what is gonna make you stand out is telling leaders what they don't know about their

business.

You cannot automate something that one you can't do yourself.

I don't go out and make coding agents because I don't know how to code

and I don't want to step tell people how to do their job when I've never done it,

right?

All right, welcome ~ to another episode of the RevGenius Podcast.

I'm joined with Angelo Quiambao,

go to market engineer at Intellistack

Excited to chat Jared.

So one would love to understand a little bit about your background and then tell

us that to help set some contacts with the discussion.

Sure y Yeah, absolutely. I mean running as the AI go to market engineer over here

at Intelestack, what that entails essentially is,

you know, my job is to create agents,

scale them across our go to market motion and most of all help our team see success,

pipeline revenue and efficiency across sales,

marketing, rev ops and all the other underlying go to market teams.

And what is Intelestactive?

Essentially they're historically they're a document workflow automation platform

for the last twenty years known as FormStack,

but as of recently getting into the agentic AI workflows,

primarily helping, you know, healthcare industries,

insurance, higher education industries with workflow,

form fill, all those sorts of documentation that require a lot of strict

security implications like HIPAA,

SOC two, all that.

I love it, yeah, because I I remember form stack.

I I didn't catch the rebrand. ~ that's super cool.

Same company, yeah.

Now why did they decide they needed a GTM engineer?

Yeah, that's a that's a good question.

I think the world is shifting towards go to market engineers and you know,

when I approached their leadership and w we had a great conversation,

it it wasn't the conversation of can we replace our sales team with a

go to market engineer, it's how do we boost their efficiency and scale

and I think that's the way to approach it and I I like the way

they had their goals lined out and that's where a go to market engineer

I feel can help is to scale processes that work and scale high converting workflows

~ or create them, you know, if need be.

So before Telestack, ~ did you have experience in GTM engineering

or like what's your background?

Yeah, I came from a company called Qualified.

they they were recently acquired by Salesforce.

I started there as a B D R. I did sales outbound prior to that for

the last like five, six years.

I think the benefit of that is like I was able to see outbound back in the day,

you know, with your headset and your trackpad.

And then

to the motion where people are starting to use chat G P T to help with emails

and then now it's fully integrated into enterprise scale.

~ at qualified as a B D R was able to be there at an early period.

helped

them sort of frame a lot of their outbound frameworks to where we've seen

a lot of success. then we identified that on the outbound realm.

we saw a ton of success with our flagship product that qualified being

an inbound agent and then us creating outbound agents on the side.

I think we were able to see over like sixty million dollars

of agent generated pipeline there.

but yeah, it was great.

Now, the GTM engineer role is one of the fastest growing roles in go to market.

I think you mentioned 200% growth year over year.

There's so much noise around what it actually is and does.

Is it RevOps renamed? Is it outbound agencies rebranded with the use of some

~ some better products than they had years ago?

Out of

Curiosity, what what do you think the biggest myth about go

to market engineering is?

That's great question. I don't wanna hurt anyone's feelings.

But I think the biggest myth is that you can automate everything.

You can automate your go to market with one person or a number of agents.

I think that's an absolute myth that needs to be distinguished because it's also

hurting, you know, the expectations of what a go to market engineer does and

biggest part of my job currently is to set expectations with different levels

of our go-to-market and departments,

but also operate on that level.

So if you think I can automate outbound and get us unlimited revenue,

then you're gonna be very disappointed when I tell you like we need

to figure out who our ICP and map out all these and l align this,

make sure we're in compliance with security,

like all of that is the biggest myth.

So don't try and automate outbound because it shouldn't be.

There should be humans in the game.

Yeah.

And to be clear, like I was so interested in having this conversation because

you shot me some numbers, right?

Like ~ you created a solo outbound agent,

a team co-pilot ~ pilot agent,

and I'm seeing 190 meetings generated from the solo outbound agent

for like 15.4 million in pipeline source.

The team co-pilot agent 132 and 13.7 combining for Q1.

322 plus meetings and 29.2 million

in pipeline sourced. That's pretty incredible.

How long did it take to do that?

A long time. ~ it to it took about I would say it was always a work in progress.

I think that was the best iteration of it,

but it worked in tandem because we drilled in a real outbound

p ~ process done by humans that converted at a high scale.

