第十三讲 Space and Time Explore Intersection of Data within AI and Web3

Transcript

hello my name is Stephen Hilton I am the head of solutions at space and time and I'm going to do a session on the intersection of data within Ai and web 3 so with that let's jump into it again my name is Stephen Hilton uh headed Solutions at space and time uh I've been in crypto uh for three years professional I've been an advocate and an and an Enthusiast for many years um but I've had 10 years in managing solution engineering team especially in data driven Technologies uh particularly in Big Data 20 years in

big data and data management so uh most of my career has been in the data field in various flavors of engineering right um recently I've flipped over to doing that in the web 3 space uh but again I've I've been kind of following the industry and we'll say an Enthusiast for a long time so that's kind of me um and the topic here exploring the intersection of data within Ai and web 3 and so we're going to do this in a few different chunks I'm going to start with um just kind of putting together some thoughts on why I

think data uh traverses web 3 and AI very neatly how they kind of fit together uh we'll look at how AI is empowering web 3 and particularly using space and time and then turn around and look at how web 3 can Empower Ai and then we'll do some lightweight coding at the end so this will be um kind of one part discussion and one part uh technical so we'll say it's lightweight technical so with that let's jump into the first topic here web 3 and AI so I think a lot of people are starting

to recognize how well web3 and Ai and AI agents in particular fit together particularly when you start looking at the pros and cons um and how much of this is driven by the different characteristics of the data Behind These platforms and so um you can see on the left we have blockchain on the right we have ai um and the pros and cons and you'll notice the pros and cons are flipped right the things that that blockchain and web 3 are very good at such as being decentralized or giving user control over their own data these are things that uh AI today uh struggle with right

AI today is very centralized and that that means that the data is largely controlled by large Tech corporations right the the large Tech Giants and so um if we're going to get to a world where AI is really integral to your day-to-day existence uh that means you're exposing all of your data in kind of an uncontrolled way to large corporations not that we think corporations are inherently evil but they have they have different motivations than than maybe you uh on blockchain on the pro side free movement of value and that the data is deterministic and publicly provable

right whereas the cons of AIS today is that uh the usage the the usage cost of AI uh and the the movement of value is largely from the consumer of AI to again these Tech Giants because it's so centralized uh the data where in blockchain it's deterministic publicly provable of course in AI today it's obscured requires trust that's not necessarily a characteristic or a decision by these large corporations but because they're corporations they have certain regulatory statutes they have to follow so they don't always have uh the option in other words you're not going to find Google simply publishing all its user data out in the web because that would

be a gross you know misuse of of of their power but nothing's perfect blockchain has a number of cons um it's and and most of these kind of could be bundled up by saying it's just very complex it has a long learning curve uh it's unforgiving of mistakes if you send money to an address that doesn't exist it's just gone right there's no undo in this environment and really what this means it's just not really made for humans blockchain and web 3 uh time and time again you see very talented companies

trying to wrap web 3 in a way that makes it more usable more more forgiving for the imperfections of humans whereas AI this is where AI shines right this is exactly what AI was designed to do was to make uh make systems easy to use make them self-managing basically mimic human interaction that's really the whole point of AI is to I shouldn't say the entire point but one of the the fining characteristics is to be able to go to an AI system and say I would like a picture of a a Purple Panda uh eating oatmeal and just like just describing what you want

as if the AI agent was another human is typically enough to get what you want so in other words you're not programming anything you're not drawing a you're not opening up other interfaces the interface is a human interface right and so if we could take these to all of the pros that blockchain brings that really you know very cleanly uh mitigate the cons of the AI agent and likewise all of the pros of AI can very neatly mitigate the cons of blockchain right we start seeing what the future actually might look like if you could bring these two things together right

like now we have Pros like decentralized imagine if you and by the way there's lots of companies out there working towards decentralized AI right but imagine if you're going to have an AI buddy an a AI agent uh that's going to kind of follow you around digitally and it's it's it kind of almost becomes it becomes a personal assistant but you know po po could even have a personality that you find tuned to suit your style and it's going to collect all the information about you you may not want all of that information being sent to a centralized Corporation but if you could send it to a you or

