Full Funnel as a System of Action with RevSure

0:13 Next up is Pinder Sang Dengra, founder and CEO of Rapture AI and who was the first one I think in the industry to deploy agents forums effectively and will showcase today fullfunnel system of action. Super excited for this one. Always love your sessions. How have you been, Defender, and what's new? >> I've been doing great. Thank you so much, Julia. Great to be here. Um, it's all about agents deployed for the enterprise, uh, where we come from. So, um, just to kick it off, hi everyone. I'm Deependur.

0:49 I'm the founder and CEO of Refshore. We are on this journey to build the only enterprisegrade revenue system of action for B2B um enterprise revenue teams uh across the marketing the sales the SDRBR and the customer success motion. So when we mean fullfunnel we mean across the end to end um GTA motion not just a marketing motion not just the sales motion not just prospecting and beyond or not just um you know hey for sales teams or for SDR teams etc. Today you know I'll be talking about how do you think about going from systems of record to systems of action.

1:35 Um we all know that for the past 100 years revenue was run by human beings. Um but the current G and uh GTM techstack was never built for agents. It was built for humans. How would you go from a GTM tech stack built for human beings, built for specific user personas, specific teams to a GTM text stack built for agents that can actually act right from anonymous visitor to close one revenue and beyond. And that's what Revshore really focuses on. What's happening today if you think about the past couple of decades is that there are number of point solutions. We have all heard about the GTM franken stack the text sprawl on GTM but essentially B2B GTM is broken across fragmented buyer journeys fragmented and disconnected tools messy data siloed teams each of the tools that were built in the past whether that the CRM system the marketing automation system the buyer intelligence tools the sales automation prospecting tools the paid ad tools ABM intelligence tools etc in and of themselves they're great tools but they only focus on specific parts of the GTM motion. They're only specific on uh serving specific needs of different user personas and teams and therefore they create their own systems of record.

2:58 Each of these systems of recovered is very deep in its own perspective and the part of the GTM funnel that it kind of addresses whether it's hey attracting visitors converting them into leads or accounts and nurturing them into marketing qualified then nurturing them into sales accepted CS qualified etc whatever whatever GTM motion you have whether it's a inbound outbound product marketing led leadbased accountbased PLG community-led um in fact the kind of customers that we work with, they all have hybrid GT emotions. I've not seen in the enterprise um a pure PLG company or a pure ABM company or a pure leadbased company every enterprise that has multi-reion, multi- channelannel, multi-segment, multi-product GTMotion, hybrid GTMotions are operating in parallel, right? And so what ends up happening is that even though we try to connect these tools through different integration technologies, these are more band-aids than solutions.

4:01 And in a world of agentic AI, this actually causes more problems than solutions because it creates more noise. It amplifies the noise. When an agent doesn't have full context, you actually give that agent a reason to go rogue. Because at in the enterprise level, if you're going to act on millions of contacts and and execute billions of actions at enterprise um scale and enterprise complexity, if you if the agent doesn't know the full context, if the SDR agent that you probably, you know, best of breed agent that you bought or an inbound qualification agent that you bought a best of breed agent from some vendor, right, does not have the full context of what's going on in the marketing motion or what's going on in the sales motion or whether that visitor actually came to came to me through a website uh expressed some intent or went to my trade show, my webinar, my field event, etc., right?

4:53 They're going to cause the same problems because they're going to act on incomplete data. And although you might automate a lot of outreach and qualification and campaign execution, they're going to end up creating the same siloed problems, you know, and lead stagnation. Um, not great pipeline quality. Why are my leads not converting? Why is my pipeline not progressing through the stages? those same problems are going to recur but this time actually double uh at double the scale than what was there earlier. So how do you take this existing broken B2B GTM text stack in the enterprise and make it agentic AI ready and convert these individual systems of record that have deep but partial data into systems of action for the enterprise. You might say, "Hey, you know what? We're going to rip out all of this tech stack and we're going to have some, you know, uh, one technology rule at all, AI native CRM or AI native marketing automation system." But that's not practical because heterogenity is, uh, is the fact of life in the enterprise not because of just the DTM text, but because the number of channels we work with enterprises and abovement market companies that have 50 channels.