So our our human team was really out there,

pipeline generation machine and we just scaled the highest converting workflows into

an agent and scaled that across a higher TAM.

instead of the other way around.

is

scaling across a higher TAM with unproven workflows,

right?

Yeah, and and you built quite a few workforces,

so to speak, different

Yeah.

agents or teams of agents to do different things.

Would love to be walked through like each of them,

like the end to end agentic outbound workforce,

the ICP qualification agent, and the problems you were trying to solve along

the way.

Sure, yeah, it's ~ it's I don't have access to the actual agents anymore,

obviously 'cause I'm I'm at a new company,

but I think there's

A great way that I look at look at it and I think of agent building

or an agentic layer to your go to market motion in seven different layers,

I think. In order to have any sort of agentic motion within your go to market,

you need these seven layers. And it's like layer one's the targeting,

the data, your ICP, right? Your triggers,

your signals.

Why are you reaching out to these folks?

your routing, decisioning which agent does what,

what is their decision-making capability?

Now then you need the content,

the assets. Are they just generic outreach or gong sequences or

are they hyper personalized? What are the campaigns and outbound plays they'll run?

And then the knowledge for that,

right? Agent battle cards for competitors,

for replies, for objections, value statements.

And

then I decide where does the human intervene?

Where is the decision making for approval?

And that always stays no matter what,

because if you know what you're doing,

you look at an email, you'll be like,

that sucks. If you don't know what you're doing and you see an email,

you'll be like, It made an email,

that's great, send it, right? we can also ingrain those into agents later

on as QA agents. But the biggest thing too is there needs to be

a layer for feedback. Like every week with my agents,

the reason

They

got so good so fast is because I treat them like their employees.

I'll ask them at the end of the week,

what did you do well?

What did you not do well? Where can you improve?

And how do you feel you did towards hitting your goal and your KPIs

and your metrics? And th they sometimes they'll outline everything you need

and then you can rebuild upon that.

And long term memory about all the highest converting ones,

similar to a human, like you get a good email that leads to a meeting,

let's study that. Let's replicate that.

So those are like the seven layers that I went by and how I build

all my agent workforces.

And they're all necessary.

So one cool takeaway there, and and I love I I think you have this written out,

right? Like in terms of the

Right.

framework. Can can you share your screen?

Cause we'll we'll dive into

Sure.

each layer.

Absolutely.

The part that I wouldn't be surprised if more teams aren't doing is they

is the goals you set for your agents,

similar to employees. Like can can you walk me through like the goal setting,

how you figured out what appropriate goals would be?

Did you just pick a a number out of the sky?

Did you use what the board wanted you to hit type of thing and then back channel,

which you don't want to give to any human directly because it's probably inflated.

~

How how'd you how'd you go set and level set?

And then how did you set that up?

Yeah, that's a that's a great question.

I think for any go to market engineer to establish goals you need to know,

first of all, what the goals are from the board level to the C level to

the VP level to the manager level to the B D R level,

right? Depending on what agent you're making,

because all of them have different goals aligned to the same goal.

So if our

Like I report to our chief sales and revenue officer.

Typically go to market engineers report to VP RevOps.

Like if their goal is to boost pipeline by one million for the next,

you know, two quarters, like that's their goal.

The managers that trickles into a metric goal of we need fifty more meetings

per month, something like that.

So understanding the world of every single leader and every single part of

the go to mo ~ market motion will help you sort of speak your language

and then establish

The goals for the agent. So if we have an agent that does only emailing

and the main goal

Out of curiosity, if in

your example, if you have to do 50 more meetings,

take your number, right? Like assuming 20% conversion rate,

whatever A C V.

How do you pick what the team owns and how to push humans?

And what does the agent take from that?

And is it like a sliding scale?

As the agent starts, you want the team to own more.

And then as the agent picks up,

the agent can own more.

Yeah, I think that's the the right way to think about it.

A crawl, walk, run approach, right?

