Here's how Anthropic uses AI for GTM (GTM AI Summit 25 - full talk)

My name is Ash Al Hashim.

Thank you Tyler for that lovely intro.

I lead one of the sales teams at Anthropic that supports our technology customers.

So my team supports some of the biggest technology companies in the world.

Companies like Netflix Palo Alto Networks, Shopify, Uber, et cetera.

We help them really transform their businesses using AI.

But I'm not here to talk about what we do for our customers today.

I'm gonna talk to you a little bit about how we use Anthropic and AI in general ourselves to transform the way we go to market.

And it's quite interesting.

We've made a lot of progress over the last few years as these models have gotten smarter.

So let's get right into it.

So I think just being honest about where things are is an, a good way, a place to start.

So a lot of, in my experience, a lot of sales teams are really flying blind when it comes to how they go to market.

There's broken maps, stale data a ton of silos of information.

So your AE is not gonna know, if their champion left on a certain call, if they're managing a bunch of other deals.

Your manager's spending all Friday morning.

Herding cats getting information ahead of the forecast call.

There's just data everywhere.

Account plans are gathering dust from months ago when you actually filled them out.

It's just not a fine-tuned way to go to market.

We're drowning in tools, but we're starving for intelligence.

But imagine a different world.

So a world where your market maps update themselves as new information comes up.

As the market evolves, your account plans account for all of the changes, all the conversations that you're having with your customers dynamically on their own, your sales call.

Sales leaders are getting preempted with different call coaching kind of highlights and summaries.

To help their team get better at their jobs.

And your forecasts are powered by AI to accurately keep you up to date as to like where the market is, where your pipeline is and what you're gonna bring in for the business, as well as other like competitive threats, et cetera, in this world.

And this world does exist.

We do this at Anthropic ourselves, so I'm excited to cover, five ways we use AI ourselves to go to market.

And these are all things that are actionable and hopefully you'll all be able to take.

Back to your own companies and implement starting next week.

Alright, let's jump in.

So I'd like to start with a little bit of a personal story.

Back in 2019 the very, very early stage reinforcement learning company, I I had joined as VP of sales abruptly shut down when we ran outta money and had didn't have product market fit.

And it was about a month before my first son was born.

And I definitely just did not want to go through the interview process and have to.

Talk my way through, like, all right, I need time off to be with my family.

I really needed balance in my life and I wanted control, so my family and I decided that it was a good time for me to, bet on myself, work for myself, and start a sales consulting practice.

I'd done fairly well as a sales leader in the past, and this felt like a good opportunity to bet on myself, do something that was a little bit more manageable while I spent time with my family.

So I was working with a lot of startups and scale ups, very technical founders who had, most of them had never sold before.

They were trying their hand at sales founder doing founder led sales, and I was helping them build repeatable, predictable, scalable revenue engines.

The number one lesson that I taught all of these founders was you have to be absolutely ruthless about understanding your ideal customer profile.

Your ICP is your company.

That's North Star.

If you don't have clarity around what your ICP is, you're gonna be spending countless cycles talking to the wrong people, building the wrong things.

You won't have the opportunity to get deep with the companies that matter.

That could be your big biggest deals and biggest accounts.

You just absolutely need to nail your ICP.

And that's where a lot of our effort when I say our, my and the founders I worked with that's really where we focused.

Was building and defining ICP.

So this is the kind of work we would do.

We'd get very specific about, what kind of companies are you targeting?

What industries do they belong to?

How large are they?

Oftentimes, again, we're talking about companies that I'm working with that are between five and 50 employees.

They don't have a lot of the certifications and compliance stuff to go sell into the Goldman Sachs and the JPMCs of the world.

So we gotta go down market.

We gotta go into, the geographies that are relevant here are, the ones where English is the first language.

So that limits, where you can sell in the world.

You shouldn't be chasing that deal in France or Germany.

You're just not gonna be able to support it from a time zone perspective as well as like a language translation perspective.

And there's a bunch of other stuff like Tech Stack and what technologies these companies are your top customers are using, and how well they correlate to the likelihood of them using, future companies using your products.

