E42: SaaS is changing with Yash Chavan

‍

Five Months, Five Product Surfaces

‍

Five months ago, Rishabh's company was a single product business selling funnels generated against paid ads. Primarily mid-market customers.

‍

Five months later? Selling to large enterprises. A platform based on consumer behavior. Five different product surfaces live with customers.

‍

The go-to-market shifted. The number of products shifted. The platform shifted. All in five months.

‍

Then you look at Twitter and Claude Code exploded over the holidays. OpenAI shipping at a new pace. The whole market confirmed what Rishabh was already living.

‍

The speed with which you can do things has changed fundamentally.

‍

Your Team Needs An Athletic Stance

‍

Teams used to operate on a time scale based on quarters. Annual planning. Goals set quarterly. As a salesperson or a marketer, you would wait for the product, then go sell it. You could think about a product, deploy it, and sell it for three years.

‍

That time scale is compressing.

‍

A year is too long. A week is too short. It is somewhere in the middle. And it looks very different than it used to.

‍

The way Rishabh thinks about it is an athletic stance. Keep up with the market at the current pace. So when someone interacts with you about a specific topic, you are informed enough to respond in a useful way.

‍

That is a completely different ask than "here is your task for the quarter."

‍

It is about readiness. About being in motion.

‍

EOS Is Being Stress Tested

‍

SARAL runs on EOS. OKRs, L10 meetings, bi-weekly check-ins, standards reports. The whole playbook, taken seriously.

‍

And for the last year, that system has been getting stretched.

‍

You set a goal. The market shifts. Someone launches something faster. The quarterly cadence gets outpaced.

‍

The question is what replaces it. Maybe a few org-level goals and everybody runs toward them. Maybe something new entirely.

‍

The AI world EOS playbook is still being written.

‍

Buy The Agentic Platform

‍

Here is the new buying criteria.

‍

Rishabh's company churned off Gong, Fivetran, and ZoomInfo. The vendors they are growing with are the ones building agentic platforms.

‍

When the head of finance asked which applicant tracking system to buy, the answer was simple: buy the one that is an agentic platform. Is there an agent? Is it getting better? Are they investing in it?

‍

That was the entire buying criteria.

‍

The question of whether this translates beyond tech is really a question of time scale. Maybe e-comm brand owners will buy this way in 2030. So all we are debating is when.

‍

And you would rather be early than late.

‍

Manual Work Gets The Conversation

‍

At Rishabh's company, if someone on the marketing team does something manually, it becomes a conversation.

‍

The focus is on the agentic workflow that was implemented to create the thing. The system matters, because the system compounds.

‍

The production rate changes. The output rate changes. Everything changes.

‍

That learning flywheel is the thing worth optimizing for. A team that builds agentic workflows is on a fundamentally different trajectory than a team that does things once by hand.

‍

The Leading Edge Of The Normal Distribution

‍

The default output from LLMs is verbose. Most people skim. Ask the next question. Move on.

‍

The better frame is to think of an agent's response as a normal distribution. There is a leading edge that is great. There is a middle that is fine. There is a bottom that you skip.

‍

Only the leading edge matters. The top 10%. The top 1%. That is where the value lives.

‍

And you can tell the AI to give you information however you want to receive it.

‍

Here is a concrete example. ChatGPT integrated with Slack in the product releases channel. Engineers and PMs write detailed specs. The prompt: summarize the last week like I am a go-to-market salesperson, give me the top three points about what we released, what it can do, and how it impacts customers.

‍

The quality is great. It summarizes correctly.

‍

The key is giving it a specific prompt with a specific point of view.

‍

Same way you tell a human teammate how to communicate with you. Send it in email. Keep it to three things. Same instructions for humans and agents.

‍

The Best Equipment Makes The Best Athlete

‍

I showed the group a video Zach put together recently. Stunning creative work. The kind of thing that makes you pause.

‍

The question came up: would Zach be a top 0.1% designer if he had been reliant on AI tools from year one?

‍

100% yes. Having the best equipment makes you the best athlete.

‍

You have to compare the 1% to the 1%. The top end of AI-assisted work against the top end of manual work. The ceiling keeps going up.

‍

The RunwayML bear video that went viral on Twitter is a perfect example. A short film made entirely in RunwayML that looks like an Oscar-nominated piece. That is the 0.01% of AI-assisted creative. And it is incredible.

‍

There are always going to be people producing things at every level of the curve. That is true with the tools and without them.

‍

The leading edge is what matters.

‍

Customers Are Pulling You Into Multi-Product

‍

The traditional SaaS multi-product playbook is you hit 10 or 15 million ARR and launch another product. You decide to build the next thing.

‍

What is happening now is that the customer is asking for the next product. That is a different dynamic.

‍

Customers can see that your production velocity has gone way up. They land with you for one thing. Then they ask: can you do this too? And this? And this?

‍

And you say yes. Because it takes six to eight weeks to ship the next surface. It is engineering implementation work on the same foundation.