For any organization. We have the agents start targeting known areas that

are lacking

like bandwidth from the team. So say maybe like pre and post event follow

up or there's a you know, a bucket that needs to be a attack but isn't

for some sort of reason, that's where we would put the agents to prove efficacy,

to prove, you know, their architecture and how they work across

the different systems. As they start to show more and more positive outcomes,

then they become more and more integrated with

all of our team. I think at my last gig we started that way and but by

the end of it, you know, it was a required KPI.

It was ingrained into the workflows of every single rep,

like you must use the agent for X amount of emails.

Now walking through this layer one targeting and data.

We know that this is an underrated challenge,

right? Like is it's not just my c ~ I'm working with companies over

a hundred employees in North America,

VP of RevOps, Tech.

go into Apollo and and and no shade on Apollo,

like go into your your tool of choice,

let's say, a single tool. Let's let's size up.

The TAM or the or the Sam, right?

Like how what what do you set up intricately to fine-tune that?

Right, I you hit the nail on the head,

right? Anyone can say we we target marketing folks,

we target IT folks, we target,

you know, cybersecurity folks in enterprise.

what does that mean? And again

For go to market engineers who come into a process where maybe they don't know

who they target or maybe there is no ICP,

you must have the knowledge to cr like sort of create that,

right? Understand the product that you're selling.

Know how that aligns with marketing.

I mean IntelliSec's been around for twenty years,

so they don't have, you know, that sort of problem,

but I can see areas where, you know

That's not clearly defined. You know,

tools like Apollo and Zoom Info have done a great job of incorporating their tools

into MCPs for you know agent use and for go to market engineers,

but they cannot decide the process nor the ICP for you.

it needs to be a mixture of RevOps,

marketing, sales, and then speaking with the BDRs themselves,

like who generates meetings? ~

Well, you s you said something interesting.

Your company's been around 20 years,

so they don't need that. I think that's a perception,

right? Like, like there might be more meat on the bone and there might

be quicker deal cycles to be had and more focus.

Did you find that? Like, like by by just like adding some a bunch of nuances there?

And again, I've only been here about six weeks,

so that is T V D.

Okay.

But I I think you hit the nail on the head.

There maybe there are other verticals that we appeal to that

we historically have not been able to reach out to,

and I think that's the beauty of the agents.

You'll learn very quickly that we may appeal to these or no that messaging does

not work, right?

What ~ what what tools are you using to create layer ones?

we use the info internally. obviously all the other AI tools like you know,

the L LMs like Claude, GPT. I'm a big fan currently of this tool called Vaughn.

They're made by ~ Rattle Rattle.

Yeah, I use them heavily. big fan of them and everything they're doing over there.

But I think that's h been helping me sort of establish this targeting layer,

so to speak.

Awesome. Walk me through layer two,

the triggered signal.

Probably one of the most but depending on the product,

one of the hardest things to identify,

right? The trigger we used to go based off intent,

right? Sixth intent, demand based,

marketing intent, MQLs, but I think with how much data is out there,

it's intent is almost dead, right?

It's more like why do we need to reach out to these people and why would they want

to take a meeting with us if we're talking about outbound?

So triggers and finding them

You can find this externally and I try with like z zoom info,

but I think it really comes from understanding the product.

I've watched over five hundred gong calls,

you know, with different industries and first ten minutes of any gong call they'll

tell you why they took the meeting.

To me that's a trigger if we see similarities there.

it's something that

Y like our internal data, every company's internal data is something that

no one else can query and what we have is a weapon.

External data is what agents and every other AI can find.

It's what makes it generic.

Yeah, so I understood. I I almost call them commoditized signals,

right? Like job change, funding announcement,

blah, blah, blah, blah.

Right.

They're a signal, but they're not necessarily signal you should act on.

the talk to me about operationalizing,

like so the first ten minutes of the call,

theoretically, you're probably able to just if if if you have the right setup,

query the call recording, the the the slew of call recordings and

say what's what's the first reason that they said or the most prominent reason.

Right. And then from there, how do you make an agent find that?

Like like what are some of the examples theoretically of of things people have said?

Yeah, and one caveat with that is you can query it and you can just ask an

AI to do that. But I like to watch them because the the translation of what

the AI thinks is the problem versus me having done outbound or having done sales

could be different, could differ,

could be the same, depending on the model you use,

but

You know, when the AE starts a disco calls,

like what prompted you to take the meeting?