This is the work we did, and it was a lot of hard work to nail ICP, but with that hard work, we could actually do it and we felt really good about our ICP definitions, but here's where we struggled.

Once you know your ideal customer, you need to go map your addressable market, and that's much harder because the ICP question tells you who to target.

But market segmentations tells you how to basically organize your go to market motion.

And so we would spend a lot of time, this is like weeks and months of work to basically build out all these lists, fill out Excel files, figuring out the mapping, scoring the different accounts that exist, trying to find, figure out what industries and what business models they have and how well they relate to the kind of businesses we're trying to sell to.

And this is where I stumbled upon this, this concept called mei, which is a way to segment the market mutually exhaustive or sorry, mutually exclusive and collectively exhaustive.

Mutually exclusive means that each company in your market map should fit into exactly one category.

There's no overlaps, there's no confusion.

Collectively, exhaustive means every company in your known universe or targeted universe is mapped and accounted for.

So when you have MECE segmentation, your entire team can execute with precision.

But this was, again really hard to do and it took a lot of time and data.

What's cool is right now we live in a really special time.

We live in a time where AI just makes this dramatically easier and here's.

That's the collectively exhausted part.

But here's like the cheat code for this is a prompt that I use when I'm mapping different parts of my team's market.

So I used a variant of this to map our cyber cybersecurity vertical, which is a key vertical for Anthropic And basically you just give Claude the instructions.

Hey, I want this to be a MECE map Here's some examples of what companies exist in this world and how we've labeled them.

Tell me if these labelings are incorrect.

Tell me if there's a better way to, combine vulnerability detection with remediation companies and developer operations with security operations.

There's, you just give Claude the kind of the prompt and the context and the time to figure all of this out.

And it does a remarkable job.

Like one, I'm talking one-shotting this, you get a, the good remarkable feedback around what your market mapping should be.

And again, this is pretty incredible in that you take something that used to take me weeks or longer to do with a lot of support from my, my founder friends.

And we're just doing this, we're doing this really quickly.

I'm talking in an afternoon, and we get going and we get started.

We're talking about everything is scored, prioritized, ranked, segmented, and you just get going and your whole team is ready to just start executing.

So what's changed, right?

It's not that the methodology is different, it's just that AI makes all of this very fast, accurate, and scalable.

So you can move real quickly.

Part two.

So account plans.

Who hear, who hears in sales?

AEs.

All right.

Got at least one or two hands.

How do you feel about account plans?

Arnold, I'm looking at you.

I can't get enough.

Can't get enough of them.

Great answer.

Account plans suck.

Account plans are really hard.

Here's the reality for most AEs.

I, when I was an AE, this was certainly my reality.

You get assigned a major account.

Maybe it's a, an account where you have a $5 million deal in pipeline for it, or they're already spending a ton of money with you and you just want to go super deep.

And you're supposed to basically do a few things.

You're supposed to map all the stakeholders.

That matter in the account.

Usually you wanna map those stakeholders to internal stakeholders so that you have a line, a connection between their CEO and your CEO, their CTO, and the right persons on your end, etcetera But you gotta map the stakeholders.

You have to understand the org structure and business priorities.

You're supposed to understand all of the different relevant use cases that exist assuming you're selling multiple products or can support multiple use cases.

And you're supposed to track those use cases across opportunities and you gotta stay current on a lot of things headwinds in the market.

You know how their stock price is doing competitive threats and Intel.

So you create this beautiful detailed account plan in your first week on the job.

And then you have three discovery calls.

You're scribbling all of this information all over your notebooks and Google Docs and notepad, and it's wherever what, whatever you can get your hands on in that moment, right?

And then you never do anything with the account plan.

It's just sitting there collecting dust.

And in a few months time, your your sales manager's asking you about that account planning, you're scrambling to find all this information which you have collected and put it into, Salesforce or your account plan, we don't have to do that anymore.

So this is what we do.

This is a diagram of or a workflow of like how we work at Anthropic.

Every call that we have, every is recorded by Gong or Google Meets, and what happens is once that call gets processed, it automatically feeds to Claude, which provides a summary, and then that summary gets automatically uploaded to Google Docs.

