everybody had a different inbox and we could support a few, but you guys could a ton.

So being able to plug into your infrastructure to serve some customer's needs was hugely helpful. So I appreciate your partnership.

Christine: Of course. We're really happy to have been able to be a part of the Salesloft journey and I think, we're seeing especially as we go into this crazy AI world and like everything's getting super fast and chaotic that like, people is still, basically their apps they need to do so many different things that like APIs are something that's really helpful for people to build, ship features quickly.

And our role, kind of going into this world is really to kind of power this rapid innovation and help people get connections to that data and build some of these features like this.

Isaac: Absolutely. Well cool I think we're about out of time, but first we have picks, right?

So just to recap, as always, every week we bring something cool that we're excited about.

It could be a new application you started using. It could be a book you read, it could be anything.

It could be a random new fidget toy that you picked up or something like that.

Scott: As he hears my fidget toy clicking in the background.

Isaac: We all have something on our desks for that. Spang, do you have a pick this week?

Christine: I wish it were a fidget toy. I could use a few of those. My pick for this week is the reMarkable 2.

It's basically a e-reader that you can write on and I use it all the time for taking notes.

I also use it for reading things like technical papers or even long blog posts in the bathtub, on the couch, out on my porch.

It's a great way to get off the computer 'cause you know, it's just like an e-reader where you can use it in direct sunlight and I just find it's a really great way to get work done and not have to be sitting at my desk all the time.

And also if you're writing down during a meeting, it helps you remember things better so great device.

Isaac: What's the battery life on that Spang?

Christine: I'm not sure how much it is actively using it, but I probably only charge it every couple weeks.

Isaac: Oh my god, nice. All right, that's cool. That sells me on it. Anything e-ink is just automatically awesome in my book.

My pick this week is someone got me Hanukkah actually, the Ember mug, which is a mug that automatically keeps your drink at a certain temperature and the battery life is like six or seven hours or something like that.

And as someone who is a mug addict and like collected silly and stupid mugs, it's completely cured my mug addiction.

And it's actually really changed the way I drink coffee where instead of slugging it down while it's still, at a acceptable temperature, I really feel like I nurse my coffee now and my caffeine consumption has naturally gone down to acceptable human being levels and I don't have these periods of crazy spazziness throughout the day.

Christine: That's funny. I have one of those too. A VC sent me it like 6 years ago.

Isaac: It's at the price point where I wouldn't have bought it for myself, but I would definitely buy another one now.

Christine: Yeah, yours is probably like three generations newer than me.

But it's funny, I have a friend who's like a security hacker and he was like, "You got a software update to your mug? I got to like figure out like what's running on the backend of this thing" and he got kind of distressed and excited at the same time.

Isaac: Yeah, he's trying to figure out how to create a botnet of mugs or something.

Scott: Right, exactly. What's the security exploit for your mug?

Isaac: Yeah. Exactly. Scott, what's your pick this week?

Scott: Well, mine after having completely unplugged from the tech scene for months, right, six months or so, mine is last week or two weeks ago, I started writing an iOS app.

First time of writing any new software in ages and I'm doing it with Cursor.

And lots of people are probably already familiar with Cursor, but man, what a cool experience to step into a platform that I have very little experience with, right?

I've played with Swift a couple of times, but I've never built anything and now have this conversation with a bot that builds an app and in four hours, I think, I have a fully functioning game that I wanted to put together, kind of recreating something we used to do with kids and it was a great introduction to it.

It was like it just blew my mind. Now obviously, how's that scale out to a real industrial strength code base, enterprise scale?

Probably not great, but boy for getting started and getting a code base going, it's been really, really cool.

So yeah, that's what I've been playing with over the last few days, but it's been fun and a neat way to kind of dive back in and write code for the first time in ages.

Isaac: That's awesome. Well, I think we're out of time. Thank you so much for joining us, Scott.

Christine: Great to have you, Scott. It's been fun.

Scott: Absolutely! Really enjoyed it.