Everyone Has AI, Context is Your Only Edge

Julia Nimchinski:

And… welcome to the show! Zach Vitibor, co-founder and CEO of Octave. What a treat! How are you doing?

Zach Vidibor:

It’s the latest and greatest. Excited to be here.

Julia Nimchinski:

Excited to see your attack in action!

Zach Vidibor:

Is that the drum roll, please? Let’s get started?

Julia Nimchinski:

Yeah, heard a lot of great things from your VCs.

Zach Vidibor:

Cool.

Julia Nimchinski:

Well, let’s…

Zach Vidibor:

let’s dive in. We’ll get in the app very quickly, but I’m gonna show a few slides, just because we’re talking about context, which can mean a million different things, so I want to tell you about Octave, how we see the world. introduce myself very quickly.

So, one of the co-founders here, before starting Octave, I was in the trenches of go-to-market for many, many years myself. So, been a sales rep, been on the phone, been in BD roles, growth roles, for over 15 years, so trying to bring a lot of, I’d say, the, hard-won learnings from on the field into what we’re building at Octave.

So let me… Share my screen here. And… Hopefully, it’s coming through. Thumbs up? Okay, cool. So, we’re gonna talk about context engineering and how we build smarter go-to-market agents. So, what the hell does context mean when we’re talking about it?

So, I think the best way to think about this, how the world has really changed, is we’ve moved out of the prompt era, the build better automations, right? Like, how can you build a task list in GPT or some column in an orchestration tool to push something along, right?

That’s the difference between, like, giving someone, you know, directions to the store, or can I give them a map of the city? Can I let this agent go find its own way to a desired outcome? Whether that be… go to the movies, or go to the supermarket, right?

So, that’s kind of the difference of how we think about where we’re in now, is like, how do we give agencies maps, right? And what’s tricky in go-to-market is things get… complex extremely quickly.

So if you just think about a company, let’s say we have 2 products, we have 8 different personas, we might want a slightly different message or approach for a different strategy. We have a few different segments, right? Hey, we sell to healthcare, and we sell to SaaS companies, and we could sell to financial services.

Oh, wait, actually, financial services, we sell to banks different than insurance companies, right? Like, you can imagine how this grows very quickly in your businesses, and then, oh, we have… We have a few different motions, right? We have a net new business motion, we have an upsell motion, expansion motion, a cross-sell, right?

There’s, like, all these different ways it changes, and then, of course, oh, we have a bunch of playbooks based on all these different signals we see. Are you hiring? Are you growing? Did you… expand to a new country, right?

There could be all of these, and like, those all create added dimensions of complexity, and your agents are not going to know this out of the box. Clawed will get you pretty close, but it’s gonna get you 80% there. All your competitors are at 80%, right? Everybody has Claude.

If you want to get to 100%, if you want to be a leader, you now need to give them more context. So that’s what we’re talking about, is how do we give them this map? And what we talk about is there are two types of context. There is type one. This is all your data. This lives in your CRM, in your data warehouse. It is about what happened, right?

It is all your enrichments, it’s about the world. Type 2 is what we’re gonna focus on today. These are your opinions, this is your strategy. This is when we see these types of patterns in our customers and these types of problems, this is how we win, right?

And that needs to be captured somewhere, otherwise your agents are going to take a guess, sometimes they’ll be right, many times they’ll be wrong, many times they’ll just be off the mark enough that it’s not gonna be what you want, right? So, this is where we sit.

So you can think of, at the, kind of, bedrock foundation, you have your systems of record. We’re not crazy, we’re not saying, replace your CRM, replace your call recorder, replace your data warehouse, all that’s gonna stay there. What you’re missing is this system of context, right?

People are… kind of doing this today in maybe Notion or Google Drive, or they started to put Markdown files in a Git repo, but it gets… turns into a mess very quickly. And then you have your systems of action on top, and then we are feeding this intelligence wherever your team lives, wherever your agents live, we don’t care.

You want to be in Clay, you want to be in Claude, you want to be in GPT, you want to build Slack agents, they all need to be smart, we can make them all great.

And this is kind of what we’re talking about, right, is how do we be that layer of judgment that takes in all of these signals, all of the what-happened data, plus your opinions on how to best win?

based on learning from, obviously, all your calls, all your deals, what’s happening, how do you win when you talk to a manufacturing company versus a HR tech SaaS company, right? Octave is gonna help learn all that for you, feed it into all your tools so you can do all types of things across the funnel, right?

So… Very quickly, wanted to set the table, and now let’s jump in and see Octave. So, we’re in a kind of demo instance here. This, by the way, think of Octave as, like, Stripe, in that This is a headless platform, your reps don’t log in here, this is for a special class of business user.