Okay.

Names of the attendees, some of the key themes you can upload the transcript transcript itself though the rolling notes doc we try to keep somewhat clean with just the summaries and key takeaways and action items. Business challenges that are mentioned, technical requirements that are discussed, any competitive intel that is important.

And again, those action items and next steps.

These get appended to the rolling notes document and.

Those are then fed into the account plan that that Claude keeps dynamically updating.

We keep our account plans in Google Sheets.

I dunno about you all, but you can do it in Docs.

I like sheets just because of the structure and we'll get into what, where that structure and that format really comes in handy.

But it's not just again, the Gong transcripts.

It's Claude pulls everything from Google Docs, from Gmail from internal and external Slack channels.

Again, we have MCP connectors that connect Claude to all of these data sources.

So now it doesn't really matter that whether or not things are all in Salesforce or it's a Claude just again, gathers all this data across all your silos and it just keeps updating your account plan dynamically week after week.

So if a new stakeholder is mentioned on a call you're gonna get, Claude talk added automatically to the relationship map.

New use cases come up.

Automatically gets mapped.

In this case, we also keep track of last meeting, the last time we met people, key people in the organization, we had like color coding schemas.

So like anytime you haven't met with someone in 30 days or 60 days, depending on executive level level within the organization, you'll get that red box again.

It's just highlighting, Hey, we need to take action here, right?

The toil of doing this goes away and now a salesperson can spend more time accurately and in an informed way, executing against the their account plans.

This is not janitorial work.

It becomes actual sales activities.

It's like the fun part of the job, right?

It's, " Oh, I need to, I haven't talked to this person, or This new thing came up.

I need to go talk to my champion and figure out like what threat level this thing poses based on what I'm hearing from that previous."

So you know, this is again, like more things like business priorities being mentioned by the CTO updates, the company overview.

Here we got the executive summary, which we click into.

You can see that there, this captures specific strategic priorities from stakeholder conversations, not just from the website account headwinds that are br brought up new use cases discussed.

These are all uploaded in real time.

The use case tracker in particular is really, this is the lifeblood of a lot of how we work because at many of our accounts, I think Bloomberg, Thomson Reuters, these kinds of conglomerates where you have media, you have, their information services division, you have their legal services division, you have their government services division, you have all of these different lines of business.

You have all of these use cases within the organization.

There is no way to track this in your head.

But Claude does a great job of helping us keep all of this information inside of a spreadsheet ranked and and prioritized and segmented by status.

So this gives leadership real time visibility into both pipeline health across the entire account.

And it really just transforms again, the way AEs work so today, whereas if an AE is actually keeping their account plans up to date, which most AEs truthfully are not in my experience just, you just need to basically make sure in this new world that Claude is doing a good job of accurately representing what's going on in your account plans and taking action.

All right.

Part three, call coaching.

Who hears in sales leadership again how many hours would you say you spend either sitting on calls a week or like listening to call recordings?

Less than five?

Alright.

Some real high achievers.

I love it.

More than five.

About five to 10.

More than 10.

Yeah.

So caps out about five to five to 10.

That's about right.

I think especially for busy managers right now I have a big team.

I have 12 reps on my team.

And I just don't have the time anymore.

I love.

Sitting on calls, I give this weird conflicting message to my AEs, which I'm sympathetic.

I'm like, I love to be on customer calls.

Please add me to customer calls.

By the way, I have no time to be on customer calls, so look at my calendar and if you ever find opportunity, please add me, but you won't.

And it really does suck 'cause I do love that part.

Actually, one thing I've started doing over the last few years that I have a lot of fun doing is I'm a big runner and I'll just go, I'll have the Gong app on my phone.

And I'll go and I'll listen to a call on 1.25x or 1.5x speed.

And I'll like quickly talking to Claude on my walk back home.

And I'll just say here's my feedback and I'll share it with the AE.

Which is pretty effective.

But again, it still doesn't scale all that much.

You still need that, if it's an hour long call at 1.5x speed, that's 45 minutes.

That's a long time.