‍

ACVs are going way up. Decagon raised at 3x their previous valuation and the only data point they shared was that they signed 100 enterprises. They chose that data point on purpose.

‍

Because investors understand that if you land 100 enterprises and your product velocity is high, every one of those customers becomes a multi-million dollar account over time.

‍

Total reframe of how to think about market opportunity.

‍

Software At Its Peak

‍

There is an interesting way to look at what AI represents to the customer right now.

‍

Software has always been about minimizing inputs and maximizing outputs. Save time. Unlock capacity. Do more with less.

‍

AI is that same idea, at a different level. It is software at its peak. At its simplest, it is point and shoot. Ask ChatGPT what you should have for dinner. At its best, it races ahead of you.

‍

And it represents the cutting edge. Early adopting. The bleeding edge of what is possible.

‍

AI is software. A different kind of software. And to the consumer, it represents the ceiling.

‍

The Parker Launch

‍

Alex Cooper launched Parker, an AI creative director that lives in Slack. 1.4 million views on the launch video in three days. Every big AI page on X was posting it.

‍

We signed up right away. It connects to your ad account, knows your spend, and gives specific recommendations based on your data and customer reviews.

‍

Alex has been on top of AI from the very beginning. Back when ChatGPT could only give you text, he was using it for creator scripts. He has pushed himself and his organization to use AI at its maximum capability at every stage of maturity over the last three years.

‍

He deserves every success in the world because he has tried to push it as much as he could at every step of the way. More people should be doing that.

‍

Saral's Refueling Year

‍

SARAL is an influencer operating system. Brands find influencers, contact them, manage relationships, track, pay, everything you would want to do with a creator program. The software biases for less risk over time the more you use it.

‍

Zero to a million in 14 months. Then the business model completely changed. Self-serve to sales-led. Tripled prices. Went upmarket.

‍

Over the last 12 months the early SMB revenue got replaced with committed mid-market customers. Entirely new business model.

‍

Three years bootstrapped, cash flow positive, going multi-product this year with a Shopify App Store launch.

‍

Shipping faster than ever with AI-assisted engineering.

‍

The name Saral means "simple" in Sanskrit. The space was crowded with complicated tools and the name needed to be clean. Also picked intentionally so it could be anything. Saral could be a DTC brand if needed.

‍

Smart naming. Very meta.

‍

Path Dependence Is Everything

‍

Once you do a thing, it informs the next thing. The path dependence is more important than the statement of the objectives.

‍

It is like the scientific process. You run experiments. You do the next thing, then the next thing. And all of a sudden you discover something.

‍

From a product perspective it works the same way. There is a big picture of where you want to go. But the steps between now and then are really about getting to the next step as quickly as possible.

‍

Agents work this way too. You ask a question, the agent does something, it comes back, and that creates a new set of questions.

‍

There is less planning. And that might actually be a good thing, because part of the value of moving fast is that you start with incomplete information and let the path fill in the gaps.

‍

Take one step. Get to the next thing. Move as quickly as you can.

‍

Hiring Efficiency And The Gig Model

‍

There is an interesting overlap between full-time hires and the project-based gig economy.

‍

With an Uber or an Upwork hire, the employment is extremely efficient. One task, one transaction, done.

‍

Full-time hires have a different structure. You are committing to a suite of tasks on a regular basis. The role evolves. The goalposts move.

‍

AI agents work more like the gig model. Point and shoot. Get this thing done. Infinitely more scalable because AI can work around the clock and handle complexity at speed.

‍

The interesting part is that whoever sets the direction is always the one in charge. Whether it is an employee, a contractor, or an agent.

‍

Building And Replit

‍

Replit improved massively in nine months. $400 in credits in a week. And it is way better than it was.

‍

What changed is that it took out a lot of the planning. You look at the software you have, see an obvious thing to improve, improve it, and get to the next thing.

‍

That mirrors how agents work. Less upfront planning. More iterative progress. More path dependence.

‍

Quarters still make sense as a frame for what you are trying to accomplish. But the exact steps to get there are about taking it one step at a time and moving as fast as possible.

‍

CS Is Where AI Gets Interesting

‍

Some platforms are on the complicated end of the spectrum. There are many things you could do with an influencer platform and many ways to use it. The best analogy is HubSpot. Buying HubSpot means you bought a tool, and there are so many other things that have to go right for the software to be useful.

‍

Compare that to something like Intercom where the use case is so specific that the UI is almost the entire product.

‍

For platforms with a lot of surface area, customer success and onboarding matter enormously. And that is exactly where AI agents are being deployed right now.

‍

Early experiments are already running.

‍

The Pace Of Change Is The Opportunity

‍

Replit went from a dashboard builder to a real app builder in nine months. Rishabh's company went from one product to five in five months. Saral completely rebuilt its business model in a year.

‍

The pace is the opportunity.

‍

The teams in an athletic stance, building agentic workflows that compound, focused on the leading edge of quality, saying yes when customers ask for more, are the ones creating the most value.

‍

The speed at which you can build, ship, and expand has fundamentally changed. And the founders who lean into that pace are the ones pulling ahead.