You know, it's like does the A does the AI know if it's inbound,

outbound, partnership, renewal,

right? These are all different buckets.

Like if I'm in the business of net new meetings,

I need to know why they took this meeting and most commonly people have

a pain sort of like, you know,

this competitor is not working out for us or lately,

AI mandate, right? We have an AI mandate.

Or, depending on who it is, like C levels have more of a

broad

level pain, but then the tool owners,

the ones who feel the pain have direct product level pain that they need,

you know, help with. So understanding that,

then we can make our b battle cards,

value statements, intros, why,

right?

So different pains for different personas.

So you can get like real granular with with your outreach to a founder versus

a sales director.

At the enterprise level, like generic is not enough,

right? Like everyone is reaching out to that IT director or that,

you know, chief marketing officer of Netflix.

Like how many people are emailing them?

Like what is gonna make you stand out is telling leaders what they don't know about

their business. From five hundred calls we've seen,

people in your role typically feel this,

right? I mean obviously that's not the intro,

but that's how we become experts,

right?

more potent than I saw you commented on this competitors p of ours post.

You might be in the space. Although that signal shouldn't be discounted,

but it's probably not the main one.

Yeah. I mean it it's all just layers,

right? Like it's not the what we base things off of,

but the more the better.

Now layer three, routing and decisioning.

Taking that raw data and turning it into a decision.

Putting your best outputs or best decision makers brain into an agent,

right? If I'm the best outbounding rep on this company,

then I'm probably should be making the outbound agent.

If you're not and you walk into a company and they have a you know,

a bunch of BDRs doing well, like find out what the top BDR does and scale that.

where are the real inconsistencies within their own process and

Having the agent have that sorry,

go ahead.

no, go on you just stopping on the point,

did you say you should have the best outbound rep build the agent themselves?

Or work in tandem with them, right?

Like 'cause they know the ins and outs

A hundred percent.

of it, like, when you dupe check a lead in Salesforce,

this is the process, versus asking someone above that or using what you done before.

Like the every company has a different process for how CRM intake

or dashboards are made or how we generate leads.

Like those need to be ingrained for the company specifically,

not just, we can add fifty leads tomorrow and blow all through your Zoom info

or Apollo bank, right?

I love that. ~ working in tandem with the rep or the rep doing it

and and you bringing it into your ecosystem and fine-tuning it or integrating

it or both is something we don't hear enough about,

I think, right? Like it's kind

Yeah.

of like

if reps are doing something, they kind of I f I sense that they hoard

it as their secret sauce. building this culture.

Of finding the best way and letting the people involved do it.

Like I I just did that with my team.

Like we we brought on ~ somebody to to handle some GTME stuff,

but like I realized there was a disconnect between them and the team.

So I'm pushing the team to build what they need when they need

it and like communicate and like really take an ownership of that because.

Nobody who's not involved in the process is gonna know it as well as you.

They might know how to build it better than you,

but they're not gonna know what to build better than you.

One thousand percent. I think that's the best way to approach it.

And you know, companies ask me,

like, what's the first thing you'll you'll build when you get here?

I have no clue. I don't even know what the problems are,

right? Like, I'm gonna spend a lot of time understanding internally what's going on.

And areas where I'm not an expert in,

like I don't know product marketing as well as it should.

I don't know the intricacies of RevOps and building and all doing all that.

So I will lean on their expertise.

Like the agents in Go to Market are

a culmination of every single department's highest converting workflow.

So that way everyone agrees upon them.

It's not competing, it's only helping accelerate what's working.

E either highest converted or ~ i i if they don't have any or many,

biggest impactful one that you create together,

right?

Hundred percent.

Now what's the fourth step?

Content and assets. I think something that gets ~ skipped upon.

Can you scroll down a bit

the same way you train a BDR, like you give them sales enablement material,

but I think what a lot of companies have now because of AI is a lot of data,

a lot of AI enablement material,

all that good stuff, but they don't have context.

The agents need context, and that is what the AI go to market engineer would bring.

when you see a head of IT you say this.

You

when this reply comes, you say that.

If this comes, do that, right?