have it operated by a decentralized collective of node operators all of a sudden the incentive is different the incentive to be honest it's cryptographically it cryptoeconomic secure uh and often times cryptographically secure at that so you know the future of web 3 plus AI it's decentralized that gives you more user control over your own data allows the free movement of value because imagine if you had an AI agent that you could give an allowance to or you could give a budget to and it could make purchases uh or or or sales on your behalf right because there's no intermediary it has

to contend with uh and then also deterministic is probably um a little we're still talking generative AI so you know deterministic but what it did the logs of what's going on uh are there and it's publicly provable right it requires no trust from a private entity you don't have to say well uh my AI bot did some data movement on on my behalf or did some movement of value on my behalf and I trust that it happened you can always go back and verify that now unlike the previous slide where there were pros and cons we really get pros and more Pros because not only do we get all of those Pros

merged together with with the synchronization of AI and web 3 data together you get pros and you get more Pros right because now you get all of the ease of use the self-management right having AI agents operate independently you just give give them instructions and and they know how to go operate in the sphere of web 3 right and so you can imagine uh AI agents acting as financial advisors or operational managers or technical developers a lot of a lot of work in the development space But even things like domestic coordination right you could uh um you could imagine you could have a an AI that um operates within your home can do things like like

like take a photo of your pantry or of your refrigerator determine what food you have at what levels you have them and then you just simply give this AI this domestic AI a budget for food and say when I'm low on these certain items uh place an order with my local grocery delivery service and the food just shows up all of a sudden you don't have to manage that that doesn't you know doesn't exist today but we're talking about parts of the future so you can see where these two things really work well together and and where you know where there's so much potential for merging these two web 3 and AI together all right um and by the way if you want

to do a little Extended Learning um there's a couple podcast and and YouTube channel here that um I think are really good the pod podcast goes into the combination of web 3 and AI very deeply it's a I don't want to say it's old it's a year and some I think it's a year and a half old uh but but the ideas that are in that are really great and then I just thought it was this YouTube video was fun because it proves that like the first AI bot uh that got to be a crypto Millionaire right so like these things are happening and kind of the all of a sudden the surprise ing choices and and conundrums that you have to deal with

when you have a an AI agent that owns a million you know a million dollars worth of crypto anyway some some fun extra reading for this if you're into it all right with that we're going to move on to the next section here which is talking about how uh we can use AI to empower web 3 and in in particular we're going to look at space and time natural language to SQL agent this is how space and time one of the ways space and time uses uh AI to help make web 3 easier to uh to engage with so actually for this I am going to uh flip over to a demo so if

you give me just a moment let's jump over to space and time here and we'll do a quick demo let's see here there we go um I'm all logged in so I should explain what space and time is space time is a decentralized data warehouse that has ZK tamperproof queries so what does that mean ZK tamper proof ZK stands for zero knowledge and and tamper proof just means uh that the tables cannot be tampered with once they're loaded and so you put those two things together essentially it allows space and time to

or allows users of space and time to query a data set in fact we'll jump over here and we'll just drop in uh this data set here and run this and you can get the query result uh as well as a proof and you can take that proof to a verifier who will verify it for you it's it's a third party so you can take it to any verifier in the world and they can verify the proof because essentially the ZK tamper proofing it's math uh it's it's the way that for example ZK sync does its rollup activities it doesn't rely on being

optimistic doesn't it it doesn't even necessarily need to rely on um consensus although I think it does uh but it's ZK means that you can mathematically calculate the T the the fact that the the table and the query in SpaceTime has not been tampered with so um what is that like what's the practical application of that imagine that you have a smart contract and it's a defi and you're going to accept a wallet and a token and an amount that the user wants to borrow right the requested amount in that scenario today most defi most

smart contracts they have a little bit of compute and a little bit of data and they don't they don't have access to the full history of the chain and so what they do is they just treat all of their users identically everyone gets the same yield on that particular offer what you could do is [clears throat] use space and time or or use your smart contract to query space and time uh go over teram mites of data uh like for example get the entire history of that wallet address and calculate you know return how many times has it been liquidated how many times has it repaid in full how many times is it paid early

the behavior of that wallet and the the characteristics of it return that back to the smart contract and then make an informed intelligent more sophisticated decision right you could maybe you want to offer a good user of better yield maybe you want to just flat turn down a user who gets liquidated all the time right so you can start making those more sophisticated business decisions and smart contracts because you have access to all of the data and by the way web 2 the Enterprise world today they make all of their decisions using big data right all of the decisions are proper

