Bryan McAnulty, founder of LatchLoop and Heights Platform, has gone deep into AI agents — using them to build software, research, automate work, and rethink how his team operates.

We talk about moving beyond simple prompting into actually delegating work to AI, where humans still matter, how to safely give agents access to your tools, and why understanding the work becomes even more important when AI can execute it. Bryan also shares a simple framework for AI adoption — automation loops, goal loops, and feedback loops — and explains why the real competitive advantage may not be the AI model you use, but the processes and knowledge you build around it.

TIMESTAMPS

00:00 — From Heights Platform to LatchLoop

02:38 — What does 10 billion tokens actually mean?

05:14 — More code doesn't necessarily mean better

07:15 — AI makes experimentation incredibly cheap

08:35 — What happens when AI gets expensive?

12:40 — Can we actually trust AI agents?

18:08 — How much access should you give AI?

19:48 — Why human communities become more valuable

22:58 — Is AI killing online courses?

25:29 — What should creators actually automate?

27:59 — Should you replace employees with AI?

34:52 — Learn the job before automating it

35:49 — "AI does the work. You come up with the ideas."

36:30 — How to get your team to actually adopt AI

38:00 — Automation loops, goal loops & feedback loops

40:39 — Why AI workflows break

42:52 — How to build better AI processes

46:31 — What is an AI harness?

50:12 — Competing with ChatGPT, Claude and the AI giants

54:29 — From software as a service to "service as software"

56:53 — What's exciting outside the AI bubble?

Bryan McAnulty:

LinkedIn - https://www.linkedin.com/in/bryanmcanulty/

LatchLoop: https://latchloop.com/

Heights Platform: https://heightsplatform.com/

─── sounds good. — a podcast with Anna Nadeina

https://soundsgoodpod.com

sounds good. is sponsored by saas.group — serial acquirer of profitable SaaS businesses. If you're thinking about an exit, they're worth a conversation.

https://saas.group/

What if the best B2B marketing strategy in the age of AI is… being more human?

Steve Armenti has spent 16+ years in marketing, including years running demand programs at Google. Now he’s the founder of twelfth agency and Howdy, where he’s betting on a very different version of AI-powered GTM: let AI do the research and repetitive work — then let humans create the relationship.

Timestamps:

00:00 - Meet Steve Armenti
02:40 - What ABM actually means
05:10 - Should a $9/month SaaS even think about ABM?
07:59 - When does ABM make sense for a startup?
10:47 - The economics behind a $1.4M pipeline campaign
12:00 - How Howdy turns personal signals into real experiences
15:42 - Why ABM can build more than pipeline
17:38 - Personalization vs. creepy: Europe, GDPR and trust
22:27 - Can AI research people without losing the audit trail?
25:42 - What should you actually automate with AI?
29:52 - Can AI audit AI?
34:52 - Why complicated AI workflows break
37:37 - Is field marketing becoming more important than demand gen?
41:14 - Why big B2B purchases still need humans
42:18 - How personal is too personal?
45:34 - Steve's AI disaster story
49:37 - What Steve is excited about that has nothing to do with AI
50:38 - The case for going for a walk
55:07 - Final thoughts

Steve Armenti:
LinkedIn — https://www.linkedin.com/in/stephenarmenti/

twelfth agency:
https://twelfth.agency/

Howdy:
https://heyhowdy.ai

───

sounds good. — a podcast with Anna Nadeina
https://soundsgoodpod.com

sounds good. is sponsored by saas.group — serial acquirer of profitable SaaS businesses.
If you're thinking about an exit, they're worth a conversation.
https://saas.group/

AI has made it easier than ever to build a product.

But has it made it any easier to build a company?

Elizabeth McCalley, Founder & CEO of StartStak.ai, thinks we're confusing the two.

In this episode, we get into what founders are missing when speed becomes the default. You can vibe code an MVP, launch faster than ever and automate huge parts of the operation — but you still need to know what company you're actually building, who it's for, what problem it solves, how it makes money, how it operates and why it should continue to exist when someone else can recreate the technology.

