sarin

automating marketing depts @indeliblegtm | ex @rippling @moveworks @salesforce

[sarin @dudesarin]

attention is all you need

[Jeremiah Prummer @JeremiahPrummer]

If I were starting a SaaS in 2026 I would start it as an agency.

Charge clients thousand per month to solve an acute problem.

Build systems and tech to support the backend.

Slowly phase out human time with automation.

This is effectively how building software has always worked.

The difference is that us software builders used to subsidize all that work for our customers because we believed the long term value of the business would make that worthwhile.

Customers would haggle over $50/month for our software, but wouldn't hesitate to drop $20k/month on an agency retainer.

So I wouldn't mess with that.

I'd go get the service revenue, automate it, and eventually push most of that value back to my customers.

But I wouldn't call it software/SaaS. I'd call it a service.

[sarin @dudesarin]

The marketing dept is morphing in real time - soon we will see orgs consisting of generalists with good taste, judgement, understanding of ICP and product - with specialist AI agents running underneath

[Cristina Cordova @cjc]

Marketing leadership might be the hardest role to fill and the hardest to keep filled. A few reasons I find this to be the case based on my conversations with founders:

1) "Marketing leader" isn't one job. It's brand, demand gen, product marketing (and more). Different muscles, different people. Brand seems to be having a moment right now, probably because a lot of categories have gotten incredibly crowded and there isn't much left to differentiate on besides how you show up.

2) Companies struggle to prioritize. The job description asks for many different disciplines within marketing and no one is great at them all. So, you hire a generalist who's fine at everything, or a specialist, and then quietly resent them for the parts they don't cover.

3) Everyone thinks they can do marketing. Nobody weighs in on the database schema. Everybody weighs in on the landing page. And the scoring is asymmetric: weak adoption is a marketing problem, strong adoption is a great product. Same outcome, different owner depending on which way it broke.

4) At early stage, the founder is the CMO whether anyone says so out loud or not. So hiring well is mostly about figuring out who can work inside that, which has little to do with the resume.

[sarin @dudesarin]

@cjc These roles are already starting to morph and over time will be a blend of generalist with good taste, judgement, understanding of ICP and product - with specialist AI agents running underneath

[Parker Conrad @parkerconrad]

Huge congrats to my cofounder Prasanna on the launch of his new co!

[Prasanna S @myprasanna]

The models are so smart now that they have outgrown your laptop. All the dev is going to happen on the cloud. Get out of the driving seat. Let the agents ping you when they need you. Close your laptop. Answer on mobile. Massively parallel ideation is the new builder job.

Congrats to @cursor and @elonmusk -- this space is wide open now.

We're building a roadster. Launching it next week!

[weber @weberwongwong]

Introducing FLORA v2.

We've rebuilt FLORA from the ground up for a faster, more performant experience.

This sets a strong foundation for Phase 2 of FLORA's evolution.

[sarin @dudesarin]

Ability to save $5B per year if construction RFIs were automated with agents

[sarin @dudesarin]

High pain + value opportunities for AI agents in construction

[sarin @dudesarin]

@LM_Braswell Praying robots/AI don’t surpass us in poker for a while

[sarin @dudesarin]

@saranormous

[sarin @dudesarin]

@jaltma Translates to venture in so many ways - being present, patient, persistent, sociable

[sarin @dudesarin]

@nikunj It’s helpful to hear shortcomings about particular markets and what’s done well or not worked in the past, but I agree that your own personal taste is what separates you from the rest

[sarin @dudesarin]

In construction, there are 8.3M total workers in US with ~2M (25%) doing work AI could potentially automate

- 520K project managers (40% time on documentation)
- 75K estimators (manual takeoffs)
- 50K schedulers (manual resource allocation)
- 180K superintendents (coordination)

$75B annual labor cost across these roles

[sarin @dudesarin]

researching construction industry

construction software exits over the past decade:
- Procore IPO (2021): $11B at 24x revenue
- PlanGrid acquisition (2018): $875M (raised $69M)
- Aconex acquisition (2017): $1.2B at 9.7x revenue
- Viewpoint acquisition (2018): $1.2B at ~6x revenue
- Textura acquisition (2016): $663M at 7.6x revenue

$5B+ in M&A at 6-10x multiples. question is what the AI-native version looks like

[sarin @dudesarin]

SaaS digitized industries.

AI agents will operationalize them.

The companies that get this sequencing right will quietly define the next decade.

[sarin @dudesarin]

Every platform shift rewires how industries run.

Cloud software created the first wave of modern Industry OSs — Epic, Procore, Toast, ServiceNow.

AI agents will create the next wave.

That’s where we're spending our time. ↓

[sarin @dudesarin]

The defensibility isn’t hype or branding.

It’s utilization × complexity.

If it’s used every day and coordinates multiple systems, it’s almost impossible to replace.