6:09 Each of those channels are generating data and generating signals and generating interactions from their individual data sources. So that hetrogenity is going to say. So our perspective is to convert these systems of record into systems of action. One needs to think about a harmonized context layer. without a harmonized context layer and a unifying substrate that can unify the data intelligence across the disparate and fragmented GTM text stack which is going to continue to operate right in their specific parts of the DTM ocean. Your agents will not have fullfunnel context and therefore will not reach their full potential and you won't have a coordinated of system of action that will share the same context.

6:56 Every agent, imagine every agent, whether it's your prospecting agent, your campaign refinement agent, your deep funnel optimization agent, your deal intelligence, lead account research agent, nurturing agents, etc. having the same context of the lead account opportunity, the past interactions, the the movements, the decision traces, etc. And that's what our perspective is. You need to create a harmonized context layer as a unifying substrate on top of existing GTM text stack to convert your current systems of record to systems of action. Now, how does this context layer work? What the context layer does is ingest and harmonize the full BT business and GTM context as well as all of the events, interactions, decisions, GTM funnel stages, your GTM life cycle, etc. So imagine the millions of visitors, leads, contacts, opportunities, hundreds and thousands of campaigns that one might be running at the enterprise level across multiple CRM systems, multiple paid ad systems, ABM tools. We work with customers who have 65 systems. We work with customers on an average who have 20 plus systems across the GTM motion. And we are harmonizing all of that data, getting all of the business context, the ICP segmentation, the key buyer personas, the buying group definitions, uh the definitions around the campaigns, the channels, the products, the regions, but also getting all of the events interactions and decisions that are taking place, whether the decisions, first party interactions, second party interactions, online, offline interactions, product usage interactions, conversation calls, paid media interactions. Today you do find solutions that will focus on the prospecting to conversion which is just from okay hey I identify a lead I identify an account I do some research I start prospecting and then try to kind of take that through the uh generate a pipeline then convert that and you find solutions that are very good at the on the marketing side that are focused on hey let me get all of the interactions around paid media and maybe online offline but there is no solution today that can integrate end to And the reason I'm talking about this is because for you to achieve fullfunnel optimization of your buyer journeys, one needs to bring the context across sales and marketing. Because if one doesn't, then the same problems of the past, sales and marketing misalignment, leads not converting, leads stag stagnating, decreasing ROIs, pipeline quality not keeping up, win rates going down, less than 50% of sales rep meeting the CO that kota is going to continue to exist although we might keep automating more and more. So bringing all of this into one context layer, an AI reasoning engine that can orchestrate autonomous as well as coordinated agentic actions.

9:47 That is the missing substrate for most enterprises. Right? Most enterprises what they like to do is either have a rip and replace strategy. In the enterprise actually that doesn't work, right? It might work for SMBs. So if you're an SMB, you could try a rip and replace strategy with with one, you know, system that can try to do it all. But you really need to have a more practical approach where you're actually converting your current systems of record and your GTM text stack to become agent AI ready. Now, how does this look?

10:16 It looks it starts first by bringing all of the context across your fragmented GTM tech stack across this marketing motion, the SDR BDR motion, a motion. And I'm going to show this to you for real. and how this works harmonizing all of this context and we are big believers in eating our own dog food right and we do this at scale right we do this at enterprise level right where the complexity right can really uh be daunting but can be solved for so I'm going to switch to actually showing you how this works right here is um an example where we are essentially harmonizing context across the GTM text stack So we integrated um across the CRM system, the marketing automation systems, right? You could bring in activities from G2 reviews etc. Senoso the the gifting campaigns first party pixels the paid ad campaigns an dearizing visitors uh on the website fingerprinting tracking all of the anonymous visitors to first conversion and beyond Twitter Google search console uh bringing across your demo tools right so we use demo tools um ourselves right bringing data across that bringing the the Google Analytics data search interactions data etc bringing all of this we have over 22 data sources integrated.