Timestamps:

00:00 — What StartStak is building
01:19 — Lean Startup vs building an actual company
04:33 — Does business planning still matter in the AI era?
05:59 — Go fast and break things — but what are you actually building?
07:00 — Building a product isn't building a company
08:25 — What's missing from most founder business plans?
09:42 — Can the one-person AI company actually exist?
11:07 — Where early-stage founders usually struggle
13:33 — The rise of “pufferfish companies”
15:00 — Is there a perfect founding team?
16:14 — The most common problems founders bring to StartStak
16:32 — Why marketing problems are often product problems
19:24 — How do you diagnose what's actually wrong?
20:00 — A $9M company that didn't know who its customer was
21:48 — Product-market fit is a continuous process
24:01 — How AI changes company building
25:33 — Where does your company sit in the AI stack?
27:58 — SaaS, AI agents and outcome-based services
29:08 — Niching down in AI
31:10 — What if you know your AI startup will eventually die?
33:30 — Building in the blind spots of larger companies
34:37 — What should AI replace — and what still needs people?
36:44 — What happens to accelerators in the AI era?
39:12 — The 99% of founders existing programs don't serve
41:34 — What can founders actually automate?
42:12 — People do the work. AI is part of the process.
44:37 — Building an AI-native company
45:02 — Why are we still struggling with AI adoption?
45:58 — Start with what you're already good at
48:17 — The right AI model for the right job
49:16 — Is there anything exciting left that isn't AI?
49:47 — AI hasn't made us work less
52:28 — Why sometimes you have to go slow to go fast


Elizabeth McCalley:
https://www.linkedin.com/in/emccalley/

StartStak.ai:
https://startstak.ai/


sounds good. — a podcast with Anna Nadeina

https://soundsgoodpod.com

sounds good. is sponsored by saas.group — serial acquirer of profitable SaaS businesses.

If you're thinking about an exit, they're worth a conversation.

https://saas.group/

A year ago, Jason Zigelbaum was running a SaaS business at roughly $1M ARR completely alone.

A year later, he's approaching $2M — and he's still alone.

So, how?

Jason is one of the clearest real-world examples of what running an extremely lean SaaS with AI actually looks like. He uses AI to build product, analyze customer behavior, understand churn, identify his ICP, create marketing campaigns, build landing pages, analyze ad performance, and automate much of the work that would traditionally require a team.

But this isn't a story about handing your company over to a bunch of autonomous agents.

Jason is still very much in the driver's seat.

We talk about how he structures AI workflows, why domain expertise makes AI dramatically more useful, and why the best tasks to delegate are often the ones you already know how to do — but simply don't have enough time to do yourself.

We also get into one of the biggest questions facing SaaS founders right now: if anyone can build your product in a weekend, what makes your company defensible?

For Jason, the answer is depth. Knowing your customer incredibly well, listening to their weirdly specific problems, and building solutions that someone starting from a blank prompt would never know to ask for.

And despite automating almost everything, there's one thing Jason still refuses to hand over to AI: customer support.

Because when you're the person answering every frustrated email, your customers are effectively writing your roadmap for you.

Finally, we talk about the strange side effect of all this new leverage. AI was supposed to give us our time back. Instead, Jason is working harder, doing more, and moving the goalposts again.

The next target?

$4M ARR. Still solo.