Again, they get the benefits of Octave and this intelligence wherever you want them to live. So in here. This is where the system will start helping you define your, kind of, key definitions, you’ll see in a second, around what products you sell, your personas, etc, things like that. motion playbooks, right?

This is how… how do we take our strategy and cover our market? How do we actually use this in action? And then we have some agents out of the box, but more often than not, it is how do we connect all your tools, right? So we are wrapped in APIs, MCP, we can plug this thing in and make it work wherever you want. So… The library real quick.

is kind of where your core spinal column, right? This is kind of, like, where our company brain lives. So you can kind of see we’re building this context graph for you, so how do all our products and use cases and playbooks and competitors interrelate, right? You can drop a new node in the brain very easily.

You don’t need documents, you can do it with just a chat, anywhere, anytime, and we know how that relates to everything in your go-to-market. So you can see, you can have products and services and solutions. Octav actually understands the differences between those things. You can have your target ICPs, right?

Segments and personas, and then we can start mapping all of these things around the use cases and proof points and reference customers, right? And we know how all of these things interrelate. And the cool part is these aren’t just flat, one-dimensional definitions, right, about how we talk.

They also help agents understand, how do we qualify, right? people can come in here, wait different questions, right? Help the agents understand, how do we think about this thing, not just how do we talk to them, right? Which most solutions are just thinking about, what are the words we say?

This is more about how do we identify, how do we strategize, right? So, this is the library. I’ll show you now how that translates to a strategy. So here’s where we have motions. So you can have a bunch of different motions around net new business, upsell, cross-sell, etc. I’ll show you what this starts to look like.

And again, this is for auditability. None of our customers live in here, right? You’ll see it’s overwhelming, we try and make it pretty, but this is really a graph for agents to walk, right?

So agents can find, hey, when we’re in net new business, and we’re talking to one of these types of companies, and we’re talking to a go-to-market engineer, what’s our strategy?

What is cool for our customers to see is, obviously, we’re tracking your activity and things like that, so we can start showing you interesting heat maps of where you’re winning, where you’re losing, where you have the most activity based on all these dimensions. But what agents are able to do is we give them the tools to walk this graph.

Find this strategy, understands what’s happening. Do we have evidence and validated kind of proof around all these characteristics? What are we learning? What are we changing? You can see here, the agents bringing in key language that’s being used on calls, pain points, etc, and then agents can pick this up and run with it, right?

So… just to land the point here, you can see what we’re doing on the backend. We’re turning this all into code, so your agents can pick this stuff up and read it, right? That’s the magic here. People don’t live in this, this is for your agents to read off of, so you can build smarter systems faster. So… Lots of cool stuff, in here.

So these are motions. We do have, or actually, let me show you playbooks first. So the cool thing here is you can layer playbooks on top of this. So you can imagine, you might want a strategy for how you go after business in the UK, or for a specific type of trigger, but you can very quickly layer on different strategies around this, right?

So it could be some thematic-based… trend you want to speak to, hey, you’re probably adopting Claude, and you want to develop a strategy around that, or some specific trigger. geo-based, you want to go after a competitor, you can layer playbooks, over this strategy, and again, tell your agents how you want to go to market.

Is one single point of control. Make them say whatever you want them to say. The last thing I’ll show real quick, is agents. So we do have some agents out of the box that we know for very common, workflows. Things like sequencing, enrichments. qualification, call prep, these things, again, we’re not here in the agent selling business.

These are, we give them to you for quick time of value, and you can see you might plug them into places like Clay, where we have an integration. and you want to qualify, leads, companies, people based on that intelligence. You can see here, we have a great, solution out of the box. You can use our integration with Clay.

And you’ll see these agents will do a lot of work, scoring across your products and segments. things like this, and you can see they go into deep, deep, qualification here. They’ll come up with 20 different questions they can ask and figure out and generate confidence, and so you can give this stuff immediate use in the real world, right?

We want to give you context that you can apply to your workflows right away. And then, of course, got lots of APIs for all the builders in the room. You can create all this stuff, again, headless platform, at the end of the day. So… Very quick, I know we don’t have, too much time here, and we’re gonna do a Q&A as well, so I will, pause there.

Julia Nimchinski:

Zach, this is phenomenal. One of the best, Zach. solutions I’ve seen in a while. Do you have, like, direct competitors? Just curious.

Zach Vidibor:

It’s a good con- it’s a good question. There is not a, like… pure play, oh, Octave or this? Generally, our competition is teams are starting to home roll something, right? They are building with Claude, with Claude code.

You know, again, as we’ve moved out of prompt era into context era, most people, they find themselves in a build mode without even planning it, right?