Enterprises they are datadriven decisions they don't just offer you a mortgage on a house based on the fact that you know they set the rate at 7% and anyone who walks in the door gets 7% regardless of their history they obviously you know uh do a lot of work to uh to determine what they think the credit worthiness is um until web 3 catches up with its ability to make those kinds of decisions the web two world is going to be very hesitant about adopting the technology because it becomes a step backwards for them space and time bridges that it allows smart contracts and other dap uh kind of systems to leverage a a big data

a large data set in real time very low latency and interject that data result into a decentralized process and because it's ZK proven it it maintains uh the trust factor of a of a blockchain right it's a trustless system you don't have to trust a blockchain because you can prove it you don't have to trust a query from space and time because you can prove it you just prove it with math instead so that's kind of um the high level space of time we obviously index different blockchains into the system I'm showing them up right here uh these are again large

tables so this is a 400 gigabyte table um there's where's our big one here I think our log tables right here that's 800 gigabytes so it's encroaching on a terabyte I think if we go to a higher volume chain like maybe polygon yeah so the logs table and polygon's four terabytes just that one table and then the transactions wherever it is is half a terabyte so you can start doing really big series on blockchain data because we go from block zero to about 10 seconds after block finality in that ball part so um where doeses this intersect with

AI let's run a query and I'm going to just grab this one right here this is going against uh Bitcoin transactions and what we're doing is getting the date we're going to look up the in count the number of blocks and then count the number of transactions for the last 14 days and we'll put it in order order by one that'll order it by the block date here uh in about two and a half seconds it traversed this transactions table which is 1,000 gigabytes otherwise known as a terabyte uh so you know you can see it's obviously a big data system it's

designed for traversing large amounts of data pulling out information quickly cool what if you don't know SQL right what if you said uh neat but I want to look up something on aptose right um I want to look up the ANS the like the DNS but the ANS on apos uh let's see if we could do that but instead of writing SQL we can just type in a uh a natural language prompt show me how many distinct aptose ANS domains exist that are still active by year and month now I'm going to run

this and it's going to uh reach out to an llm llm of choice and it's going to construct have the llm construct a query uh based on my question just my natural language question and you can see uh it's a little more vertical than I might otherwise read it but it looks like it got it right it's deleted false I asked for active uh year month and the count of active domains so there they are um I didn't have to know anything about the Aptos data model I didn't have to

know even how to write queries it's useful if I can look at it determine you know whether it's doing what I thought it would do uh but I don't this is an example of AI making something that's otherwise a little complicated right space and time we try hard to make the models intuitive and easy to understand right but it's still blockchain has some complex ideas and so as long as you understand what the ANS is you understand that it's like a DNS but it's an Aptos name server

then uh then you can ask for it very easily uh you could you could say all instead of active that are still active I think if I just said how many exist I can run this again and I'm going to get squl that looks similar this is generative and so uh if if you notice it took a different way to structure it this time but you know the end result is the same I mean it took out the Weare criteria of having it not have been deleted so it did exactly what it was supposed to do so it's an example of where AI uh can take something that can be fairly complex or at least

intermediately complex both you have to know SQL but you also have to understand the data model behind Aptos and uh and made it as easy as just saying show me how many domains exist right by this type of by this time dimensionality as long as you understand what that is uh you're good and this is just one example uh we have other uh AI products that are going to be released in the near future so I'd encourage you to check back uh often and if you want to go play around you're certainly welcome to uh the URL here is appspace and time.