So what we built is something, called Claude coaching, and we use a couple of different sales methodologies, MEDDPICC and Proactive Selling by the late great Skip Miller, if you guys are familiar.

And the really cool thing about what we do is that every call gets again recorded, transcribed by Gong, summarized by Claude.

And every call gets automatically evaluated against our sales methodology.

Claude watches the calls, it provides structures feedback.

It says, "Hey, here's some call highlights.

Sarah, effectively uncovered the business drivers decision making dynamics and strategic value areas of improvement.

The rep could have done a better job of exploring technical requirements like security constraints, for example." It tracks things again in the kind of above the line, below the line kind of framework, which is again, how we're thinking about like executive level and below, executive level connections and relationships.

So here we see sarah doing a great job talking about business impact above the line, but maybe she could do a little bit better at talking about technical requirements with the engineers and the product folks.

The cool thing about this is this.

When you do enough of these, it's more than just like the individual benefit that you get to go talk to Sarah about this feedback.

You'll start to notice patterns again.

You get enough data, you start to get like a distribution of like where things aren't working and where things are working, and you can start to double down.

If Sarah's really good at talking, the business impact language, then I bet you based on the training or the way we sell or just the culture of the organization, many others are similar, but they're also struggling with the things she's struggling with.

Or it's usually the inverse, right?

People are usually good at talking about features and talking to engineers and champions, but not good at translating it, the conversation into a business impact.

With AI, you can just see these patterns emerge.

You can coach much more than five calls a week.

Every call gets reviewed, every AE gets consistent and methodology aligned feedback, and then you just build a much better sales culture overall.

So also one where feedback is seen as like more of a gift than as a thing to worry about.

That's another.

Concern with that, with not doing something like this is, when you only ever so often jump on calls, pe the team starts to build anxiety around oh, this is the first call Ash has shadowed in a month and a half.

They start to behave differently.

But if it's just known that we're watching, we're listening, we're there to help, we're not trying to get you, it, it creates a culture of feedback is much more sought out and embraced.

All right, part four.

Part four is pipeline reviews and forecast.

So lemme talk, tell you another quick story.

This one's a little bit more painful.

In some ways it was pretty embarrassing for me.

A few years ago I was working at a company where we hired a new sales leader to join the company.

A really smart guy.

Ton of experience, a little bit intimidating, but still remember one of the early pipeline reviews that we had with our management teams. We had just gotten in the habit of doing pipeline reviews in a very specific way, and.

He came in and he just started asking questions that were totally fair questions.

They were just not the questions we were used to discussing in those pipeline reviews.

So things like, " What's driving urgency on this opportunity?

Who's the economic buyer?

And when's the last time we talked to her?" That second part was harder than the first part.

What's the competitor share of wallet and why?

We had no answers or we had terrible answers.

And what really sucked about this wasn't that we didn't have answers or that we had bad answers.

It's that we did have the answers.

It just wasn't in Salesforce.

And it wasn't in that report.

It was in Slack, it was in email.

It wasn't Google Docs.

It was, sometimes in, in the heads of the AEs.

But it was somewhere.

So we try to systematize this and again, bring all the data together at Anthropic and in order to like surface off all this information so that forecast and pipeline review calls are a much more fun exercise of trying to help each other win rather than nitpicking about the fields and things that are not filled in.

Maybe taking a step back though, first I think one of the things I really love about Anthropic is we're really intentional about how we approach sales discipline.

And we've defined what matters.

We believe that, we need clear visibility into forecast pipeline and activities in order to all do our jobs effectively.

We believe in rigor and our cadence rigor and our cadence and hygiene needs to be there in order so we can focus on relationships and closing.

We think that accountability needs to be mutual, right?

These processes support the team.

So like management, the customers and our sellers all need to be there, do their jobs.

'cause we're trying not to catch them.

We're trying to help them.

And finally, we do try to make sure that we're protecting selling time.

We I've worked at companies where you're just constantly uploading and updating spreadsheet after spreadsheet, and you gotta make sure you're protecting seller time.

So we set up Tuesdays and Thursdays we be meeting free by design.