Like th these are the contextual knowledge that needs to be built

in from understanding of the role.

you take your sales enablement material on all your playbooks and then turn that

into agent what I call agent battle cards,

right? Like quick knowledge bases or context files that the agent

can reference when it comes across any scenario.

This will save you a lot of time in reiteration,

hallucination of the agents, misfirings,

bad emails, just a ton of like probably one of the more important parts.

Like the only thing that predicates an agent doing well is if it says

the right things.

Otherwise it's just AI slop like everything else and that is the last thing

you want and because you lose trust within the organization but then

it also hurts your your email domain,

all that good stuff.

Now, I think the word context,

you know, in the header here might be missing because sometimes content

and assets alone are based what the internal team thinks is best.

And and if you take that as is,

you'll stay on brand a hundred percent,

right? But it doesn't necessarily like like you don't know if that's

the best thing to show them.

Understood that if you're working for a bigger company going to the enterprise,

that the the best thing that's not in your queue of things might need

to get approved before before using it.

Understood. But ~ are you building in feedback loops in each one of these steps

to like say, like, okay, cool,

you gave me 50 pieces of content,

you classified it based on verticals,

personas, company size.

One like

I want to know because essentially your content and assets is your

hypothesis to tested hypothesis,

right? Like of what's

Yes.

of of what they need to see. And maybe it's work to get you here,

but you might not have a motion to get you further with with with that.

And and I'm say I'm I'm layering into layer four because sometimes the context

Isn't always used in the content because there might be new context that came

out since Claude dropped, you know,

their newest model, right? Like s like if if you're ~ if you're Canva

or something selling to the enterprise and you're having a fine time,

and then ~ Claude design drops,

and now you're hearing that time's not fine,

theoretically you could act quicker.

Than a content team, perhaps, like in real time.

You could test quicker, that's for sure.

Hundred percent. And

I lean on the expertise, like it say I were to create better content,

top of funnel content that I need for my messaging,

I'm in contact with every single leader of the go to market,

right? Like, can you approve this quickly?

Is this okay? Is this on brand?

This is my goal, this is my you know,

MO, this is what I'm trying to do,

this is what I'm trying to evoke,

and these are the results. Like every single change,

every single thing that I do needs to be aligned with every single part

of the organization. And that's from security

to marketing, to sales, to RevOps,

you know, on all different levels,

right? Like without the buy in of the leaders,

it it's n it's nothing.

basically what I mean by content and assets.

Like there's internal content and assets,

but then there's agent content and assets that you would need to create as a or,

you know, as a go to market engineer and like a value bank,

persona, buyer committees. If these aren't built yet,

you need to build them. All right.

Proof point libraries. You can probably get these for marketing.

If not, you gotta create them.

But then

even on a deeper level, you know,

outbound reps know, you know, if you're doing cold email,

exactly what hook or exactly what sort of plays to run.

GTM context, everything about your company.

Like these are our leaders. These are basically as if you're training a real BDR.

and nothing more, nothing less,

right? Enough to do its job. You don't want to give it,

you know, MCP access to every single system and then say,

just go find it.

You know, it's a waste of money,

and credits for sure. But yeah,

Are you are you having your agents add

these are sample

add suggestions to the knowledge base,

like this type of content or this came up in five different calls and

I don't see any content for it and stuff like that?

Yeah,

so like that's the the seventh layer which is like the the feedback loop.

Mm.

So if like they go about their week and they're like,

Hey, I don't know if you've noticed,

but like we we appeal a lot better,

you know, higher open, higher reply rate for these ten people,

I'm like, Okay, let's add that in and it'll add it in for you,

right?

I love that. ~ what's the fifth layer?

Yeah, let's go back to that.

And what I love the most about this,

all the best GTM teams operate like this,

but some people, when they're started,

they think that they could just take like

A tool, entered name of your favorite tool that allows you to create tables

of tons of ICP, layer in signals.

probably that everyone else is going after and just rip it and have success.

This isn't the year twenty twenty anymore.

That that actually probably worked in twenty twenty,

twenty twenty one, but like you have to be way more thoughtful.

Yeah, and maybe there is a difference between like an AI go

to market engineer versus just a GTM engineer,

right? Like are you orchestrating a system or are you operating a system?

and I think there's like maybe levels to that.

