Your Moat Is Your Data Model — Mike Phipps, Gates Foundation

Yes. So my talk today is about the title Yes. So my talk today is about the title your data models remote. We have a your data models remote. We have a your data models remote. We have a enterprisewide platform that we had just enterprisewide platform that we had just enterprisewide platform that we had just rolled out here this past month. And so rolled out here this past month. And so rolled out here this past month. And so I'll go into details on this. I'll give I'll go into details on this. I'll give I'll go into details on this. I'll give you some hopefully some practical you some hopefully some practical you some hopefully some practical lessons here and why we made decisions lessons here and why we made decisions lessons here and why we made decisions we made for this uh how how you could we made for this uh how how you could we made for this uh how how you could picture your processes within a similar picture your processes within a similar picture your processes within a similar type framework. type framework. type framework. So first just a quick introduction. So first just a quick introduction. So first just a quick introduction. So this gets into the the title here the So this gets into the the title here the So this gets into the the title here the talk and the the framing of you know talk and the the framing of you know talk and the the framing of you know what I hope you take from this but with what I hope you take from this but with what I hope you take from this but with AI moving very fast at the frontier what AI moving very fast at the frontier what AI moving very fast at the frontier what what's defensible you know you can move what's defensible you know you can move what's defensible you know you can move you can build things very quickly with you can build things very quickly with you can build things very quickly with clogged code um but once you push things clogged code um but once you push things clogged code um but once you push things to production there's constraints you to production there's constraints you to production there's constraints you find how much of your your uh deployed find how much of your your uh deployed find how much of your your uh deployed stack do you want to actually own you stack do you want to actually own you stack do you want to actually own you know there's monitoring there's upkeep know there's monitoring there's upkeep know there's monitoring there's upkeep there's uh you know people building there's uh you know people building there's uh you know people building dependencies off your stack that you dependencies off your stack that you dependencies off your stack that you have to be prepared to handle. How much have to be prepared to handle. How much have to be prepared to handle. How much uh appetite do users have for uh appetite do users have for uh appetite do users have for decentralized access? This gets into decentralized access? This gets into decentralized access? This gets into I'll show you what we built, but you I'll show you what we built, but you I'll show you what we built, but you know what's what's the access point for know what's what's the access point for know what's what's the access point for users? You know, is it another chat app?

users? You know, is it another chat app? users? You know, is it another chat app? Is it clawed? Is it chat GPT? Uh is it Is it clawed? Is it chat GPT? Uh is it Is it clawed? Is it chat GPT? Uh is it something else? What's your product something else? What's your product something else? What's your product differentiation from from those differentiation from from those differentiation from from those different SAS products? And so our team different SAS products? And so our team different SAS products? And so our team then you know with this context in mind then you know with this context in mind then you know with this context in mind you know thought through here you know you know thought through here you know you know thought through here you know what's our skill set here what's our what's our skill set here what's our what's our skill set here what's our competitive advantage in this competitive advantage in this competitive advantage in this environment and this is what I you environment and this is what I you environment and this is what I you really hope that you you take from this really hope that you you take from this really hope that you you take from this talk and you picture yourself in this talk and you picture yourself in this talk and you picture yourself in this but our moat here was our understanding but our moat here was our understanding but our moat here was our understanding of our internal processes the tacet of our internal processes the tacet of our internal processes the tacet knowledge that you need to to run knowledge that you need to to run knowledge that you need to to run successful AI and this is true I think successful AI and this is true I think successful AI and this is true I think no matter how good AI gets how good no matter how good AI gets how good no matter how good AI gets how good models get and new releases that models get and new releases that models get and new releases that different companies put out when when different companies put out when when different companies put out when when Mythos comes out or when there's a new Mythos comes out or when there's a new Mythos comes out or when there's a new app from Claude. Yeah, I'm not I'm not app from Claude. Yeah, I'm not I'm not app from Claude. Yeah, I'm not I'm not worried because the part that we've worried because the part that we've worried because the part that we've built is the defensible part that that's built is the defensible part that that's built is the defensible part that that's that's durable. that's durable. that's durable. So, these are the I I'll tell you what So, these are the I I'll tell you what So, these are the I I'll tell you what this means here in more detail, but this means here in more detail, but this means here in more detail, but these are the the processes tacet these are the the processes tacet these are the the processes tacet knowledge that we've modeled into what knowledge that we've modeled into what knowledge that we've modeled into what we call the strategic intelligence we call the strategic intelligence we call the strategic intelligence platform or SIP. and it rolled out here platform or SIP. and it rolled out here platform or SIP. and it rolled out here this past month in production for this past month in production for this past month in production for enterprise use across uh the Gates enterprise use across uh the Gates enterprise use across uh the Gates Foundation. So about 4,000 people.