And again, we're, what we're trying to do here is, we're trying to solve this fundamental problem of getting intelligence from data where despite where it lives so that it's accessible when we need it.

This is, again, another workflow that shows kinda like how we do pipeline pipeline reviews and AI weighted forecasts using Claude, which pulls data from all of these different sources and and their MCP connectors.

And, this way when you ask, review my pipeline and flag any gaps, Claude isn't working from what you remember to tell it.

It works from all the context that you have all across all of these places.

And we've tried to build all of this into a, an operating cadence, an operating rhythm that really is just universal across the organization so that we're all on the same page.

The heartbeat of the organization is very it's again, very much in sync.

So Monday we try to do team meetings and one-on-ones.

Tuesdays and Thursdays, again, our protected selling time.

Wednesdays we're doing pipeline reviews.

Salesforce updates are due on Thursday and forecast calls are on Fridays.

So again, every Thursday afternoon, basically an AE asks Claude, Hey, review my pipeline flag anything, sale or incomplete.

Now they don't have to go hunting for this information.

It comes to them.

Claude scans everything and just automatically helps 'em see it and feeds it to the to Salesforce.

From there for managers we're basically asking doing the pipeline reviews on Wednesdays.

And then on leader on Fridays we're doing our forecast calls using the Claude generated forecast that comes straight from Salesforce, as well as other data.

Alright.

Let's take a look at this.

So this is what this looks like.

This is a weekly pipe health pipeline health report that Claude has generated.

At the top, you'll see some of the team metrics, $125 million in total pipeline.

What the percentage of Salesforce compliance is, how many still opportunities we have.

And below that we have, the flagged five opportunities that are in advanced stages that are missing, critical qualification fields.

Now these are gaps that would've made us look unprepared, right?

In that pipeline review session that we had a few years ago.

This would've helped me really get ahead of some of this.

And again, we would've had the information we needed in that moment.

And this is a report showing kind of the sale opportunities one by one.

We're seeing here some of the underlying concerns surmised by looking at the available information across systems. And it's not just problems, of course, that Claude highlights, it highlights when things are going well too.

Here we have a few reps that are really improving with respect, velocity.

We have a lot better compliance across these folks in Salesforce.

And, the Claude's also great at identifying, again, these competitive intelligence patterns.

It's always, it's basically like having an always on virtual frontline manager that's spot checking and trying to basically assess data and make sense of data in a way that human beings just aren't able to do.

And so this is again, every Friday.

This is our forecast call cla.

Claude does the same thing.

So look at this AI weighted probability analysis.

Claude evaluates every deal using 14 factors and basically gives you scores and likelihood to commit to close by certain dates and calls out any, re-forecasting that you should do, whether it's a stage, date or amount based on the data that's available.

So here you got threat watches currently, in best case, but based on.

POC success and fiscal year urgency.

It should be a commit, whereas Claude Defense is currently in best case, I dunno if we're gonna see it.

There we go.

Currently in best case but, probably shouldn't shouldn't be.

So this really transforms the way we operate.

We operate completely.

Instead of really again the tone of the forecast call or the pipeline review call being like, Hey, go update these five next step fields.

We're having conversations around, " Hey, it seems like these deals are stuck at stage two across this product line in this kind of industry segment."

What are we doing on from a product perspective, from a sales engineering perspective and a sales perspective?

To try to get ahead of this.

What are the underlying concerns here?

We have a much more fun and thought provoking set of conversations at all levels of the go to market organization.

All right, last topic, and this one might surprise you.

'Cause it's not really flashy but it's really important.

It's brainstorming.

So I believe this is probably the most overlooked thing in sales.

We talk about the importance of hiring a ton.

We wanna hire smart people, creative problem solvers, people who are great with customers, and can get arrive at solutions, especially when you're like basing down the barrel of a really hard set of problems. But then you know, what good is in that intelligence if we're not really giving people the time and the space and the tooling.

To think deeply about their problems. And this isn't, of course, just about sellers, right?

This is for everyone in the organization.

Strategic thinking I think is something that is just not done nearly enough or not emphasized as far as importance goes.

Nearly enough.