Timestamps:

00:00 — From $1M to nearly $2M ARR — still solo
02:04 — What actually drove the growth?
03:23 — What Jason automated with AI
05:16 — Using customer data to find your real ICP
06:06 — Where should AI stop?
08:08 — Jason's actual AI stack
10:34 — Is this finally the one-person AI company?
11:29 — What happens when AI stops working?
13:07 — Why domain expertise makes AI better
14:35 — Learning how to actually talk to AI
16:28 — Can AI replace expertise?
17:48 — Using AI for marketing without creating slop
20:11 — Does building a founder brand actually drive growth?
20:50 — Google vs Shopify vs LLMs
22:20 — The one thing Jason refuses to automate
24:11 — “Anyone could build this in a weekend”
25:08 — How to make a SaaS defensible in the AI era
27:56 — What happens if AI changes e-commerce completely?
31:39 — Where should a bootstrapped founder spend the money?
32:26 — The surprisingly hard problem of cash flow
35:25 — Letting AI manage ad campaigns
37:41 — Everything changed. Everything stayed the same.
38:24 — Why Jason is finally building a personal brand
40:07 — Automating your own content strategy
42:32 — Why AI gets harder when teams get involved
44:17 — What actually performs on LinkedIn?
47:21 — What Jason is excited about outside AI
49:13 — Is AI actually giving us more free time?
50:43 — The goal: $4M ARR
51:31 — See you next year

Jason - https://www.linkedin.com/in/jason-zigelbaum/
Zigpoll - https://www.zigpoll.com/

sounds good. — a podcast with Anna Nadeina

https://soundsgoodpod.com

sounds good. is sponsored by saas.group — serial acquirer of profitable SaaS businesses.

If you're thinking about an exit, they're worth a conversation.

https://saas.group/

Is SaaS dead?

Dirk Sahlmer spends his days looking at SaaS businesses through the eyes of buyers — and his answer is no. But the things that make a SaaS company valuable are changing fast.

In this episode, we get into what the SaaS M&A market actually looks like in 2026: which businesses buyers still want, what kind of multiples founders can realistically expect, why vertical SaaS is getting more attention, and why a business growing hundreds of percent a year can still be difficult for an acquirer to underwrite.

We talk about AI-native companies with incredible top-line growth but brutal churn, whether buyers actually care how much AI you're using internally, and why the dream of running a multi-million-dollar SaaS with three people and a small army of agents is still much more common on LinkedIn than behind the scenes.

Then there's vibe coding. If you can build a product in a weekend, can you sell it? Dirk explains why a little revenue isn't necessarily enough to make a vibe-coded product attractive to an acquirer — and why distribution, brand, customers and market access are becoming increasingly valuable as the underlying technology gets easier to replicate.

We also talk about the increasingly weird world of LinkedIn, why Dirk gets fewer impressions but more actual leads, and why reach might be one of the least interesting metrics for a niche expert.

Finally, we get into something founders often think about far too late: preparing for an exit. Increasing growth and profitability isn't the whole story. Customer concentration, geography, pricing, platform dependencies, hiring and product decisions can all influence what a buyer sees — which is why Dirk would rather start talking to a founder one or two years before they want to sell.



Timestamps:

00:00 — Is SaaS actually dead?

03:19 — What makes a SaaS business durable now?

06:41 — What happened to the “prompt engineer”?

08:08 — What are SaaS companies actually worth?

11:16 — SaaS multiples in 2026

13:24 — Can you really run a SaaS with three people and AI agents?

16:40 — AI growth vs sustainable growth

18:51 — Why great growth can't hide terrible churn

20:05 — Can you sell a vibe-coded SaaS?

21:39 — Is distribution becoming more valuable than product?

23:02 — Why strategic buyers increasingly care about distribution

24:54 — What happened to LinkedIn?

31:16 — Why reach is a vanity metric

32:49 — When personal posts outperform useful content

34:33 — What founders should ask an M&A advisor

35:08 — Why you should prepare for an exit years in advance

38:00 — What actually makes your company more attractive to buyers

40:42 — Don't let the AI bubble drive you insane


Dirk - https://www.linkedin.com/in/dirksahlmer/
FE - https://www.feinternational.com/


sounds good. — a podcast with Anna Nadeina

https://soundsgoodpod.com

sounds good. is sponsored by saas.group — serial acquirer of profitable SaaS businesses.

If you're thinking about an exit, they're worth a conversation.

https://saas.group/

What happens after you sell your SaaS?