Foundation. So about 4,000 people. Foundation. So about 4,000 people. So first I know this is an engineering So first I know this is an engineering So first I know this is an engineering talk but the the the scope of this talk talk but the the the scope of this talk talk but the the the scope of this talk gets into data modeling internal gets into data modeling internal gets into data modeling internal operations processes and so I want to operations processes and so I want to operations processes and so I want to give very quick background here over give very quick background here over give very quick background here over what the Gates Foundation does because what the Gates Foundation does because what the Gates Foundation does because then this is what we're we're modeling. then this is what we're we're modeling. then this is what we're we're modeling. So, as you're probably familiar, the the So, as you're probably familiar, the the So, as you're probably familiar, the the Gates Foundation has a has a very wide Gates Foundation has a has a very wide Gates Foundation has a has a very wide scope and it's a very ambitious work scope and it's a very ambitious work scope and it's a very ambitious work that we've been doing for the past 25 that we've been doing for the past 25 that we've been doing for the past 25 plus years. And there's all kinds of, plus years. And there's all kinds of, plus years. And there's all kinds of, you know, broad initiatives that we're you know, broad initiatives that we're you know, broad initiatives that we're doing, you know, whether it's for uh doing, you know, whether it's for uh doing, you know, whether it's for uh child mortality, whether it's for child mortality, whether it's for child mortality, whether it's for nutrition, agriculture, uh education. nutrition, agriculture, uh education. nutrition, agriculture, uh education. And these are kind of broadly the the And these are kind of broadly the the And these are kind of broadly the the different buckets that these different different buckets that these different different buckets that these different initiatives fit fit into. creating initiatives fit fit into. creating initiatives fit fit into. creating market incentives, spurring an market incentives, spurring an market incentives, spurring an innovation, collaboration between public innovation, collaboration between public innovation, collaboration between public and private sectors. And then the fourth and private sectors. And then the fourth and private sectors. And then the fourth one here kind of gets into the the lens one here kind of gets into the the lens one here kind of gets into the the lens that we're building here. You know, high that we're building here. You know, high that we're building here. You know, high quality data trying to derive datadriven quality data trying to derive datadriven quality data trying to derive datadriven insights from the from the actual insights from the from the actual insights from the from the actual investments, the the grants that we've investments, the the grants that we've investments, the the grants that we've put out. And over 25 years, there's a put out. And over 25 years, there's a put out. And over 25 years, there's a ton of structure. There's a ton of data ton of structure. There's a ton of data ton of structure. There's a ton of data that's developed. And trying to extract that's developed. And trying to extract that's developed. And trying to extract those insights at scale is difficult.

those insights at scale is difficult. those insights at scale is difficult. And that's what we're trying to solve. And that's what we're trying to solve. And that's what we're trying to solve. So this slide here is a snapshot of the So this slide here is a snapshot of the So this slide here is a snapshot of the of some of the different uh of the work of some of the different uh of the work of some of the different uh of the work that went out in 2023 within the that went out in 2023 within the that went out in 2023 within the foundation. This gives you an idea. I foundation. This gives you an idea. I foundation. This gives you an idea. I just put this here to to show some of just put this here to to show some of just put this here to to show some of the structured the structure that we the structured the structure that we the structured the structure that we have that we're working across. So you have that we're working across. So you have that we're working across. So you have over 2,000 grants in one year. Many have over 2,000 grants in one year. Many have over 2,000 grants in one year. Many of these are 5 million plus uh many 100 of these are 5 million plus uh many 100 of these are 5 million plus uh many 100 plus countries that uh that that are plus countries that uh that that are plus countries that uh that that are targeted with these grants uh alumni. So targeted with these grants uh alumni. So targeted with these grants uh alumni. So there's 4,000 different employees of the there's 4,000 different employees of the there's 4,000 different employees of the foundation. Um you know many different foundation. Um you know many different foundation. Um you know many different strategies within the foundation the US strategies within the foundation the US strategies within the foundation the US within the US across almost all the within the US across almost all the within the US across almost all the states grantees the total annual states grantees the total annual states grantees the total annual dispersement over 7 billion dollars. dispersement over 7 billion dollars. dispersement over 7 billion dollars. And so this gives you some idea of And so this gives you some idea of And so this gives you some idea of structure that we're we're working with. structure that we're we're working with. structure that we're we're working with. And this one just finally here when I And this one just finally here when I And this one just finally here when I show the data model this will make more show the data model this will make more show the data model this will make more sense. But we have different divisions sense. But we have different divisions sense. But we have different divisions that that funding goes out through that that funding goes out through that that funding goes out through different divisions. And so this breaks different divisions. And so this breaks different divisions. And so this breaks down some of those divisions. So you can down some of those divisions. So you can down some of those divisions. So you can see different priorities and it'll make see different priorities and it'll make see different priorities and it'll make more sense in a second here. But global more sense in a second here. But global more sense in a second here. But global development, global health, uh gender development, global health, uh gender development, global health, uh gender equality, USP are just a sample of the equality, USP are just a sample of the equality, USP are just a sample of the different divisions.

