Scaling Sales Teams in the Age of AI with Dini Mehta

Dini Mehta has spent the last two years advising GTM leaders at AI-native startups. Here's what's actually changed about building a sales team in the AI era—and what hasn't.

For five years as a revenue leader at Lattice, Dini Mehta sat in board meetings where the message was always some version of “go faster.”

The solution back then was always the same: hire more reps.

Dini ran the hire-more-reps playbook about as well as it could be run. She joined Lattice as VP of Sales at $3M in ARR and left as CRO with revenue past $100M and a team twenty times the size, at 250+ people. She was then fractional CRO at HeyGen through its run from $5M to $40M.

She spent the last two years on the other side of those board meetings. As Operator in Residence at Peak XV Partners from July 2024 to August 2026, she advised founders on scaling go-to-market—and frequently caught herself delivering the same message she used to receive.

“I find myself saying, ‘You can go faster! What are you doing!?’” she says. “Then I remind myself that it's so much easier to be saying this from the other side.”

But unlike her time at Lattice, “hire more reps” isn't the default playbook anymore. Across the AI-native companies she's worked with—Wispr Flow, Nooks, Relevance AI, Checkbox—Dini says she's consistently seen headcount treated as the last lever a founder pulls rather than the first.

Dock CEO Alex Kracov ran marketing at Lattice while Dini ran revenue, so he wanted to catch up and compare notes on what's actually different about building a go-to-market team in the age of AI—and what's the same.

Growth Bytes

Top takeaways:

• Headcount is now the last GTM lever founders pull, not the first. AI has absorbed the research, enrichment, and ops work that used to require hiring, so teams stay smaller and the bar for talent rises.
• Reset your ARR growth baseline, but don't benchmark against outliers. What used to be a great year—$2M to $10M—is now $2M to $20M. But you can't benchmark against Cursor and Lovable without knowing the economics of a business.
• Expect to build a sales org even if you start product-led. Individual productivity tools spread bottom-up. Anything that changes how a group works brings governance, access, and security concerns—which is why the fastest-growing AI companies keep layering on SaaS-era sales motions.
• Reps should use AI time savings in front of customers. AI can handle research, enrichment, call summaries, and CRM hygiene. Trust-building, discovery, and navigating the buying group should stay with the rep.
• Audit where seller time actually goes before you buy AI tools. Map the workflows bogging sellers down in admin work, automate against that list, then put your head down for six months before deciding on phase two. Otherwise you're in permanent shopping mode.
• Hire for learning velocity, systems thinking, and comfort with agents. Volume-based prospecting is machine work now, so what's left of the job needs account strategy and the ability to navigate an org.
• Teach new reps how to learn, not what to know. Product and messaging change too quickly to justify heavy ramp cycles. Spend onboarding time on how to find internal knowledge and research with AI.
• Spend the manager time AI frees up on people, not more pipeline review. Most reps leave because they stopped learning or don't like their manager, so career growth is your biggest retention lever.

Go-to-market is a systems problem now

Dini says lots about running a GTM org hasn't changed—understanding a customer's business, building trust, running a real sales process, finding where the organizational chaos lives, proving value inside the product.

When it comes to scaling, Dini says founders used to start with headcount. Now they start with systems.

“Go-to-market is increasingly being treated as a design and systems problem versus a people problem. In the past it was like, ‘Oh my God, hire a bunch of reps, go get revenue.’ Now AI can reduce a lot of the cost around research and enablement and operational work, and teams are thinking about where they can leverage AI and make adding people the last lever in the business.” — Dini Mehta

Dini says founders aren't necessarily avoiding making sales hires. “Less delaying hires in general, more being super picky and delaying bad hires for sure.”

Meanwhile, year-over-year VC growth expectations keep rising. A jump from $2M to $10M ARR would have been considered a great year in the past, but now the expectation is more like $2M to $20M.

Some of that comes down to the boon around AI—and some of it comes from the availability of GTM tools that we didn't have in the past.

However, Dini doesn't think Cursor and Lovable's growth curves should be the new benchmark—they're category-defining outliers.

More of those outliers exist in this era because the whole technology market is shifting, but expecting that growth rate from a “normal” company, without knowing the mechanics and economics underneath it, is how founders end up making bad go-to-market decisions.

Winning AI revenue is easier than keeping it

Dini also cautions that the top-line growth numbers everyone is citing haven't been through a renewal cycle yet.

“In this new world with the AI tailwinds, getting revenue is almost easier than keeping that revenue, because you've got experimental budgets and people are very excited to test and run with it.” — Dini Mehta

Budgets opened fast for anything attached to AI; buyers were eager to run pilots, so there was a lot of spend to go around—but that experimentation phase won't last forever. Now, buyers want accountability.

Which makes the second (or renewal) year of these contracts the real test, not the first.

AI companies still need traditional sales teams

Dini said that every AI company scaling through bottom-up adoption is still building a traditional sales org on top of it.

“Literally everyone, from the labs to all the companies that we know of that are scaling really fast, is building the exact same sales orgs that we know from the SaaS era.” — Dini Mehta

Any tool that makes one person better at their job spreads bottom-up—and that describes most of the current wave of AI tools. Anything that changes how a group of people work together arrives with governance concerns, access control, and data security attached—and those deals require a sales process.

So the question isn't whether you build a sales team. It's when, and in what order.

If the goal is a billion in revenue as quickly as possible, her answer is product-led at the entry point with sales-led layered on top. That's the call she made at Lattice, which ran a product-led motion early before committing fully to sales-led (a call she said she'd make again today).

Alex added that the rise of the forward-deployed engineer—an engineer who sits inside the customer's org, building alongside them—is the least “product-led” move imaginable, and yet it's working for many revenue teams.