and this will just bring you to the homepage um you can sign up for a trial account you just come here you can either connect to wallet and then use your wallet to um to drive your user experience or go away there or you can just sign up for a username password and uh it will ask you for an email just to confirm that and then um and then you're in then you'll be a trial user for a while it'll give you a certain amount of queries for free and then um you can try it out all right with that little Excursion into how space and time is using AI to help make web 3 easier to

understand and easier to work with let's jump back to our slides all right here we are back at the slideshow here we just got done covering AI empowering web 3 in particular on space and time our AI engine and like I mentioned lots of good things coming up we're you're going to continue to see both space and time but I think the entire industry um lean into using AI to abstract away complexity especially in the web 3 world but let's talk about how web 3 and Empower Ai

and so probably the the first way that um space and time but but I think I'm going to say web 3 more generically uh is doing things like value empowered AI agents when I say value I mean like Financial power like like having AI have the ability to to earn and spend and move money on behalf of users right and there's like some obvious examples like trading Bots these things are being built today um I've not actually tried it but uh it it seems like a really interesting idea right you it's almost

like having a financial advisor except that it's Ai and again using something like space and time where um if you can see all of these inductor nodes down at the bottom are writing data into the AI or sorry the uh the space and time chain or the SpaceTime Network um and then from that issuing either optimistic queries or ZK proven queries um you can you can see where we're feeding the data from these different blockchains which are in and of themselves Financial systems like they're you know they're distributed ledgers and making that available for AI agents

to consume very easily now you get kind of this flywheel of being able to have ai agents understand what's going on take some action make a purchase or make a trade that is then um executed on chain which then goes back up into space and time which goes back to the AI which then can reassess make a change or trade or purchase or sale which then goes back on chain and you just keep going round and round uh so you know having these systems having these comprehensive Data Systems in web 3 that can directly power AI is a very powerful idea uh business budgeting bots so having basically a a a an AI agent that's a

Financial Manager for you for your business or even for your Home Integrated domestic bot I think I mentioned earlier having essentially a manager for the components of your your house or your apartment right having something that again can evaluate the stock in your you know of your food or of your dog food or of different components or monitor how many streaming services you have and and you can do things like assign budgets to these you know to these different elements and have the AI um come through and say you sure you really should be uh ordering that door Dash because that's going to put you over budget and and by

the way I you know I went out and looked up there's a sale you know uh at your local pizza place maybe you want to go out for pizza instead of ordering door Dash I I'm making things up but it's the idea that you have an AI agent that's there that can help in the domestic setting as well I think is sometimes Overlook and I think that's going to be the bigger game Cher trading and and business and everyone who's kind of in the in the web 3 world like we're all trying to strive to make this reality we sometimes lose sight of the fact that this is potentially lifechanging for you know everybody even if they don't realize it

uh so that's that's how we see SpaceTime providing a data infrastructure to power AI agents that themselves can move around value let's talk about user control data [clears throat] so users user control data it's another good place where space time really stands out space time has a component that's blockchain we also have a component that's a like a dpen but with the ability to do these DK proven query results means that uh we can do things where we can provide the answer to a question without giving the underlying data

however we can give you a proof that that underlying data has not been tampered with so this this is like really interesting I think it's it's a use case we'll see a lot more sometimes confusing so let's maybe do an example let's say that um you have a service where um you're I don't know you're you're you're collecting users for I don't know some online service doesn't even matter what um so you have this user information and you want to share certain elements with Partners right but you don't want to give your or you don't want to expose your customers

data um you want to be able to answer questions about your users but not actually expose any more data than you need so for example maybe a partner says I need to know uh when there's a user I have to confirm that they're over 18 years old because there's regulations in place and you know that's a requirement they can run a query against your user data in space and time that simply says if they're over 18 answer yes otherwise answer no pretty straightforward um they can run that they have no access to the underlying data they only have

access to that yes no but they also can ask for that ZK proof and the proof itself proves the query and the underlying data without them having to see it again this is ZK zero knowledge so they get the answer without seeing the data so um now you have a system where you could have a number of of different entities that maybe are competitive or they for whatever regulatory reason they Cann not share data directly but they can do things like cooperate and and um answer questions for each

other so like let's say uh you have a number of financial uh service services like staff services and uh fraud is an issue in your area you can't share data with each other that wouldn't be appropriate but what you can do is sayy hey I have this user taking this doing this activity seems like fraud to me you could ask a question of another one of your competitor entities and say you know here's this behavior from this user I think it's fraud what do you think and they could give you back a fraud score right like U that behavior to us appears to be fraudulent