different divisions. different divisions. Okay. So the the fun stuff here now I Okay. So the the fun stuff here now I Okay. So the the fun stuff here now I hope the uh strategic intelligence hope the uh strategic intelligence hope the uh strategic intelligence platform so platform so platform so in a in a in a nutshell here structuring in a in a in a nutshell here structuring in a in a in a nutshell here structuring operational data for agentic retrieval. operational data for agentic retrieval. operational data for agentic retrieval. So we're building a knowledge graph with So we're building a knowledge graph with So we're building a knowledge graph with the idea of the agent consumer and here is an endto-end look of what and here is an endto-end look of what this looks like. So we we have different this looks like. So we we have different this looks like. So we we have different systems of record structured systems of record structured systems of record structured unstructured unstructured unstructured these have been siloed traditionally these have been siloed traditionally these have been siloed traditionally the so part of our team here the work the so part of our team here the work the so part of our team here the work has been to create what's essentially a has been to create what's essentially a has been to create what's essentially a data lakehouse putting everything under data lakehouse putting everything under data lakehouse putting everything under one roof. This is our internal one roof. This is our internal one roof. This is our internal enterprisewide data. It's also different enterprisewide data. It's also different enterprisewide data. It's also different different programmatic data that are different programmatic data that are different programmatic data that are uh outputs of different investments. uh outputs of different investments. uh outputs of different investments. Once it's there it's easy for us to Once it's there it's easy for us to Once it's there it's easy for us to consume. So we have a data curation consume. So we have a data curation consume. So we have a data curation layer that does different processing to layer that does different processing to layer that does different processing to it and then finally SIP here at the end it and then finally SIP here at the end it and then finally SIP here at the end with aentic chat agentic workflow as the with aentic chat agentic workflow as the with aentic chat agentic workflow as the as the the UX you how users are as the the UX you how users are as the the UX you how users are consuming our platform and so it's a consuming our platform and so it's a consuming our platform and so it's a cross system semantic graph layer that cross system semantic graph layer that cross system semantic graph layer that agents can res across agents can res across agents can res across okay uh so some of this I'll try to okay uh so some of this I'll try to okay uh so some of this I'll try to speed through here just for the sake of speed through here just for the sake of speed through here just for the sake of time but this one is critical time but this one is critical time but this one is critical when you're dealing with systems of when you're dealing with systems of when you're dealing with systems of record with lots of complexity record with lots of complexity record with lots of complexity engagement ment is critical. This is engagement ment is critical. This is engagement ment is critical. This is something that we've we've found here something that we've we've found here something that we've we've found here repeatedly. We have to engage data repeatedly. We have to engage data repeatedly. We have to engage data owners to understand, you know, this owners to understand, you know, this owners to understand, you know, this tacet knowledge we're trying to to tacet knowledge we're trying to to tacet knowledge we're trying to to model. What's the full meaning of model. What's the full meaning of model. What's the full meaning of different fields, the structure of the

different fields, the structure of the different fields, the structure of the data set, how do we join things data set, how do we join things data set, how do we join things together? How do we uh understand together? How do we uh understand together? How do we uh understand limitations, systematics of the data, limitations, systematics of the data, limitations, systematics of the data, safeguards, security trimmings, uh safeguards, security trimmings, uh safeguards, security trimmings, uh reporting conventions? You know, it's reporting conventions? You know, it's reporting conventions? You know, it's not enough just to answer it a question not enough just to answer it a question not enough just to answer it a question a certain way. You have to answer it the a certain way. You have to answer it the a certain way. You have to answer it the way that it's been answered in the past. way that it's been answered in the past. way that it's been answered in the past. And so, this is the comes back to the And so, this is the comes back to the And so, this is the comes back to the moat here. This is the procedural moat here. This is the procedural moat here. This is the procedural understanding tacet knowledge that AI understanding tacet knowledge that AI understanding tacet knowledge that AI needs and it's yeah it's the part that needs and it's yeah it's the part that needs and it's yeah it's the part that we that we own that's you know that's we that we own that's you know that's we that we own that's you know that's ours that um and that's what we're ours that um and that's what we're ours that um and that's what we're modeling here. modeling here. modeling here. Okay. So going back here just very Okay. So going back here just very Okay. So going back here just very quickly for this one this is the a quickly for this one this is the a quickly for this one this is the a snapshot here of different data curation snapshot here of different data curation snapshot here of different data curation considerations that we're that go into considerations that we're that go into considerations that we're that go into this pipeline. So you have [sighs] for this pipeline. So you have [sighs] for this pipeline. So you have [sighs] for different data sets whether it's different data sets whether it's different data sets whether it's structured unstructured there's structured unstructured there's structured unstructured there's different pre-processing filtering different pre-processing filtering different pre-processing filtering dduplication there's an order to dduplication there's an order to dduplication there's an order to different documents there can be uh different documents there can be uh different documents there can be uh inconsistencies across documents those inconsistencies across documents those inconsistencies across documents those need to be uh handled up front there's need to be uh handled up front there's need to be uh handled up front there's extraction so structured field extraction so structured field extraction so structured field extraction semantic chunking for extraction semantic chunking for extraction semantic chunking for unstructured documents if you have unstructured documents if you have unstructured documents if you have figures you need to convert this into figures you need to convert this into figures you need to convert this into text in some way so you can do retrieval text in some way so you can do retrieval text in some way so you can do retrieval across this uh various forms of tagging across this uh various forms of tagging across this uh various forms of tagging that these can form connections in your that these can form connections in your that these can form connections in your graph structure extra metadata that you graph structure extra metadata that you graph structure extra metadata that you create during this pipeline. Then that create during this pipeline. Then that create during this pipeline. Then that becomes different properties in your becomes different properties in your becomes different properties in your graph. And then the third bucket here, graph. And then the third bucket here, graph. And then the third bucket here, governance. This is a important one that governance. This is a important one that governance. This is a important one that I think AI makes more acute things that I think AI makes more acute things that I think AI makes more acute things that were that were accessible previously.