11:43 Once the data sources get integrated, it's not just about the integration. It's about bringing this and harmonizing the data into a one common data model. Right? Let me just refresh the screen. Um right, bringing in this to one common data model that can then allow you to build the overall context graph. do the linkage resolutions, do entity resolutions, do all of lead to account to contact mappings, etc. Once you have that, what that sets the stage towards is your ability to bring together for every lead, account, and opportunity, bring together all of the interactions that are happening.

12:27 Right? So once you've harmonized the data for every lead and account and um just to show this to you all of the funnel activities campaign touches funnel moments across all of the different data sources right the unique timeline journey of interactions across the sales motion marketing motion SDRBD motion customer success motion for every lead for every account every lead has its own unique journey the details of those interactions right not just one um one set of interactions across online offline channels. Not just the interactions themselves, but all of the details of those interactions across all of the data sources across your GTM text stack being integrated. Further, being able to extract more signals, right, from conversations, from emails, being able to infer what are those buyer personas, being able to bring together all of the technographic interaction attributes, um the time traces of events, as well as all of the GTM funnel life cycle decisions that were taking place, bringing together all of the predictive pipeline propensities that based on the behavior of every lead based on its unique journey. What are the propensity?

13:43 So, extracting all of the signals at the lead level as well as extracting all of the signals even at the account level, right? Similarly, aggregating all of the interactions at the lead, account, opportunity, campaigns level. This builds a foundation of the context layer. Once you have the context layer which is unified across the GTM tech stack, then one can start building agents on top of it. Now your GTM text stack is harmonized. All of the systems of record are brought into one context layer and an enterprise context layer that appreciates the nuances of your business model of your lead life cycle definitions of your um of your operating calendars of your buyer personas of your specific attributes dimensions etc. Once that appreciates that then you are ready to run agents on top of that. And now the difference is these agents are not agents that are individual agents. This is a team of agents. Each agent so is built on the same context.

14:57 Every agent shares the same context. But now you can build a team of agents across your GTM tech stack, across the context that's embedded in each individual system because now it's brought together, right? So you can build agents for account research, you can build agents for lead scoring, you can drive agents for prospecting, you can drive agents for upfunnel, top of the funnel campaign refinement, reaching out to ICP visitors on your website, reaching out to people who have requested a demo. All of the signals are brought together, right? Each of those agents can get orchestrated and configured further. Now, this doesn't have to be the agentic orchestration layer doesn't have to be in one environment. It could be and you could use an agentic orchestration platform.

15:45 Um, it could be one of those hyperscaler agentic orchestration platforms. It could be claude, it could be from Google, chat GPT, etc. But the idea that made this impossible that made this team of agents possible based on the shared same context was your ability to bring build this unified substrate of context across your GTM tech stack because once you have this then you can build these agents you can expose these agents you can build agents on top of this right realtime agents schedule agents all of the whole agentic workflow can be orchestrated with the whole AI and intelligent logic within it but you can also expose agents into external environments. You can expose the context layer to for example claude where all of the context within your context layer is actually exposed as tools. So now any external agent can access all of these tools to be able to further drive agentic workflows on top. Right? And that's what helps you convert systems of record that you have. All of these systems of record that you have which are have their own deep but partial data. Each of these tools is great for its own purpose. Some tools different tools are doing different uh aspects of your GTM motion. But none of these tools can drive coordinated agentic actions across your full funnel. And so once you have that then you have a real system of action with the ability to deploy teams of agents. You are harmonizing context from your GTM systems from external data. Harmonizing that context into a unified substrate.