Peter Leonard founded MyWorks, built it into a successful SaaS business, and eventually joined saas.group through an acquisition. The original plan was simple: sell the company and move on.

But Peter stayed.

In this episode, we talk about what happens when a founder moves from building and running their own SaaS to becoming part of a larger group — and how that changes your relationship with the company, the product, and your role as CEO.

We also get into what Peter is building next with Stockroom, a new product focused on B2B purchase orders and inventory management for Shopify. As Shopify continues pushing further into higher-volume B2B commerce, there are still plenty of operational problems that existing software doesn't solve particularly well.

From purchase orders and inventory workflows to the reality of competing with spreadsheets and manual processes, Peter explains where he sees the opportunities — and why “boring” software can still be incredibly valuable.

And then there's AI.

We talk about what AI actually changes for established SaaS products, where it creates leverage for small teams, and where it doesn't magically eliminate the need for deep product knowledge, reliable infrastructure, and software that businesses genuinely depend on.

Peter also shares what has changed for him personally since the acquisition: moving from founder to CEO, growing into a different kind of leadership role, and figuring out what it means to build the next chapter rather than simply starting another company.

Timestamps:

00:53 Introduction and Journey to Leadership
03:55 Navigating Change in a New Structure
06:40 The Role of AI in Business Growth
10:01 Building Relationships and Product Focus
12:43 Adopting AI: Challenges and Strategies
15:49 AI in Product Development and User Experience
19:03 Introducing Stock Room: A New Venture
21:59 Partnerships and the Future of E-commerce
26:12 The Role of Expertise in SaaS Development
30:42 The Future of Retail and AI Integration
35:46 Building Partnerships for Success
40:37 Finding Balance Beyond AI


Peter Leonard - https://www.linkedin.com/in/peter-leonard-tx/
MyWorks - https://myworks.software/

sounds good. — a podcast with Anna Nadeina
https://soundsgoodpod.com

sounds good. is sponsored by saas.group — serial acquirer of profitable SaaS businesses.

If you're thinking about an exit, they're worth a conversation.

https://saas.group/

B2B marketing doesn't have to be boring.

Renée Shaw from tl;dv has turned humor, personality, and relentless experimentation into one of the company's most recognizable growth channels. Instead of treating social content as something marketing teams have to “do,” tl;dv built a culture where creators are given the freedom to experiment, fail, and figure out what makes people pay attention.

We talked about why a B2B company decided to put creators and comedy at the center of its marketing, how to measure something as messy as brand attention, and why trying to attribute every view to revenue can actually make your content worse.

Renée breaks down how their team comes up with ideas, why TikTok, LinkedIn, and Instagram require different instincts, how trends become original skits, and why volume matters when you're still figuring out what works.

We also get into the less obvious side of creator-led marketing: how to hire people who can actually make funny content, why they don't necessarily need to understand your industry, how a distributed team maintains the same voice without a giant tone-of-voice document, and why psychological safety and trust are essential for creative work.

And then there's the AI question. Renée explains why she's skeptical of AI-generated creative at scale, why she thinks human-made work is becoming more interesting, and why sometimes the best creative strategy is simply to stop optimizing everything.

Plus: the mistakes that taught her the most, the dangers of over-engineering content, creative burnout, why your ideas aren't as precious as you think, and what happens when you stop performing for an imaginary audience.