And these things, these can be multiple things.

It could be like as a sales leader, you might be wrestling with, why isn't this ICP definition resonating with our messaging?

What's actually blocking kind of the specific enterprise deal?

Should we bet big on this new vertical?

Or should we not?

These are hard questions and a lot of the AI tools that are out there, in my experience, they're great at helping with execution work, but they're not as good with brainstorming work.

So it's all about, I think in this case, like the idea of having a great virtual thinking partner.

Who helps you think more deeply?

So this is Tempo.

It's a product that's currently in alpha.

It's designed to be exactly this, a brainstorming partner.

It's not designed for quick answers or task execution.

It's for helping you.

It's there for helping you think through complex and ambiguous problems. So here's how it works.

By the way, I just noticed this was here.

I've been like looking back here the whole time.

I. So with Remy which is our thinking partner here, you start by you start by talking you basically voice dump everything.

You just talk, right?

And you the way you could use it a few different ways, but the way I like to use it is by speaking into it and letting it write back.

It creates this sometimes with like voice to voice, there's like a weird latency, like up.

I'm talking and then it wants to talk.

And so this kind of avoids that.

It'll start typing and keep talking and it'll stop and it'll slow down.

It'll just let you go.

And all of this messy and unclear and half form thoughts, just get 'em out.

Get 'em out, and and Remy transcribes the product.

So over as you continue to go, it'll clean up all your thinking, all your rambling, it'll pull off the signal, it'll summarize and tell you what it's hearing, and then it'll ask really sharp.

Like Socratic questions that you haven't asked yourself.

And these are really helpful and they force you to articulate what you're actually wrestling with.

And once the problem is clear, in my experience, the decision becomes more obvious.

Sometimes it takes a few rounds.

Sometimes you might wanna go talk to a physical person as a follow up to speaking with Remi, but it just gets that path cleared for you to arrive at these decisions yourself.

So really, again, with Remi and this kind of product it's not that the output is like a document summarizing what it's learned.

It's, it gives you the clarity is the byproduct that you're seeking, right?

That clarity leads to better judgment and it leads to stronger outcomes.

So this is different from everything else we've talked about in the sense that, everything we've talked about from market mapping, account planning, pipeline reviews, these are all about better execution, which matters a lot in an organization that is growing and trying to grow even more quickly.

But brainstorming is about better thinking, and in the long run, the quality of your thinking really determines the quality of your execution.

Alright, so wrapping up, I wanna leave you with three key principles here and that guide how I think about using AI in, in go to market.

First thing is the AI and human relationship is symbiotic.

There's things that AI can do that human beings just cannot.

AI can, our haiku model, which is our fast.

A cheaper model can read like multiple textbooks in like seconds, right?

Obviously we cannot do that, most of us.

But humans can do things.

They have common sense, they understand the physical world, they understand relationship dynamics in ways that AI cannot.

So these are, it's a real, it's really powerful when you're using AI and humans in the loop together.

And it's a really symbiotic kind of relationship overall.

It just gives us superpowers.

Number two, reimagine everything.

A lot of companies are falling into the trap of just injecting AI into what they're already doing, which is fine if you're looking for marginal improvements or if you're optimizing for like a local maxima where, there's a ceiling, but.

AI we should really push ourselves to just reimagine everything and build from scratch wherever we can.

Hey, what would this look like?

Should we be doing account plans and sheets?

Is there a better way to do this?

You should really break everything down and then think from first principles about how to get, how to do it better with AI.

And then finally using the right tool for the job.

Again, if you want fast answers, you can use Claude if you want deeper answers.

And you want like.

All the connective tissue across your whole, whole go to market stack, push into Claude, then use Claude with maybe extended thinking and deep research and connectors, all to your Salesforce, Slack, Asana, et cetera.

And if you want a brainstorming partner, you can use a tool like, like Tempo.

So here's what's important, I think for you all to remember, is the companies that master go AI power to go to market are just gonna, they're gonna beat their competitors.

Not 'cause they have better AI.

'cause they've learned how to like, amplify the human experience and capability in a way that ai can do and unlock for you in in a way that not using AI cannot.