17:27 You further overlaying AI learning engines that can help learn signals at scale right because some signals have to be learned at scale across the GTM motion. And then that drives a system of action where you can deploy agents across your funnel as a team of agents and a team of coordinated agents. So that's what I had to show today. Yeah, >> super impressive. Always love your sessions for the reality of what's actually going on in enterprise deployments. Um I'm curious from the perspective of um you know your experience actually operationalizing this shift in an enterprise um how do you yeah >> sorry go ahead >> yeah uh how do you see the ultimate um you know futuristic stack >> um our perspective is that there'll be there'll be a bifocation in the stack there will be companies that will try to uh rewrite the architectures across the GTM tech stack and those be and those will be companies that uh either they will adopt some vendor who's trying to kind of do the complete end toend GTM motion orchestration because acting on incomplete context is no longer an option because agents are not limited by us human beings right CRM was was created to help sales teams log deals marketing automation was created to help marketing teams execute campaigns and manage their leads, accounts and prospects, right? Conversation intelligence tools were was was built to help sales teams review calls and their leadership review calls. Now, each of these tools was built with a persona in mind with a specific team member uh uh with a specific team in mind and for specific stage of the GTM motion, right?

19:19 But that ended up creating its own problems because now on an average we have 23 tools across the GTM tech stack. It created problems of silos and all of the misalignment problems that revenue leakage, pipeline leakage etc. But agents we have the unique opportunity that agents are not limited by the by the individual teams that they are part of right they can adopt the persona of the team but then they can go across to understand context right so our perspective is there'll be companies who are actually going to try to rip and replace but that will mostly work for SMBs and midm markets what we are seeing in the enterprise the architectural approach is to create an ecosystem of technologies but create one context layer so while enterp enterprises will use multiple if you think about the three layers of uh agentic AI for for revenue the there's the foundation models and all foundation models are converging every 12 months right that's at the bottom layer then at the top layer there's the agentic execution layer and the agentic orchestration layer and the companies like claude and other tools that are helping orchestrate that and that layer also getting fragmented and also getting commoditized so people have different options right the only layer that is remaining and that is not yet actually built out and companies are trying to do this internally at least enterprises are trying to do it internally right we had a customer for example who was trying to solve this problem for 10 months right with a mix of team data engineering data ops marketing ops data science analytics digital marketing revops right coming together to try to bring the context of the full funnel and you know It's really tough to build this because you need to have a very dedicated focus. It requires you to understand data harmonization, entity resolution at lead account, option a campaign across the multiple channels and the multiple systems. You need to do semantic data standardization, linking of leads to accounts or campaigns to campaign members, fingerprinting, all of that needs to come together, right? And so the the primary architectural approach that we see in the enterprise is to try to build a context layer right in between the GTM tech stack because it's not going to be possible to replace for enterprises who have heavy investments in workflows and processes that are already part of your current GTM tech stack right you can't just rip and replace that GTM text that GTM text stack what you have to do is to build a layer on top of that that enables that GTM text stack to act as a system of action Right. Uh while every every tool will continue to build agents on top of it, you need a coordinated substrate to help you build systems of action. That's our thought process. Yeah, that's what we are seeing.