Well, we're talking about all fundamentals and

~ conversations internally that any engineer should probably be having.

And if they're not having it and and

People that use AI the most, probably will have the biggest internal conversations

before doing it. Right. And like the deepest and and spend a day like diving in,

talking tomorrow, like talking to different teams and then automating

the hell out of it. Right. Like.

Yeah, yeah.

I think that's what ~ Davida

at Clay said. He's like, you know,

I don't care about tools. Like we talk to understand

Yeah.

of of course we use clay, but if we didn't,

it would be fine. And ~ you know,

we're talking about these processes so much,

so in depth with humans. And then we're of course we're automating the heck

out of it.

That's the key thing, right? You cannot automate something that one

you can't do yourself. Like I don't go out and make coding agents because

I don't know how to code and I don't want to step tell people how to

do their job when I've never done it,

right? And I still stick to that,

right? I know how to do outbound very well,

you know, probably with the best of them.

So I'm able to automate that and I've learned about RevOps,

I've learned about marketing, I've learned about sales in general,

full cycle, right? And these are things that could be learned,

but we're not automating our AE's jobs because it takes a

long

time. Those are jobs that humans need to stay in the loop.

And I think, you know, that's a good segue like into layer five,

right? Delivery and compliance for say you're d I mean this is for email shops

or email outbound is probably easier than creating a voice agent.

But one of the more important things is delivery and compliance,

you know, with your IT department,

you know, with your company's domain and how you're sending emails through,

right?

when I see people say they've automated outbound,

they've sent ten thousand emails per day,

like the only thing I can think of is like ~ fried their domain,

right? And every y

So they have a they they have a hundred

rotating inboxes in a smart lead or something.

I

right. I mean that's great too as well,

but it's also like Google and what they're doing with Gmail

is preventing cold outreach. Like,

you know, they've never been a fan of it,

it's not a secret. Like they do a lot to protect their users in boxes from spam.

And there's a lot of it. AI slop going out at scale.

You know, and

You need to get past that by showing,

you know, familiarity with your prospects and not making it generic.

Like I used to train younger BDRs and I would tell them

Would you reply to that email or like what's different about that email?

And like the agents can do that at scale.

So when we think about delivering compliance,

it's how many inboxes are you using?

How many emails do they send per day?

Are they intelligently sending this?

Is it in compliance with your DCAM or your SPIF requirements?

What is your current domain health and then what tools are you using?

You know, outreach, gong and gange or pure Gmail,

right? Like does automation hinder open rates,

reply rates?

all that good stuff because if you're not tracking pipeline and revenue

you can't track pipeline and revenue in the first two weeks.

Maybe you can, right, depending on the organization.

But at the enterprise level, the first metrics you're gonna sort

of track are replies, opens, engagement,

sentiment of that, and overall efficacy of

the agents in architecture. You can't and if you said two thousand emails

and none of them got through, none of them got replies,

you have to know if that's you know,

a domain thing or if it's a messaging thing or if it's an agent thing

and you need to know fast. So you can fix it,

right?

Now are you also cold calling and LinkedIn in the delivery and compliance?

Because I I I I get that you might not have built an outbound agent

and and all indicators are it's not a good idea yet.

But

You are doing a lot of other things so that they cold call the right people

and maybe there's a signal for when they should pick up the phone or or

use their ~ power dialer or whatever there.

And then also there's looking at when cold calls come in spam and understanding,

you know, that's probably a function of the dialer,

I'm sure, but managing up to those people to

either get that to stop happening or or move to another solution.

And then with LinkedIn there's different nuances for deliverability

and different levels. Like are are

Yeah.

you

And it so I guess the question is like what are you doing with those networks?

And then like understand or I'm sorry,

expound on

Where the humans have to take initiative and and do something.

Yeah, that's that's a good point.

LinkedIn, very complicated channel,

but very lucrative in return. There's a lot of tools that quote unquote automate

and dodge their filters, but I am not a fan of those because LinkedIn is for what?

For connecting with professionals.

So if I'm an AE, if I'm a BDR,

like just do that yourself. Like you can't outbound it I you can outbound

at scale on LinkedIn, but you would have much better results if

you just personally message two sentences to the actual

high value contact that you want and research them,

right? Like we have agents that can identify who to reach out to.