You can make more of your team resemble more of the top performers on your team by assessing what the like signals are that make them top performers, and then like building coaching and scaling programs around, around those learnings.

So what used to be tribal knowledge is now is now something that is available for you all to be able to using and win.

All right.

That's my session for today.

Happy to hang out and do some Q&A.

Yeah.

Hi.

I am not an AE, I'm the head of data for the company that I'm at, but we also rebuilt our go to market stack about 18 months ago.

What I'm fascinated by in all your slides was how much you're using Google Docs and Google Sheets and our go-to-market team is constantly pushing to get data into HubSpot or get data into all of our systems. And the CRO and I are putting, are trying to wrap our heads around this because we're like, you're asking these tools to do a little bit too much when we could probably just use.

Google Sheets or Google Docs for a lot of this stuff.

Yeah.

And of course we get a ton of pushback on that from all the AEs and all the, all the, and the TSMs and stuff.

I'm fascinated by the Google Docs thing.

What was the thought process behind that?

Like, why not do everything in Salesforce or, why not push everything Gong calls and everything into Salesforce and why, keep them in Google Docs or is it just because Claude is so powerful at reaching across all those tools?

Yeah, it's a great question.

I, this is something I've had to actually change my thinking on quite a bit as a sales leader.

I've always been of the mindset of the system of record is Salesforce.

Everything needs to be in Salesforce.

When another leader from another function asks for something in a spreadsheet or a doc format, I would push them to, Hey go build that report in Salesforce.

The data's there.

Don't make my AEs go write another doc or fill out another spreadsheet.

I hate, I hated that.

That experience.

What's really changed is a couple of things.

Number one, the models have gotten a lot smarter and their ability to parse and bring in all this information, collate it and make sense of it, is much stronger today than it was even six months ago.

So it makes a lot of this stuff possible.

And number two, with all of these MCP connectors.

Two Docs, two HubSpot, two Salesforce, two, Asana, two... there's a whole, if you go to our MCP connectors page, you'll see a bunch of different things and you can build your own MCP servers as well if you're at all technical.

But so it's less about where the data lives right now.

It's my current take.

It's more about as long as it is accessible via Claude.

It's fine.

We should still try to, the systems teams should still try to find a way to push all of the relevant data into Salesforce because Salesforce is designed to keep track of and forecast and review and inspect all of these things.

It's just not as critical that the AEs have to move it around themselves.

Great question.

I'll go there.

Hi, marketing, not in sales, but I have a question about tying the call, coaching to outcomes and if there were any surprises because we just, I see variations just in terms of the secret sauce that some salespeople have, whether that's the relationships or things that maybe aren't captured in the call recordings.

And just if you've seen any correlation there.

Good question.

Thank you.

I think the biggest takeaway, again in terms of like when you start, you don't see these things once or twice.

You see them multiple times.

Usually distributed across the team.

I would say like there are a lot of hunches that you would have as a sales leader about, based on who you've hired, the seniority of the team how you've heard them talk about.

Your product or how the company culture wise speaks about your product in the market.

Like you'll have hunches.

Like my hunch at Anthropic was that we were just talking about features and model intelligence way too much, and we weren't translating it to like business value, especially when we were talking to executives.

And we weren't talking about, Hey, this is gonna do this for COGS, and this is gonna be your, your gross margin improvements that you should expect based on, what this model will do for you.

We were just talking to, we were talking like engineers much more than we were talking like CEOs and CFOs and CTOs.

With this, you start to see those patterns and it becomes, you take a hunch and you transform it into something that's more like data backed.

And then again, you can, like your coaching and your sales enablement programs even can be informed by what you're seeing, so you drive those better outcomes.

Because you know that if we were to speak like executives more often than not, and loop them in and speak like them you're just gonna have you're gonna do much better as a sales team.

Question on somewhat related to the first one.

I get the idea of it doesn't matter where the data is, you're grabbing it from everywhere to put it.

But from a sales rep point of view, where do they see it?

When you showed like that pipeline earlier, what UI do they go to see that information and what else is there for them?

Yeah.