were that were accessible previously. were that were accessible previously. They're much more accessible now with They're much more accessible now with They're much more accessible now with with AI. And so you have to consider with AI. And so you have to consider with AI. And so you have to consider this. Your risk sphere is is larger. So this. Your risk sphere is is larger. So this. Your risk sphere is is larger. So things like PII need to be masked. you things like PII need to be masked. you things like PII need to be masked. you need to reconsider different uh need to reconsider different uh need to reconsider different uh sensitive data classifying this um sensitive data classifying this um sensitive data classifying this um making sure that there's the right making sure that there's the right making sure that there's the right entitlements for each user who's entitlements for each user who's entitlements for each user who's accessing your system. accessing your system. accessing your system. Okay, so that's the overview here. The Okay, so that's the overview here. The Okay, so that's the overview here. The this the the data model itself. Now, this the the data model itself. Now, this the the data model itself. Now, this is the part I'll walk through here. this is the part I'll walk through here. this is the part I'll walk through here. There's a a nice animation here, but There's a a nice animation here, but There's a a nice animation here, but hopefully the takeaway is you can hopefully the takeaway is you can hopefully the takeaway is you can picture your own your own organization picture your own your own organization picture your own your own organization story within what I show here. I'll get story within what I show here. I'll get story within what I show here. I'll get somewhat technical but it's only to hope somewhat technical but it's only to hope somewhat technical but it's only to hope hope hopefully to give you an idea of hope hopefully to give you an idea of hope hopefully to give you an idea of how we how we solved our problem and how we how we solved our problem and how we how we solved our problem and then you can hopefully uh model this to then you can hopefully uh model this to then you can hopefully uh model this to yours as well. Graph is very flexible um yours as well. Graph is very flexible um yours as well. Graph is very flexible um practical representation of a physical practical representation of a physical practical representation of a physical model. model. model. [snorts] Okay. So I'll zoom through a [snorts] Okay. So I'll zoom through a [snorts] Okay. So I'll zoom through a few of these here but the just the the few of these here but the just the the few of these here but the just the the entry point here we have over 80 entry point here we have over 80 entry point here we have over 80 different strategy teams. These teams different strategy teams. These teams different strategy teams. These teams have annual reviews that happen. This is have annual reviews that happen. This is have annual reviews that happen. This is how the budgeting for each year is how the budgeting for each year is how the budgeting for each year is derived. And then so we model this here derived. And then so we model this here derived. And then so we model this here in the graph. The the the meetings are in the graph. The the the meetings are in the graph. The the the meetings are where unstructured documents uh enter where unstructured documents uh enter where unstructured documents uh enter into this system from but then there's into this system from but then there's into this system from but then there's they have a structured connection to they have a structured connection to they have a structured connection to your other systems of record. Um what I your other systems of record. Um what I your other systems of record. Um what I show here is a conceptual data model. So show here is a conceptual data model. So show here is a conceptual data model. So it's flat. So you're not seeing the it's flat. So you're not seeing the it's flat. So you're not seeing the instantiation.

instantiation. instantiation. The actual graph there's you know many The actual graph there's you know many The actual graph there's you know many different nodes. Cardonality is it one different nodes. Cardonality is it one different nodes. Cardonality is it one one to n. So the actual graph it's you one to n. So the actual graph it's you one to n. So the actual graph it's you know even more complicated. But know even more complicated. But know even more complicated. But for the data model itself, let's let me for the data model itself, let's let me for the data model itself, let's let me let me show you the first different let me show you the first different let me show you the first different hier. So we have multiple hierarchies hier. So we have multiple hierarchies hier. So we have multiple hierarchies that exist within what we've modeled the that exist within what we've modeled the that exist within what we've modeled the there's different types of hierarchies there's different types of hierarchies there's different types of hierarchies you can have. In this case, this is a you can have. In this case, this is a you can have. In this case, this is a hopefully you can see all this very hopefully you can see all this very hopefully you can see all this very well, but it's a it's an it's a um well, but it's a it's an it's a um well, but it's a it's an it's a um additive DAG. So there's a all five additive DAG. So there's a all five additive DAG. So there's a all five levels here of this hierarchy from the levels here of this hierarchy from the levels here of this hierarchy from the top to the bottom matter. top to the bottom matter. top to the bottom matter. So you have to consider everything So you have to consider everything So you have to consider everything together. And so then there's different together. And so then there's different together. And so then there's different rollup patterns you can do to work rollup patterns you can do to work rollup patterns you can do to work across this this sort of pattern. In our across this this sort of pattern. In our across this this sort of pattern. In our case, we have a in path shortcut here case, we have a in path shortcut here case, we have a in path shortcut here that connects the funding path. Uh funds that connects the funding path. Uh funds that connects the funding path. Uh funds to bow is where we have the um the to bow is where we have the um the to bow is where we have the um the budget for each of these different budget for each of these different budget for each of these different funding teams that that's stored. So we have funding what's the the So we have funding what's the the internal funding teams have portfolios.