We have agents that can surface the right phone number and the right prospect

to an AE or a BDR to call. But as of right now I don't like automating those

platforms. cold calling I don't pick up cold calls and I'm not even that high

up in the ring, right? So I can only imagine the efficacy of it depending

on the product, but

If we have an AE call, someone they're familiar with based on a trigger,

it's a highly relevant cold call and I'd rather have our AE's do twelve cold calls

to highly relevant people. You know,

like the agents follow the same principle.

They send we used to send a lot of emails per agent.

I was spent s probably sending like five hundred,

you know, back in the day, but now they send about thirty,

thirty-five. And our reply rate almost quadrupled.

And are you sending it from the core domain?

what y previously yeah.

At thirty to thirty-five, you could play with that.

Anything below fifty is probably fine,

to be honest, right?

Right. But

and I was running like twenty five different inboxes,

right? So then that's how you scale it.

Got it. And ~ and even when the AE or the SDR is required to do something,

like take an action on their own,

do you have systems set up to ensure that they hit the minimum number

of actions per day? Like essentially agents managing reps?

Not really agents managing reps.

I think everything that I make is tracked on a Salesforce dashboard somewhere,

right? Like it needs to be visible for all leadership to see.

So we like an agent dash that I create,

Salesforce fields that need to get put in there for tracking fields that

the agents are allowed to mark,

not mark,

These get marked daily basis, weekly basis and then we can track KPIs from there.

Similar to how you would with a human rep,

but never an agent being like you know,

like you have

Yeah, I know.

An agent micromanaging a human probably

Mm-hmm.

isn't good. A KPI saying you need to hit eighty percent of every human activity,

otherwise you don't get your bonus or whatever,

probably better on the macro scale,

right?

You know, you you bring up a good point,

like maybe if we have a rep who's,

you know, struggling a bit on metrics,

the agents can probably maybe identify that and,

you know, kind of supplement their their production on that end,

you know.

I think thinking through it, like I think putting a KPI to ensure that humans

do human things from an activity standpoint,

and I hate to say that if you trust your process enough,

like more activity doesn't mean more meetings,

sure. But like if you're giving them gold,

they should be hitting 80 to 90 percent.

of the gold understanding ten to twenty percent whatever and and then like

if they're not hitting their numbers and they're not hitting that,

then that's a flag. If they are hitting their numbers and they're not hitting that.

Dive in deeper to see like like what's working,

right?

Yeah, a hundred percent. And i it's interesting,

I mean, I don't think I've necessarily given the agents ever,

like, you need to hit five hundred because what I've seen with the model changes,

depending on what model you use,

like they will really try and hit that goal even if they're incapable of doing it,

like, you know, with a tool broken or something,

right?

I don't

think agents should be given that call,

but I think if you

Right.

say humans, you need to call these 20 people today because something they

did within the last 24 to 48 hours happened and it's gonna be like higher value

today. And then you tested

yeah.

it to see like, okay, let's let's test one group without this,

right? Like and and see like who's calling 72 days later,

see like the success on those.

Prime ICP layered with signals,

people based on that. And then you could like reverse engineer like what

the timeline needs to be if there needs to be a timeline.

Because maybe maybe just calling that quarter is fine.

Yeah,

yeah. I mean it it's all about timing and you know,

how fast you reach out to the person during their period of interest,

right? It's like how can how can we get you in for a meeting when you're interested

and if you're not, how do we boost that intent 'til we get there and you know,

the agents can help with both of those things.

Let's go to layer six, human in the loop.

I I I feel like I keep foreshadowing the next step here.

Yeah, y you're helping me out here,

but essentially it's it's what we've been talking about this whole time.

There in my world, in my perspective,

there will never be a point in time where my agent w will run without

me in the loop. I'm not saying I approve every single task,

you know, think of it, you know,

a company that we use or a vendor that we use relevance AI,

they have a great analogy about agents being like a Tesla car,

right? Like you start off with supervised and then

Full self-driving with supervision and then you go full autopilot,

you know, where you're not looking at the road,

hopefully not. But agents are so sort of the sim similar like pathway.