right and and as long as there's an understanding that everybody's going to participate but nobody can actually see the data all you're doing are sharing answers not data but you're also getting proof that nobody's cheating right that that when you're asking the question that the data underpinning that answer has not been tampered with so there's all sorts of interesting use cases um you know this also opens the door to say uh I'm going to put some data into space and time I'm going to give access to an you know to an AI agent that's not under my direct control but and I'm going to allow that AI agent to ask questions and

specific questions without giving them access to my direct data right so those kinds of I I think that really opens the door again for being super particular about what information you're sharing um and then I mentioned earlier smarter smart contracts right uh being able to do big data use cases in web 3 because the the future of smart contracts really are in are in ZK co-processors the ability to have a data problem inside of a smart contract that would be massively inefficient to solve inside the framework of a smart contract within the context of that but man if you can push

that out to a different data system whose entire context is solving that particular problem whether it's a relational data platform like space and time or a vector database or an object store or whatever it is right uh that becomes really powerful so you know I think ZK is is I think almost it I think at some point it will really become recognized as kind of the third leg in that stool of web 3 AI and ZK um it's just a it's a it's a more complex idea and I think it kind of breaks people's brains sometimes but with that let's

talk about what might this look like this is just kind of an example right using space and time um let's assume that we had an AI marketing agent AI agent whose whose goal is to do marketing and it's tasked with achieving like let's say end number of impress Impressions across different social platforms or you know or certain clicks to a particular site or you know whatever the the marketing Target is it's given that to achieve and it's also given a budget and a deadline and access to a pool of users now the users

are the users are going to opt in with either their wallet or their did and they're going to be incentivized to particip ipate and this is this is the winwin win of this type of AI marketing or this kind of web three concept of marketing which is to say users are fully in control of their data they are opting into marketing campaigns based on what they find interesting what marketers get out of they get a highly engaged and self- selected audience they're not one of the big um I spent many years in digital marketing one of the big struggles is finding the right target market for you know for your your

brand or whatever your program your campaign is is trying to Target um big money spent on Big Data trying to pull out different clusters of users who fit different categories right soccer moms or uh you know uh dads over 50 or or you know kids on Tik Tock or whatever I guess it's not Tik Tac anymore is it um but here you have people who may fit in that or may not but they've they've self- selected the interest so maybe uh who you might normally Market to soccer

moms maybe there's somebody who's you know an 80 year-old Grandpa who really is into that particular segment and and you would never Target that otherwise but they can self select into these marketing programs and because they're self- selecting they're getting incentivized so instead of marketers paying large Tech firms to just blast out to the best guest of the guests of the target market um they're not paying the big Tech firms they're paying the participants directly not you know small amount but to say say hey we're going to do a program where uh

we want you to you know help us evaluate Footwear because we know that you are a big fan of Nike and um and you've opted in because you like to collect shoes right and now we can deliver to you new marketing on new options and shoes you know on a reoccurring basis and this is where AI comes in because AI becomes the constant engagement between the brand and the audience right it's the it's the thing that's that's curating the experience for the user who's opt in who's opted in is being incentivized and then you know and then the again the

marketing team gets a very highly engaged audience out of it so it becomes a very Sy sistic can never say that word a very um collaborative environment between between the three uh so and and you know and when the user's done the your user can just go turn it off they're like I'm done with this um you know I'm tired of shoes I I'm for whatever reason right or or probably a better example is um you have a baby all of a sudden your spending habits just radically shift but if you select out of the ones that you're no longer interested in and

select into the ones that you are uh because you're being incentivized to do that um it you know commercials and ads they feel annoying only when they're not well targeted and most of the time they're not well targeted if all of the ads were perfectly targeted they would not feel like ads they would feel like reminders right or they would you would thank them for what they're offering you it's just the fact that today they're so poorly targeted because the user ultimately has no say in what they're being advertised to so um this is just one example you know we've done a number of examples with