internal funding teams have portfolios. internal funding teams have portfolios. These portfolios then go towards These portfolios then go towards These portfolios then go towards different investments. Multiple funding different investments. Multiple funding different investments. Multiple funding teams fund an individual investment. So teams fund an individual investment. So teams fund an individual investment. So it's a endtoend relationship there. it's a endtoend relationship there. it's a endtoend relationship there. The The The investments are the thing that are our investments are the thing that are our investments are the thing that are our product. It's our it's our our business. product. It's our it's our our business. product. It's our it's our our business. But internally we have funds that then But internally we have funds that then But internally we have funds that then prioritize different different types of prioritize different different types of prioritize different different types of investments. That's what's shown here. investments. That's what's shown here. investments. That's what's shown here. And so you can take this down to the And so you can take this down to the And so you can take this down to the transaction level or you can have transaction level or you can have transaction level or you can have different uh annualbased aggregations different uh annualbased aggregations different uh annualbased aggregations that you map here as well. that you map here as well. that you map here as well. >> [snorts] >> [snorts] >> [snorts] >> And then from investment, there's a lot >> And then from investment, there's a lot >> And then from investment, there's a lot of interesting things you can do. You of interesting things you can do. You of interesting things you can do. You can map to all the different can map to all the different can map to all the different organizations and you can have different organizations and you can have different organizations and you can have different types of organizations and there's types of organizations and there's types of organizations and there's actually a lot here that is still kind actually a lot here that is still kind actually a lot here that is still kind of green space that we want to fill in. of green space that we want to fill in. of green space that we want to fill in. We have all these different observables We have all these different observables We have all these different observables that people have produced in the that people have produced in the that people have produced in the investments that we want to model here. investments that we want to model here. investments that we want to model here. So publish reports, products, you know, So publish reports, products, you know, So publish reports, products, you know, all this stuff is structured and all this stuff is structured and all this stuff is structured and connects to the entire uh organizational connects to the entire uh organizational connects to the entire uh organizational picture. >> [snorts] >> [snorts] >> So I mentioned that there's different >> So I mentioned that there's different >> So I mentioned that there's different hierarchies. This is the second type of hierarchies. This is the second type of hierarchies. This is the second type of hierarchy. At this hierarchy, each level hierarchy. At this hierarchy, each level hierarchy. At this hierarchy, each level matters in and of itself. And so it's matters in and of itself. And so it's matters in and of itself. And so it's not a a DAG necessarily. And so you can not a a DAG necessarily. And so you can not a a DAG necessarily. And so you can actually do things like precomputing the actually do things like precomputing the actually do things like precomputing the the some of these these different the some of these these different the some of these these different shortcuts. So the hierarchy it goes from shortcuts. So the hierarchy it goes from shortcuts. So the hierarchy it goes from the top to the bottom contains connects.