22:01 >> Beautiful. Community is asking about the governance element. Can you share your perspective here? >> Uh so governance are there are two aspects of governance. one is at the data governance which is all to do with um uh privacy uh and you know and confidentiality etc. So right there's a lot of focus in the enterprise around GDPR etc. And we like we have the required guardrails we can do PII reduction encryption of PII data right um using only nonsensitive PII data etc. That's one aspect on the governance side. And then there's the AI governance because to be able to enable these uh these systems of action. What one has to do is to um is to make sure that the data that if you if even if you're sending data for for agents to execute, summarize, recommend, take next next best actions and recursively learn, you have to support PI reduction. For example, we do we support that. But also you have to kind of start thinking about um you know not using not necessarily using open- source models uh using models that are vetted within the enterprise. You have to think about no prompt caching, no prompt logging, no data about that goes into the LLM model for training, right? No, no, no fine-tuning of the LLM models or the training of the LLM models based on the context that you feed that LLM models, right? So that you have to take care of those aspects, but you want to capture the agentic decision traces, right? So for systems of action to work, you need one central AI brain because you need to capture the decision traces and the AI reasoning traces back into your context layer. For example, at Refshore, right, when we run our agents, our system of agents, one of the things that feeding our campaign touch points are what are the agents that are acting on my leads and account. So we bring back our agents information and the decisions that the agents are taking and acting on into our own context layer right and that's a very important thing because you know you will need that one brain to coordinate these systems of action. So hope I answered the question both on the uh both on the governance on data and AI side but also the importance to kind of being able to capture the agent decision tracing back into your context layer while protecting that there's no prompt caching prompt logging and there's PI reduction when information is sent to the to the LLM so that the LM can reason and take action uh as part of the agent reasoning logic. Yeah%% and community is also asking about uh the integration or you know how do you work with CRM and data warehouses.

24:46 >> Yeah. So we support a whole list of integrations. So if you go back so this is the data graph but if you go back to the data sources we support integrations across sales the CRM systems like Salesforce uh Dynamics um par dot marketing automation system like par dot marketer etc because we work in the enterprise customers very often have multiple CRM systems multiple marketing automation systems for different businesses for different regions it coming through acquisitions we are probably the only company who who is able to harmonize context across multiple CRM systems, multiple marketing automation systems uh from data lake environments um across like whether that's data bricks, snowflake um uh bigquery etc red shift right um and we essentially are you know the ability because at the enterprise you could even have multiple data lake environments where context is coming from right so we support the whole gamut of enterprise the the text stack that is within the enterprise Yeah, >> can you speak to some of the customer stories that excite you the most?

25:54 Somebody who actually deployed this transitions to systems of action. >> Yeah, so we we work with customers like Zcaler uh uh normal security etc. And with all of these companies, right, uh the starting point is to build the context layer. So there are companies are um there where we've integrated over 20 systems, billions of touch points, millions of leads accounts, hundreds and thousands of campaigns, right? And then what happens when when we uh when we integrate the full context layer then that unlocks multiple AI um use cases on top whether those use cases are to kind of help you understand the impact of your marketing campaigns and marketing channels down to pipeline and revenue at the bottom of the funnel right uh to revenue um and then you know whether that is for personalized outreach which uses the full context of the interactions with that's coming from visitors leads ETA etc. Right? And the results are improvement in conversions, improvement in GTM ROI, improvement obviously in productivity. Right? Often we see like 40% 50% improvement in productivity across the board. But also there there there have been teams internal teams who are trying to build this and it takes a long long long time to build this. So we also repurpose the efforts of the internal teams on more value added activities while we take care of the context layer. Yeah.

27:23 And lastly, what's on the road map? What are you allowed to share with uh what excites you the most? >> Yeah, so we are we are building out the full range of AI uh agents across the marketing SD or a motion. We're building out the primitives which can be exposed as MCP for command CLI support and API. We already have MC support, CLI support, but we for every agentic action system of action whether that's updating your CRM, executing campaigns, orchestrating journeys, executing next best action, we already support a lot, but our vision is to kind of support the end to end GTM motion uh across that and on the context layer. We continue deepening our context layer with more integrations, but more importantly, as we deploy more to enterprises, we are getting just better and better in the whole context graph.

28:13 How do you build context graphs, right? And what information elements you capture in the context graph. So that's a huge area of focus for us. Yeah. >> Excited to be part of that journey. And where should our community go? Um your website, contact you personally. What's the best? >> Yes, absolutely. You obviously can to learn more about Ref Show. You can go to refshow.ai to contact me, can send me a direct email at deeporrefshaw.ai. Would love to chat about how we've built this and what our plans are.