If we have a low-value task but needs a lot of mass you know activity towards it,

say like cleaning out, you know,

an Excel shoe, whatever, right?

Filing these, organizing that,

like maybe you can run that on autopilot.

But anything customer facing, anything that goes out with you know

our AE's names BDR's names.

names my name, maybe we'd have a checkpoint in there.

trust comes from, you know, the quality and the production of it.

Obviously when we were sending like thousands of emails per day,

I was unable to check a thousand of them,

but I would check, you know, each campaign,

look at a couple, and then send the rest through.

I think that's sort of like the realm that it lives in.

But then as a go to market engineer you have to know what good looks like.

is this a good email? Like when leaders are saying why is it not working,

you need to have an answer that is correct and then a solution.

Not just our domain's bad or ~ the messaging sucks.

Like you're in charge of that,

no? So always knowing what approval loops to have,

anything customer facing really,

~ I have as a supervised.

And then how do you set up the feedback loops?

As of late, I mean the the building has become so,

you know, advanced these days.

I set up all of my frameworks and all of my decision making almost like

as if I'm their manager and I put that into an agent.

and I put you know today's best building practices.

Maybe I can show you what that looks like.

we'll see if I can pull this up.

But essentially I have this agent

I

used to do it manually and I still do,

but nowadays I have this agent check all the other agents running as

a separate it's separate from you know the other builds so that it's more siloed

and then it can rank, organize,

and then judge efficacy of them.

Let's see if I can pull it up here.

And we have

Yeah, here.

Can you see this agent right here?

Yes, you meant the

Yep.

This is relevance AI, the platform we use for hosting,

but essentially this

they're a sponsor of RevGenius.

Absolutely love them. What a nice platform.

Love them, yeah. Big fans of them and essentially it's like just has

all the rules and then you'll see like knowledge bases you can input here.

when it runs though, this is what I get back either via Slack,

via email, whatever have you. I mean I'm in this platform probably daily.

I said every day give me a check on each of our agents.

What is broken? So this agent that it did here

I had a broken template, tool references were wrong,

overall health score seventy five out of a hundred.

similar to like a manager looking at their their reps.

So

We need to know what to do and then how to optimize it.

This agent also has the ability through relevance MCV to fix its agents that

it sort of optimizes. So if it checks an agent says this tool needs fixing,

it fixes it. And then we're good to go for the next day.

that's how I check now, that's the feedback loop I'm doing,

but before that, obviously it's very manual,

which is okay, and I think in the beginning it needs to be.

But yeah. Arguably the most important step is that feedback loop

and that optimization 'cause goals change,

messaging changes, everything changes so fast,

especially if you're working for like a giant startup or,

you know, a fast moving company.

How much of that feedback loop skills customized to what you're building

now and like is that shareable?

~ I mean I can give you a shareable version.

Yeah. 'Cause I'm sure there there's some nuance.

If you could point out the parts that would need to get changed or deleted

or or whatever to make

Yeah.

it more universal.

Yeah. Yeah I'm

That'd be great because that would be a good

~ that'd be a good grab for this because I think right now it's probably just like

build me a skill that looks at feedback,

looking at A, B, C, and D. And I'm sure that might have been your day

one and you've probably fine-tuned it quite a bit since.

So you probably save a lot of people some time.

Yeah. I would advise people now with the Claude recorder skill is just

try it out and then give it to an agent and then see what see what it thinks.

I haven't done that personally,

but something on my bucket list.

~ I

I have conversations and then ~ and then build a skill based on that

as a starting point. To take a skill from somebody else and bring your own.

I

believe it works if you understand where to nuance and stuff and and do all that,

but just to put it in and rip is probably dangerous.

I have a a hypothesis that shared skills probably have a high churn rate because

people aren't doing the extra work to

Make it unique to them.

Got it.

I could be way off.

Who knows, right? Who knows? We'll we'll have to see.

And

Angelo, this has been a great conversation.

Thank you for for walking me through this.

How could people get in touch with you?

Yeah, find me on LinkedIn. ~ we can share my Notion page.

~ always happy to connect with folks,

especially on the RevGenius community.

I love what you're doing here and spreading the good word and there's

a lot of builders out there I think who are doing great things and it's it's awesome

that you're highlighting that.

Let's go.