Timestamps:

00:00 — How Renée became “CEO” of tl;dv
03:33 — Why tl;dv hired creators before it was cool
04:58 — Taking a risk on unconventional B2B marketing
07:00 — Can you actually measure brand-driven growth?
09:48 — LinkedIn vs TikTok vs Instagram
12:50 — Why TikTok became a place to experiment
14:28 — Where do viral content ideas actually come from?
17:21 — The philosophy behind tl;dv's content
19:32 — How a remote team maintains culture
21:43 — Trust, ownership and psychological safety
24:09 — How to hire a creator for your company
25:57 — Why you might find your next creator in comedy
27:24 — Do creators need to understand your product?
27:29 — Would you hire an AI influencer?
30:33 — The actual process behind making a video
33:20 — Why not every video needs a product plug
38:38 — Culture, optimism and surviving competition
42:01 — How to avoid creative burnout
44:46 — The content experiments that completely flopped
47:07 — Why trying to be someone else doesn't work
48:32 — What exciting things are happening outside AI?
50:46 — Stop optimizing everything
52:40 — Why old sitcoms are surprisingly good for your brain
54:43 — Why human creativity might matter more than ever

Renée - https://www.linkedin.com/in/reneeeshaw/
tl;dv - https://tldv.io/

sounds good. — a podcast with Anna Nadeina

https://soundsgoodpod.com

anna@soundsgoodpod.com

sounds good. is sponsored by saas.group — serial acquirer of profitable SaaS businesses.

If you're thinking about an exit, they're worth a conversation.

https://saas.group/

AI has made building software dramatically faster—but it hasn't made building the right software any easier.

Juan Germano, founder and CEO of Jams, builds custom web and mobile products with founders and operators. And after years of building software, he's seeing a fundamental shift in how products get designed, built, and launched.

We talked about what happens when founders can turn an idea into a working prototype in hours, why AI-generated roadmaps can be dangerously convincing, and why having unlimited building capacity can actually make products worse.

Juan also explains why the fundamentals of product development haven't changed: you still need to understand the problem, make the right trade-offs, launch, get feedback, and figure out how you're going to get your first users.

We get into why distribution needs to be designed into a product from day one, what AI means for software agencies and developers, why small senior teams may have an even bigger advantage now, and why Argentina could have a lot of SaaS opportunity hiding in plain sight.

And we talk about something that gets surprisingly little attention in the AI conversation: why human expertise, real-world experiences, communities, and authentic content may become more valuable—not less—as AI-generated everything takes over the internet.

One of the biggest takeaways: AI gives you more building capacity. It doesn't give you better judgment.


Timestamps:

00:00 - Is software development becoming a dying craft?
01:41 - Why product strategy matters more than ever
02:52 - Are we building for humans or AI agents?
04:46 - The rise of MCPs and APIs
07:02 - Why AI prototypes can be dangerous
07:17 - A prototype isn't a product
09:14 - What is the actual "product"?
11:29 - When AI creates the product roadmap
13:54 - Why founders still need product expertise
16:21 - Why six weeks still matters
17:13 - AI lets teams ship 10x more iterations
18:43 - More output doesn't mean better products
21:28 - The danger of unlimited building capacity
24:03 - The one-sitting approach to building
27:12 - Why small teams can move incredibly fast
30:31 - The new distribution problem
31:19 - The first question Juan asks every founder
33:41 - Designing distribution into the product
37:25 - What's happening in Argentina's SaaS market?
39:40 - The opportunity in legacy software
42:15 - What happens to WordPress in five years?
46:16 - How should founders work with software experts now?
47:37 - Using AI to turn ideas into tangible products
53:05 - How to use AI to write better prompts
54:47 - Juan's AI workflow
57:13 - AI is teaching us how bad we are at communicating
59:19 - What's exciting that isn't AI?
01:00:08 - Why timeless ideas still matter
01:02:33 - Escaping AI FOMO
01:03:40 - Will AI make communities more valuable?
01:05:59 - Are we actually going to work less?
01:08:14 - Why humans will still want human things
01:10:05 - Why people hate AI-generated art
01:11:09 - The value of authentic work

───

Juan Germano — https://www.linkedin.com/in/thejuangermano/

Jams — Custom web & mobile products for founders and operators

sounds good. — a podcast with Anna Nadeina

https://soundsgoodpod.com

sounds good. is sponsored by saas.group — serial acquirer of profitable SaaS businesses.

If you're thinking about an exit, they're worth a conversation.

https://saas.group/