the top to the bottom contains connects. the top to the bottom contains connects. It this is showing the investment It this is showing the investment It this is showing the investment management side of the of the management side of the of the management side of the of the organization. organization. organization. And there's concepts of direct team And there's concepts of direct team And there's concepts of direct team management. So one team at like team management. So one team at like team management. So one team at like team level two manages the investment. But level two manages the investment. But level two manages the investment. But then there's also a concept of indirect then there's also a concept of indirect then there's also a concept of indirect management. So the uh children below management. So the uh children below management. So the uh children below team level two still should be team level two still should be team level two still should be attributed to the team level two. And so attributed to the team level two. And so attributed to the team level two. And so there's different things you can there's different things you can there's different things you can different games you can play with these different games you can play with these different games you can play with these sort of rollups to precomputee. I don't sort of rollups to precomputee. I don't sort of rollups to precomputee. I don't know if you can see this, but rollup know if you can see this, but rollup know if you can see this, but rollup manages m is a is a a derived edge that manages m is a is a a derived edge that manages m is a is a a derived edge that we that we create after we create the we that we create after we create the we that we create after we create the contains and manage manages edge. contains and manage manages edge. contains and manage manages edge. So I've shown two different two So I've shown two different two So I've shown two different two different lenses for one investment. different lenses for one investment. different lenses for one investment. There's the funding lens, the management There's the funding lens, the management There's the funding lens, the management lens and you can model both of these lens and you can model both of these lens and you can model both of these here. Then within within the graph, here. Then within within the graph, here. Then within within the graph, a third hierarchy here is people. You a third hierarchy here is people. You a third hierarchy here is people. You have organizations, you have org charts, have organizations, you have org charts, have organizations, you have org charts, and you have people who are owners, you and you have people who are owners, you and you have people who are owners, you have people who are attendees of have people who are attendees of have people who are attendees of meetings, you have people who are uh meetings, you have people who are uh meetings, you have people who are uh directors. There's all kinds of directors. There's all kinds of directors. There's all kinds of different roles they have. You can model different roles they have. You can model different roles they have. You can model these here. You can have their their uh these here. You can have their their uh these here. You can have their their uh you know, who they report to, what their you know, who they report to, what their you know, who they report to, what their uh team structure is. And these all are uh team structure is. And these all are uh team structure is. And these all are structured data that connects across structured data that connects across structured data that connects across systems. Traditionally, they existed in systems. Traditionally, they existed in systems. Traditionally, they existed in a just a HR source system, but they're a just a HR source system, but they're a just a HR source system, but they're relevant for the context of the the full relevant for the context of the the full relevant for the context of the the full story.

And then that leads to this And then that leads to this connectedness. So we have different connectedness. So we have different connectedness. So we have different source systems that were siloed. We to source systems that were siloed. We to source systems that were siloed. We to understand the entire picture for the understand the entire picture for the understand the entire picture for the agent to understand agent to understand agent to understand correctly across the structure you need correctly across the structure you need correctly across the structure you need to find these common B these common uh to find these common B these common uh to find these common B these common uh these common uh these common uh these common uh entities that you stitch together. And entities that you stitch together. And entities that you stitch together. And so that's what's shown here. These are so that's what's shown here. These are so that's what's shown here. These are different source systems but they're different source systems but they're different source systems but they're related quantity entities that exist related quantity entities that exist related quantity entities that exist there. And now the agent can traverse there. And now the agent can traverse there. And now the agent can traverse here and understand this pretty here and understand this pretty here and understand this pretty complicated organ organizational complicated organ organizational complicated organ organizational structure. One last part here that I haven't shown One last part here that I haven't shown yet is the the document part. So, and yet is the the document part. So, and yet is the the document part. So, and this is still there's there's a lot more this is still there's there's a lot more this is still there's there's a lot more we can do to this part. We've just been we can do to this part. We've just been we can do to this part. We've just been uh ingesting one different document uh ingesting one different document uh ingesting one different document source so far, but this is where you source so far, but this is where you source so far, but this is where you combine unstructured and structured. And combine unstructured and structured. And combine unstructured and structured. And this gets into part of the magic here this gets into part of the magic here this gets into part of the magic here that you can model with Neo4j. that you can model with Neo4j. that you can model with Neo4j. But we have meetings that have But we have meetings that have But we have meetings that have documents. Documents then have different documents. Documents then have different documents. Documents then have different semantic sections that you can or chunks semantic sections that you can or chunks semantic sections that you can or chunks that you can uh that you can model here.

that you can uh that you can model here. that you can uh that you can model here. You can put full text indexes across You can put full text indexes across You can put full text indexes across these to to aid in the different uh these to to aid in the different uh these to to aid in the different uh search and retrieval approaches for the search and retrieval approaches for the search and retrieval approaches for the agent. There could also just be a pure agent. There could also just be a pure agent. There could also just be a pure graph retrieval that that the agent graph retrieval that that the agent graph retrieval that that the agent does. And then all these things then does. And then all these things then does. And then all these things then connect back to your your your main connect back to your your your main connect back to your your your main organizational structure. So then as a whole this is what the data So then as a whole this is what the data model looks like. So I've been zooming model looks like. So I've been zooming model looks like. So I've been zooming in here now. You can see the the full in here now. You can see the the full in here now. You can see the the full interconnectedness of this four interconnectedness of this four interconnectedness of this four different systems one graph uh one different systems one graph uh one different systems one graph uh one semantic layer that's exposed through an semantic layer that's exposed through an semantic layer that's exposed through an MCP then to the to the agents. And so this is the so if you think of And so this is the so if you think of the agent's perspective, this is the the the agent's perspective, this is the the the agent's perspective, this is the the structure that it can dynamically structure that it can dynamically structure that it can dynamically discover and reason across at query discover and reason across at query discover and reason across at query time. And for the developer, it's also a time. And for the developer, it's also a time. And for the developer, it's also a very cool thing because it exposes, you very cool thing because it exposes, you very cool thing because it exposes, you know, what you don't know about your know, what you don't know about your know, what you don't know about your your the thing you're modeling. You very your the thing you're modeling. You very your the thing you're modeling. You very soon you find out that there's a gap in soon you find out that there's a gap in soon you find out that there's a gap in your understanding or there's some data your understanding or there's some data your understanding or there's some data set that you're not, you know, fully set that you're not, you know, fully set that you're not, you know, fully including. And so this this process in including. And so this this process in including. And so this this process in in of it in and of itself is very in of it in and of itself is very in of it in and of itself is very valuable.