So Salesforce is the system of record, right?

I still will be a sales an old school sales leader when it comes to that, like.

You should live and die by your Salesforce.

And, but again, it's just less about like when you're updating your dollar amount for deals close dates, forecast categories, probability to close.

All of this stuff should be in Salesforce.

And certain information about relationships should also be in Salesforce.

It's just that, I guess the point isn't that we shouldn't have a system of record and enforced discipline around updating it.

It's that these calls, the forecast and reviews and the forecast calls and the pipeline reviews, they're meant to serve the business and you serve the business much better.

If this information is pulled in, there's a time and a place.

To hound AEs about not updating their Salesforce.

It just shouldn't be that weekly forecast or pipeline call that you have, that you gotta make that thing productive so that you can push the business forward.

So in short, Salesforce still it's just again, our time can be better spent in these conversations.

When you showed.

Are they looking in?

Yeah.

That was the second part of your question.

It was where do they see that report?

Yeah, that report is a, it's a Claude-generated artifact, so I literally, Claude will make it for me every week when I ask for it.

And AEs can make their own.

Of course everybody can make their own, but this is not a place that they go look as much as it is, you just ask Claude in natural language, Hey, tell me about these, tell me about the week.

Usually I'll have a prompt that I reuse and it just spits that out for me.

As a Salesforce shareholder, I'm terrified by what I've just seen because it feels like that's a system of record of history.

Yeah.

And Claude potentially is replacing forward looking.

I don't know.

Claude will not replace a CRM.

As a company we are a our positioning in the market is that we are building intelligence infrastructure for other companies to build on top of.

We want, and Salesforce is a big customer of ours, very strategic customer.

We want Mark Benioff to take all the good stuff that we're powering with our models and be able to make Salesforce far better tomorrow.

We are not in the business of creating a better CRM.

What, as a shareholder I think the concern should be all of the dozens of YC companies that are looking at CRM as a category and thinking, Hey, I can build something better than Salesforce very quickly.

That is AI native.

That is the concern.

But ultimately I think we need systems of record like Salesforce, like us, the ServiceNow, like the workdays of the world.

There's a lot of also compliance and regulatory stuff you have to consider.

So those aren't going away anytime soon.

It's just a matter of making them better.

All right.

Any anyone else got one in the back and one over here.

You should never ask two questions, but what does this look like in 12 months and does the ideal hiring profile of a top enterprise rep change and how.

Those are two big questions.

I don't know what 12 months looks like, honestly.

Six months is hard now at this point.

Everything is just moving so fast.

Like I remember when we entered 2024, or sorry, 2025 we were talking about like how agents are gonna, it's gonna be the year of agents.

I don't think any of us really knew what that was gonna mean.

And now here we are and it's definitely become the year of agents.

I think.

There will be a lot more, like a lot of, a lot more of what I showed will just be automatically updated in real time.

It'll be like every I think every sales manager.

We'll have a team of physical employees and a team of virtual employees, and they may not be working, in, in parallel to one another.

I think there'll be some like partnership, but every workforce, every Salesforce will be far more productive because things like this will just be automated and like in real time.

This is not dynamic.

This is this is not a real time dynamic.

It's up.

Like this stuff is like updated on a weekly basis In batch it's there.

There's a lot more, like in the moment it's gonna just like boom.

In the call you're gonna get Claude tell you actually, oh, hey, this person mentioned this.

You should mention this.

There's gonna be a lot more of that.

But who knows.

The second question, is the profile of an enterprise seller going to change?

Is it changing?

Will it change?

I think probably, there's a lot of who here's in the in the Bay Area if anyone, oh, everyone's here local.

The Bay Area, there's a lot of like corny AI billboards everywhere.

It's actually really annoying.

But one of the things that, like a common themes you'll see is AI is not coming for your job.

A person who knows how to use AI is coming for your job, right?

When I think about the, like the new wave of enterprise sellers, it's less about an age or demographic shift.

I think it's more about have you taken the time to learn how to use AI or not?

The people who know how to use AI, irrespective of whether you're 28 or 68, I think you could be a strong enterprise seller.