domestic Bots and trading Bots and other budgeting Bots um you know but this is where uh the the the data the intersection of web 3 because web 3 is required to do this incentivization web 3 is requ space and time something like space and time is required to not just have the onchain data which is going to be public it's going to have the value transfer but also that private and ZK protected data right and it's like the combination of those things which would make something like this use case really powerful so I think we uh promised a little bit of coding but before we jump over to

that I just wanted to kind of point out that of course you can go to our docs page docs. space time .io or you can go to our web page space time. and click over the docs uh we do have SDK for python go nodejs I think we have a typescript through a partner U CLI jdbc if you want to go old school uh the rest API which is this entire reference right here and then we also have uh graphql as well so we have a good number of different interfaces that you can use but if you want to just super easy get started Python's easy because um because Python's easy so in Python we'll just write a super quick script here from

space and time import space and time and actually let's do space time and SXT table we'll create a table here and we'll just say SXT equals space and time and we want to give it a an environment uh an EnV file which has all of our secrets in it so I'm just going to say EnV file path equals EnV perfect you can see AI helping me out already it does like to suggest a lot of stuff that that I don't need it to do right now let's do um authenticate so at this point those three lines of code uh should allow us to no not passw just

just settle down there we go and we're just going to run this in debug mode and what this should do if I did this right is connect to space and time and authentication success true so with those three lines of code and actually you don't even need this the table yet I just put that in there um we're able to authenticate we'll run a query we'll say SXT uh execute query and let's say um interesting but no let's say select star from SXT demo. stocks

let's do that and what it will return look at the intellisense here it's actually going to return a tuple and it's which is going to be oops is going to be a um a Boolean FL which is whether it was successful and then the data and so I'm actually going to just run this again although it's actually going to reauthenticate that's fine and we can see the data now returned the stock data open low close blah blah blah blah blah and also it will return true because it was successful [snorts]

um and that way you could do things like if Success Print data something like that turn around and run that one more time so you see how like pretty straightforward and easy it is to get started with only a few lines of code of um of python here um we'll do one other thing here which is let's say my table equals space and time table um that's not at all what I wanted yeah this AI I'm the AI I'm using here is codium by the way which is kind of aggressive but it's it's it's pretty good sometimes let's say that we're going to do an AI manager um user optin so we're going to create

the the use case we kind of talked about in the last section where we're going to have an AI uh marketing said manager AI marketing uh and and then you can go in and say Here's my wallet I want to opt into this markting program this would be one of many tables that you would need to do this but let's just create one table just for fun and here I'm going to say from file oops from file which is looking for the EnV so I'm just going to give it the same EnV because I happen to know I have um my key for my table so in space and time um every op object

needs a a public private key pair and this thing here this space of time this is kind of a convenience object that rolls a number of them together but I can go to space and time SXT the object and we can see there's actually a user object in there and the user's got a user ID and it's got um a public private key pair I'll I'll print out the public key pair not going to print out the private key uh but you know um you can see that this has a key pair the table will also have a key pair the key pair will be different for the table than the user sometimes people get that confused or they think my my user's key will also be the key pair for the table that's not

the case it needs its own um let's do access type and we're going to say SXT access table access dot let's make it a public read that way anyone can read my table actually probably it should be permission but we're going to leave it public read anyway and then um if I say space and time parent that's just to say that convenience object can be passed in and then if you set up any logging like you can do logging you can add handlers you can you can set up a lot of the environment through this xxt because it's kind of convenience object um it's convenience object the intention is to

be convenient you can just pass all of that into this object here with that one command and I think that will do it the other thing I need for my table is of course I need to be able to say create ddl this is the ddl that is needed to create the table so here we would do something like that works and um it's going okay I don't think I need the ID let's just use the user ID and I'm going to be picky here per minute I like it when we do it like this and I don't really need all of this we just need int and or actually I'm going to make it a uu ID so I'm going to use

that as barar we don't need to specify uh and then we would want the campaign ID and then we would want yeah the opt and date that sounds right although we want a date and uh we will need the primary key I also want to have a reward reward token barart and a reward uh no reward amount which could be a decimal and that probably looks good and I think that looks good so this this way I could have a user who