Okay, let me give you a sense here what Okay, let me give you a sense here what we do with this now. So this is the I we do with this now. So this is the I we do with this now. So this is the I showed you the platform, the graph, but showed you the platform, the graph, but showed you the platform, the graph, but then how does this relate to AI? So then how does this relate to AI? So then how does this relate to AI? So we've we've connected this through MCP we've we've connected this through MCP we've we've connected this through MCP and I you know I discussed earlier what and I you know I discussed earlier what and I you know I discussed earlier what the what's durable, what's defensible to the what's durable, what's defensible to the what's durable, what's defensible to us. What was not defensible was the was us. What was not defensible was the was us. What was not defensible was the was the the chat interface was the UI and the the chat interface was the UI and the the chat interface was the UI and even in some cases the the general chat even in some cases the the general chat even in some cases the the general chat cases the um you know the agent cases the um you know the agent cases the um you know the agent interaction and so we users themselves interaction and so we users themselves interaction and so we users themselves are included already or chat GPT and so are included already or chat GPT and so are included already or chat GPT and so we serve the platform where they are and we serve the platform where they are and we serve the platform where they are and so it's served here now through MCP so it's served here now through MCP so it's served here now through MCP here's an example just kind of a here's an example just kind of a here's an example just kind of a innocuous uh question here but Neoforj innocuous uh question here but Neoforj innocuous uh question here but Neoforj has some off-the-shelf uh MCP servers has some off-the-shelf uh MCP servers has some off-the-shelf uh MCP servers Here we've actually modified these quite Here we've actually modified these quite Here we've actually modified these quite a bit here. We forked it. And then a bit here. We forked it. And then a bit here. We forked it. And then there's various updates to the schema. there's various updates to the schema. there's various updates to the schema. Uh things to to pass state back to the Uh things to to pass state back to the Uh things to to pass state back to the to to our system. You know, the to to our system. You know, the to to our system. You know, the conversation uh ids, the the message uh conversation uh ids, the the message uh conversation uh ids, the the message uh numbers, stuff like this we we've we've numbers, stuff like this we we've we've numbers, stuff like this we we've we've modified in these MCP tools.

modified in these MCP tools. modified in these MCP tools. But so that's a general chat experience. But so that's a general chat experience. But so that's a general chat experience. That's one entry point. The other part That's one entry point. The other part That's one entry point. The other part that we're building right now too that's that we're building right now too that's that we're building right now too that's very exciting is more constrained very exciting is more constrained very exciting is more constrained workflow experiences. And so these can workflow experiences. And so these can workflow experiences. And so these can also be offered through things like also be offered through things like also be offered through things like co-work uh clawed chat. And you can do co-work uh clawed chat. And you can do co-work uh clawed chat. And you can do things like um you can have your you can things like um you can have your you can things like um you can have your you can have MCP apps be the the you the have MCP apps be the the you the have MCP apps be the the you the standard entry way that users access you standard entry way that users access you standard entry way that users access you know different UIs that are uh ported know different UIs that are uh ported know different UIs that are uh ported into your your your your chat experience into your your your your chat experience into your your your your chat experience and you can have different sandbox based and you can have different sandbox based and you can have different sandbox based agents that then run the the workflow. agents that then run the the workflow. agents that then run the the workflow. And so these are these are active things And so these are these are active things And so these are these are active things that we're working on. It helps to that we're working on. It helps to that we're working on. It helps to constrain the experience compared to constrain the experience compared to constrain the experience compared to chat, but it at the same time it pulls chat, but it at the same time it pulls chat, but it at the same time it pulls from that same knowledge graph-based uh from that same knowledge graph-based uh from that same knowledge graph-based uh back-end platform. Okay, I've got a couple minutes. I'll Okay, I've got a couple minutes. I'll kind of speed through this, but the the kind of speed through this, but the the kind of speed through this, but the the way eval relate to data modeling is that way eval relate to data modeling is that way eval relate to data modeling is that as you're doing eval, you find you find as you're doing eval, you find you find as you're doing eval, you find you find gaps. You find ambiguities in your data gaps. You find ambiguities in your data gaps. You find ambiguities in your data model. you find ways in which users are model. you find ways in which users are model. you find ways in which users are asking questions that uh that are asking questions that uh that are asking questions that uh that are ambiguous or it's um you not it's not ambiguous or it's um you not it's not ambiguous or it's um you not it's not returning things that conform with the returning things that conform with the returning things that conform with the reporting standards. So what we've done reporting standards. So what we've done reporting standards. So what we've done here then is we've worked with data here then is we've worked with data here then is we've worked with data owners. We've we've uh built targeted uh owners. We've we've uh built targeted uh owners. We've we've uh built targeted uh eval questions that that they that match eval questions that that they that match eval questions that that they that match their reporting standards. We've their reporting standards. We've their reporting standards. We've separated these into different separated these into different separated these into different complexity tiers. One challenge is that complexity tiers. One challenge is that complexity tiers. One challenge is that the the structured data is constantly