You just start building agents, and you go, oh, like, they need a little bit more, so I start a markdown file, and oh, they need a little bit more, stuff them all in a repo, and then… then companies start running into, wait, like, we’re gonna use this across the… oh, wait.

marketing and PMM need to update this stuff too, so then you start, like, alright, it’s not gonna be in a repo, we start needing to wrap it in a UI. That’s usually where we find customers, and then they’re baking off with Us versus something they’ve built.

Julia Nimchinski:

Amazing. Let’s address a couple of questions from the audience here. So, first one… What happens when the market changes and RICP shifts? How quickly does Octave adapt?

Zach Vidibor:

I mean, very quickly. So, I’ll show you by default, here, if we go back into here, we basically set a default cadence on these motions. This is a demo org, so it’s slightly different in here, but basically, we set these on a weekly cadence, right, where they’re gonna automatically refresh. Basically, they compact these learnings every week.

and run it automatically, you can force run an update at any time, too. So… Generally, like, what we’ve found with our customers is, you know, you don’t want the self-driving car, like, jerking the wheel every night on your strategy, but you could change it Anytime you want, instantly.

Julia Nimchinski:

Another question here, does it overlap with AirOps?

Zach Vidibor:

Yes, so we have lots of joint customers with AirOps, they use our MCP together, so they take octave context, feed it into AirOps. Great, great combination, we love Air Ops.

Julia Nimchinski:

Next question… It’s an interesting one. Can Oxford identify when we are choosing deals we should walk away from?

Zach Vidibor:

Yes, so I would say we will help, again, like, one of these things, that I was showing here in, like, Clay is, like, our qualification, right?

So, we can run every company through, and these are basically… these are trying to do these very… ICP qualitative-based scoring, like, where we would say, like, low fit, and we can identify all your leads, all your accounts, and we should tell you, bad fit, based on… based on what we know about how you sell, where people find value, where the likely pains are, no go.

Julia Nimchinski:

Zach, can you speak about the measurable business outcomes? Any improvements? Folks are asking, pipeline quality, win rates, deal velocity?

Zach Vidibor:

Yeah, deal velocity, definitely a big one with producing all the assets and follow-ups and custom call decks and, you know, stuff like that. I’d say the biggest, like, you know, probably for the group here to kind of anchor around, you know, like, where our customers are seeing true, meaningful outcomes is basically.

you know, pipeline per head, so driving a lot more efficiency throughout the team, right? They are seeing, you know, kind of… you know, pipeline per marketing dollar, like, high-level metrics are where we’re trying to really ground ourselves and, you know, drive business value. And then… teams are, you know, I’d say, you know, on top of funnel.

If you look at, like, SDR orgs, a lot more production without additional headcount. You know, things like that would be, like, where we would point to, some strong success.

Julia Nimchinski:

And a couple of questions here. What data sources create the biggest improvement in accuracy?

Zach Vidibor:

Mmm, great question. I mean, call data, I think, is the best, right? Like, these are gonna be people talking to your business about, what you do, so the richness of that data. what we do there is slightly interesting, right?

Because you can imagine Octave at any given time, like, we have this kind of point of view about all of your segments, all of your personas, like, uniquely what they care about, right? So, we can do some really interesting things, where we’re showing… actually, I meant to go back to the library.

So… what we’re doing here is, like, we can really measure against all of these different characteristics across your use cases, map to your, segments you sell into, right? All of these things of, like, what is the lift rate associated with all of these different things, right?

And so, this is just to show you what we’re really good at is inspecting your calls, because we have kind of a spine and know what you think is true. We can figure out, hey, are our reps and the team saying things that aren’t working, leading to deals slipping, and hey, like, we have a coaching problem?

are people saying new things that customers are being attracted to that’s not in our plan yet? Let’s pull those in. So, we work a little bit differently and better, because we always kind of have this point of view. It’s not just asking the LLM, what should I do? What are the pains, right? Like, it’ll give you something all the time.

Octave works really well because it’s grounded in a point of view every day.

Julia Nimchinski:

And last question, then we’re transitioning to Gainside. How long does it take to build up the context for the agent to go above and beyond the basic training, and what’s the ramp-up period before we can see better results with customer-facing activity and seeing real revenue results month to month?

Zach Vidibor:

I would say we get our customers to 80% in a matter of days. Please, book time with us, we’ll show you, we’ll build a brain for you for free, so you can try it out. And then I’d say the answer, it depends, is really based on how much velocity is in your sales motion, right?

If you are doing large deals, long sales cycles, it can take longer for the system to learn from those deals, because they move slower. If you’re much more transactional, and like, we have a lot of calls and deals to learn from, it can be… days and weeks to, like, the brain is fully tuned up. So, depends on the kind of velocity of the motion.

Julia Nimchinski:

Thank you so much, Zach.

Zach Vidibor:

Of course.

Julia Nimchinski:

Super unique tack, and yeah, we’ll share in the community.