decides they want optin for this campaign there's an opt-in date we probably ought to give an opt out date so they can decide at what point they're done playing they can escape uh reward token reward amount and we'll say let's say daily just to make it specific and with that little bit the last thing we need here is a Clos per in and then we're going to do a with statement and this will actually get created for us it's with statement but um it's a placeholder for it

with that um the last thing we need is to add some permissions my table dot now to do permissions we use biscuits biscuits if you ever want to go read up on them go to biscuit sec.org uh biscuit is an encoded token looks something like this and it will you can actually decode it and see the permissions inside of here so maybe we'll try that here in just a second for now we're going to add a biscuit you just give it a name let's say admin and then we have to give it some permissions and I'll just say uh actually I think I can say sx. Grant

Dot and then here's all the pr permissions you can grant other to all because this is my admin biscuit and if I usually you want a couple you'd probably say my table. ADD biscuit and let's say this is for my AI agent and we're going to say sx. grant. let's let them let's just let them select that they can't change anything or maybe maybe they can change now let's just leave it at that um actually let's give them one more

just for just because we'll let them insert a row too but you can't delete you can't update and we could do other things by the way we can say for certain like the time or certain requirements for the user or the subscription ID there's a lot of different variables but we'll just leave it at that because I this is starting to run a little long and I want to kind of wrap it up here but um I do want to come here and say print uh my table. biscuits and you can see what the different biscuits are so the capability accessing this particular table uh that's the private key for the table I guess that's okay

it's uh I'm not really going to leave this table so uh uh let's see the capability here's the biscuit token right here really what I wanted to do is copy that come drop it in here and you can see the different capabilities that we've assigned to our table and with that we're pretty much ready to say if if my table dot exists uh my table. drop drop it and then either way we're going to create it and then I think I guess we could insert a few rows let's just create it for now

there we go and now we should be able to say my table do exist yes it exists and I'd be able to go see it in the uh interface we had before um I could say SXT do execute execute query and just no not at all what I wanted to do but thanks for the nice try Podium I just wanted to say select star from this that's fine and it worked it's just empty excuse me anyway um that's all the coding I think

we need to do tonight I just I wanted to point out that it's actually really straightforward to get get started that um some of it if you've worked with databases in the past will feel really familiar uh you just have to create to create table statement um there is a little bit of cryptography that's added actually we could say um if I said come here I say oh I I killed it let's uh let's do that debug again we do all of this we get down here and say my table do um create ddl

actually let's do print it'll look nice you'll notice when it says width statement it has a width statement here and it's going to do the public key for the table and it's going to have um access type public read which is what I specified it up here so it kind of puts together this and other things to um to make it work uh to do tamper proof uh what is it it's immutable immutable it's true I think you also have to say tamper proof

is true and doing that will make it both tamper proof and IM mutable right now tamper proof tables have to be immutable which is why you can you have to that I will say that um if you're going to do a ZK proven table a tamper proof table um I would come back and do that last because part of it being immutable means that you cannot delete it you cannot drop the table this this command would fail um so if you have a schema that you really like and the name that you really like like um and you create it and then you realize you forgot something then sorry you have to extend the model a

different way because a tamperproof table is immutable and immutable tables cannot be dropped because if you could drop them you could effectively mutate them you could basically copy the data drop it recreate it the way you wanted it reload the table with the same name it would basically be mutable so you cannot drop immutable tables so what I do is when we're doing this I will either uh leave this as not tamper proof until the last moment when I need to actually go and test it as being tamperproof and then I'll do that so that I have time to make changes if I want or more often than not I will also do a version and so

um I can create mult you know if I version two version three but the name is preserved just a little tips and tricks on immutability all right I think um I think that's a good place to stop and we'll jump back over oops oh all right there we go thank you thank you so much again my name is Stephen Hilton uh from space and time our web page is down here on the bottom as is my email and telegram handle if you'd like to get in touch I would love to to speak with you I I think people who are the audience for this presentation understand that this moment in time is

pivotal never have human you know human race seen a a moment of transformation like this so uh we are in a very real sense redefining what the future is and uh I for one am fortunate I feel fortunate and grateful for being able to be part of it uh thank you so much and happy building