the the structured data is constantly the the structured data is constantly changing. So we have to have the graph changing. So we have to have the graph changing. So we have to have the graph query itself that we that we create for query itself that we that we create for query itself that we that we create for each of these different questions and each of these different questions and each of these different questions and then at runtime for the eval we we pull then at runtime for the eval we we pull then at runtime for the eval we we pull from the live graph and then we compare from the live graph and then we compare from the live graph and then we compare that to what the agent is delivering for that to what the agent is delivering for that to what the agent is delivering for that question that question that question and so that's what's shown here then and so that's what's shown here then and so that's what's shown here then there's a feedback loop that you can do there's a feedback loop that you can do there's a feedback loop that you can do for this. So as you're running an eval for this. So as you're running an eval for this. So as you're running an eval pipeline, an eval structure pipeline, pipeline, an eval structure pipeline, pipeline, an eval structure pipeline, you have an LLM as a judge, we've you have an LLM as a judge, we've you have an LLM as a judge, we've modeled things like pass at one uh modeled things like pass at one uh modeled things like pass at one uh stability. So if you ask the same stability. So if you ask the same stability. So if you ask the same question multiple times, you get the question multiple times, you get the question multiple times, you get the same answer back, you can use LLM as a same answer back, you can use LLM as a same answer back, you can use LLM as a judge to to to measure this. And then judge to to to measure this. And then judge to to to measure this. And then there's a feedback loop here that you there's a feedback loop here that you there's a feedback loop here that you you can update then your your data you can update then your your data you can update then your your data model. you can update your uh your model. you can update your uh your model. you can update your uh your domain rules, your uh schema domain rules, your uh schema domain rules, your uh schema descriptions to help to help uh fill descriptions to help to help uh fill descriptions to help to help uh fill those those gaps that you that you find. Then after you do this, this is a this Then after you do this, this is a this is just some some eval reporting here is just some some eval reporting here is just some some eval reporting here that we uh we show the pass at one and that we uh we show the pass at one and that we uh we show the pass at one and the the stability for our system. So we the the stability for our system. So we the the stability for our system. So we we've gotten this very very strong. the we've gotten this very very strong. the we've gotten this very very strong. the the questions that we that we end up do the questions that we that we end up do the questions that we that we end up do missing, it tends to be things that are missing, it tends to be things that are missing, it tends to be things that are ambiguous in some way. And so it's not ambiguous in some way. And so it's not ambiguous in some way. And so it's not wrong, it's just that it's things that wrong, it's just that it's things that wrong, it's just that it's things that might be right but not what the user might be right but not what the user might be right but not what the user intended. So that that's kind of the intended. So that that's kind of the intended. So that that's kind of the constant struggle that we that we that constant struggle that we that we that constant struggle that we that we that we're working around we're working around we're working around 30 seconds here. What's ahead for SIP?

30 seconds here. What's ahead for SIP? 30 seconds here. What's ahead for SIP? So we're continue continuing to fill out So we're continue continuing to fill out So we're continue continuing to fill out our existing uh data from systems of our existing uh data from systems of our existing uh data from systems of records. So things that fit into our records. So things that fit into our records. So things that fit into our current data model. We want to expand current data model. We want to expand current data model. We want to expand the primary graph to additional the primary graph to additional the primary graph to additional enterprisewide data sets. There's a lot enterprisewide data sets. There's a lot enterprisewide data sets. There's a lot of there's a lot of demand for a of there's a lot of demand for a of there's a lot of demand for a federated graph experience. So, we have federated graph experience. So, we have federated graph experience. So, we have a main enterprise system, but we have a main enterprise system, but we have a main enterprise system, but we have specific teams that have their own data specific teams that have their own data specific teams that have their own data that they want to link to this. And so, that they want to link to this. And so, that they want to link to this. And so, we're working on how to do this uh we're working on how to do this uh we're working on how to do this uh different agentic experiences like I different agentic experiences like I different agentic experiences like I mentioned as well. mentioned as well. mentioned as well. And that's it. Uh, so yeah, please if And that's it. Uh, so yeah, please if And that's it. Uh, so yeah, please if you want to ask questions, if there's you want to ask questions, if there's you want to ask questions, if there's things that you want to talk about, I'll things that you want to talk about, I'll things that you want to talk about, I'll be out back or you can add me on be out back or you can add me on be out back or you can add me on LinkedIn here and you know, keep the LinkedIn here and you know, keep the LinkedIn here and you know, keep the conversation going. Thank you. conversation going. Thank you. conversation going. Thank you. [applause]