Your Buyer Isn’t Human Anymore. Neither Is Your Seller. Welcome to AI-Led Growth. With Firebolt’s President Hemanth Vedagarbha

We just had a great session at the last SaaStr AI Annual from Firebolt on something I think every B2B SaaS founder needs to internalize right now.

The punchline: we’re entering a third go-to-market era — and most of you aren’t ready for it.

The Data Is Brutal Right Now

Let’s start with reality. SaaS growth has been cut roughly in half:

Median SaaS growth has dropped from 36% in 2021 to ~16.5% today.
Customer acquisition costs are up 55% over the last five years. Not inflation. Structural.
Customer churn is now a $136 billion drag on the SaaS industry — roughly 7-8 points across the entire market.
Public SaaS valuations have gone from ~17x to 6.7x. Private multiples tell the same story — down from 20x+ to 6-7x.

So you’ve got a triple squeeze: it costs more to acquire customers, they’re harder to keep, and your company is worth less even when you do grow.

That’s the world we’re operating in. And the old playbooks aren’t going to get you out of it.

For 25 Years, There Were Two GTM Motions. Now There Are Three.

For the last two decades, B2B SaaS has been dominated by two go-to-market models:

Sales-led growth. Expensive. People-heavy. Enterprise AEs, inside sales teams, SDR armies. It works, but it’s a high-burn model that gets more expensive every year as CAC climbs.

Product-led growth. Optimized for users. Freemium, free credits, land-and-expand. Better unit economics at the top of funnel, but still requires humans to close bigger deals and manage expansion.

Most companies run some hybrid of the two. That’s been the playbook.

But the next era of growth isn’t coming from either of these. It’s coming from AI-led growth — and it’s fundamentally different from both.

What AI-Led Growth Actually Looks Like

This isn’t theoretical. Companies are already operating this way.

Your top of funnel is now algorithms, not people. When a prospect evaluates your product today, the first “buyer” isn’t a CMO or a VP of Engineering scrolling your website. It’s an AI agent — ChatGPT, Claude, Gemini — doing research, comparing options, summarizing capabilities. Your pitch isn’t being consumed by a human with feelings and brand loyalty. It’s being parsed by a model looking for structured, relevant information.

If your marketing isn’t optimized for AI consumption, you’re invisible in the new top of funnel.

Your mid-funnel is agent-to-agent. Here’s where it gets really interesting. Firebolt described a world where buyer AI agents and seller AI agents are interacting with each other — evaluating options, negotiating terms, building proposals. Humans don’t enter the process until very late. If you’re still running SDRs doing cold outreach to set appointments for AEs, you’re playing a game that’s already being automated out from under you.

Events are going fully agentic. Some companies are running end-to-end events — from invite generation to email outreach to content creation to the actual presentation — with zero human involvement. Not many yet, but it’s happening. The Firebolt presenter even tried to have an agent build and deliver his SaaStr talk (SASTR said no, but the point stands).

Your workforce is agents + humans, not humans + tools. Firebolt shared their own three-year plan: they have a couple hundred employees today and plan to scale to 1,000+ — but the majority of those “employees” will be agents, not people. Their framing: if you have 200 humans and 800 agents, the question isn’t “how do you manage your team?” It’s “how do you extract maximum value from your agents?”

That reframes every role in the company. As the Firebolt team put it: the CISO becomes the new Chief People Officer, because they’re the ones governing what the agents can and can’t do. The traditional CPO role shrinks.

AI-Led Customer Success Is Where the Real ROI Is

Acquiring customers is one thing. But the churn problem is where AI-led growth might have the biggest immediate impact.

Firebolt shared their own setup: one human CS rep for the US, one for the rest of the world. Everyone else is agents. They’ve replaced the bulk of their customer success function with agentic AI driving in-product experiences.

Three specific AI-powered approaches worth noting:

Predictive churn models that start at Day Zero. The old model: start renewal conversations a month or a week before the contract expires. The new model: your renewal process starts the day the customer signs. For a three-year deal, churn prediction begins on day one, not month 35. AI-powered predictive models can reduce churn by ~22% when deployed this way.

AI-generated QBRs with 80% less prep. Customer success reps used to spend hours building quarterly business review decks — pulling usage data, summarizing value delivered, prepping recommendations. Now agents generate most of it. The CS rep’s job shifts from data assembly to relationship management and strategic conversation.

Next-best-action engines embedded in-product. Think Netflix-style personalization applied to your B2B product. The product itself guides customers to the next feature, the next use case, the next expansion opportunity. Instead of relying on a CSM to proactively check in and say “hey, have you tried X?” — the product does it automatically, continuously, at scale.

The Practical Churn Playbook

When asked for concrete examples, the Firebolt team laid out a framework worth stealing:

First, be honest about your churn. Most companies obscure it. They talk about “90% retention” when they mean logo retention, ignoring dollar churn. Or they blend annual numbers to hide quarterly spikes. You want to be at 95%+ on both logo retention AND dollar retention. If you’re not there, admit it and fix it.
Second, push for multi-year deals. More skin in the game from the customer. More runway for your agents and CS team to drive adoption and expansion. A one-year deal gives you almost no time to build deep in-product engagement before you’re already in renewal mode.
Third, invest in in-product guidance. Your product should hand-hold customers through onboarding, training, and ongoing adoption the way Amazon Prime or Netflix does. Don’t wait for a CSM to notice a customer is struggling. Build the intelligence into the product itself.
Fourth, blend AI-led with human-led. This isn’t about replacing humans entirely. It’s about having agents handle the routine — health checks, usage reports, renewal reminders, next-best-action prompts — so your human CS team can focus on the strategic, high-touch moments that actually require empathy and judgment. The combination of AI-led + sales-led + product-led is the winning formula.

The Provocative Question

The Firebolt presenter ended with a thought experiment: at SaaStr 2030, will there even be humans on stage? Will agents be building the presentations, delivering the talks, and organizing the event? Will the audience be agents too?

It’s half joke, half genuine prediction. But the underlying point is dead serious: if you’re still thinking of agents as “a piece of code” or “an algorithm,” you’re already behind. The companies pulling ahead are thinking of agents as employees — with roles, responsibilities, KPIs, and governance.

The shift from sales-led and product-led to AI-led growth isn’t coming. It’s here. The question isn’t whether to adopt it. It’s how fast you can move.

Every point of churn you eliminate compounds. Every dollar of CAC you cut drops straight to the bottom line. And in a world where valuations are 6-7x instead of 20x, efficient growth isn’t optional — it’s existential.

Start your renewal process on Day Zero. Optimize your pitch for AI buyers, not human ones. And start treating your agents like the workforce they’re becoming.