AI is eating sales tasks, not sales jobs

Dini argues that AI hasn't changed the underlying sales job.

“There are a lot of tasks in sales, but the job is only one, which is that you're the human interface for representing your business and helping customers buy. The job hasn't changed.” — Dini Mehta

AI can gobble up tasks like research, data enrichment, call summaries, CRM hygiene, and the context layer running underneath the whole go-to-market org.

But reps are still responsible for building trust, running discovery, mapping stakeholders, multithreading early enough to matter, and change management.

That's necessary because the customer on the other end has changed too. Buyers arrive having researched the category, shortlisted vendors, and pressure-tested your pitch against a model before even taking your call.

“Everybody's got information. It's no longer about giving them the most up-to-date information about your company. Because if you're doing that as a seller, you're failing.” — Dini Mehta

Today, a seller's main job is to facilitate decisions. The buyer still has to make a call, which usually includes convincing a group of people who don't agree.

Dini says that's where sellers can make the buyer's life easier. “The number one thing we sell is the fact that we care. And people have to feel that care in every touchpoint.”

Build an AI strategy before you buy AI tools

Dini says that compared to product teams, go-to-market is still in the early innings of AI adoption. Overall, she's seeing a lot of reactivity—teams buying AI tools based on whatever LinkedIn post the founder forwarded that morning.

Healthy AI adoption should look more like strategic experimentation. Map the workflows bogging sellers down in admin work. Pick tools against that list rather than the other way around. Get buy-in across the org. Then put your head down for six months before experimenting again.

Some companies are standing up a centralized AI function—go-to-market engineering, AIOps, RevOps—to build workflows and agents for everyone. Others are pushing agent-building down to individual reps. Dini prefers the centralized model.

“My worry with getting every person to build their own agents is you're turning into AI theater of, ‘Here's an agent I created.’ Should you be doing that with your time? It's awesome that they're excited. But I think it's a delicate balance.” — Dini Mehta

But Alex Kracov, CEO of Dock, thinks there's value in reps building their own AI workflows too. For example, Christian Corbin, an AE at Dock, has built workflows for himself in Claude that no central roadmap would have produced.

Alex said he's seen both situations play out with Dock customers. The person driving GTM AI adoption at a prospect company might sit in enablement, in ops, in a sales leadership seat, or just be a rep with strong opinions—there typically isn't a single title.

On the other hand, some organizations have set up entire AI buying committees.

“We've seen a lot of our deals go to this AI committee at a company, where we get vendor of choice, we're in procurement, and then there's this random AI committee that's going to say yes or no. Which makes sense—they want it to fit into everything. But it's a new type of discovery question we've got to ask.” — Alex Kracov, CEO of Dock

What is the ideal hiring profile for a sales rep?

If the volume-based SDR role is gone for good, how do we develop new sales reps?

SDR to AE to manager was how the industry manufactured sales leaders for two decades, and Dini's uncertain what replaces it.

Her best guess is a return to full-cycle AEs. When she came up in sales, there was no SDR role—you cold called, found your own deals, closed them, and handled renewals after the sale.

Though she's noticed a counter-signal: the AI labs are hiring lots of SDRs and BDRs right now.

So what should you look for when hiring a sales rep in the AI era?

Dini says classic traits like curiosity, adaptability, and cultural fit are still relevant. What's new is systems thinking, genuine comfort working alongside agents, and learning velocity—because products change faster now than anything she dealt with at Lattice.

“The new rep is part consultant, part operator that's comfortable with tools, part product expert, and has to be AI-pilled. It's hard to find all those pieces, but you can create an organization where you find people with the attributes to learn those things.” — Dini Mehta

Teach reps how to learn, not what to know

Onboarding used to cover four things: product, process, messaging, and people.

Dini thinks that product knowledge is the fastest-depreciating asset in your onboarding program, so heavy product training is close to wasted effort (and that it should be automated through content).

What she'd teach instead: how to research, how to prompt, how to reach the internal knowledge system, and what to automate. Reps need to learn how to learn fast, and she'd make that the explicit output of onboarding rather than a byproduct.

“The map of the organization, which includes agents, tools and people, is, I think, where enablement and onboarding should really be focused on.” — Dini Mehta

Dini believes the manager job will be similarly reshaped. Forecasting, activity review, call commenting, and coaching recommendations are all getting automated.

What can't be handed off is psyche management. The volume of rejection, combined with quarterly quota pressure, makes managing sales unlike other functions. Being a great manager is about keeping reps confident and invested through bad stretches.

Culture matters more on smaller teams

Leaner sales teams make the culture work harder, not optional, says Dini.

“Otherwise you're going to turn into a mercenary transactional org that is just focused on ‘How does this help me today? And if it doesn't help me tomorrow, I'm out.’” — Dini Mehta

She broke what makes a great sales culture down into three pillars:

1. Growth. Dini says most people leave a company because they either stop learning or don't like their manager—so growth is the biggest rep retention lever you can pull. She carefully separated career growth from promotion. Real growth means getting measurably better at what a person actually wants to improve, which requires knowing who they are and what path they're on.
2. Community. Whether people feel like they're on one team worth being around, given how stressful and fast-moving the work is.
3. Mission. Connection to the mission, or at least to a purpose that justifies the long hours.

Her bet is that AI clears enough off managers' calendars to make room for all three.

—

Dini's overall framing around AI in sales is that efficiency should mean more impact per person, not more work per person.

She believes we should strip out the parts of the job nobody likes, and hand the time back as room to think rather than as nineteen new tasks.

“Ultimately, humans want the same things out of work. Nobody wants to just come in and be a machine, because you've got AI for that.” — Dini Mehta

Watch the full episode

Watch Alex and Dini's full conversation on Grow & Tell, Dock's podcast for revenue leaders.