VALENTINA RATNER Unknown unknowns, one of the beauties about it is in a deterministic system, you're telling the system what to check for. So by definition, you kind of know that there might be a problem in that area. With an agent, a lot of our customers come back to us saying, actually, it caught something that I didn't think to check, right? That's the power of it. And it's like, you may have deterministic checks in place that are going through a list of things. That will happen when it's something that's not on that list of things, right? JUDY WARNER Hi, everyone, it's Judy Warner. Welcome back to the ecosystem podcast. These days, it's harder to know what's hyping AI and how it's really changing engineering workflows across the industry. So I'm trying to sort that out today with my guest, Valentina Ratner, who's the co-founder and CEO of AllSpice. She and her team are very deliberate about what they hand off to AI to make your job and workflow faster, better, and more accurate, and what things are absolutely off limits and should be maintained and owned by engineers forever. I think you'll enjoy this thoughtful conversation with Valentina Ratner of AllSpice. Hi, Valentina, thanks so much for joining me again today. It's been a little while, and I'm excited to get an update and hear about all the amazing work you're doing at AllSpice. Can you start out, please, and introduce yourself and tell us a little bit about your background and what inspired you to start and launch AllSpice? VALENTINA RATNER Yeah, hi, Judy, it's great to be back here for this listening. My name is Valentina Ratner. I am one of the founders and the CEO of AllSpice.io. I come from an engineering background myself, and I started this company to build a lot of the tools I wish I had based on some of the experiences and the challenges that I went through and give engineers superpowers. I feel like the engineers that are their best when they're designing and doing creative work, and there's a lot, and today's engineering workflows that's repetitive and manual, and it's not the best part of the job. I think about how can we take the worst parts of the job and automate them and give them a way to technology so we can spend more time on the things that truly matter and move the needle. JUDY WARNERThat is the theme and the goal, and, gosh, you and Kyle have been at this a while. How long ago did you start, AllSpice? Yeah. Kyle and I met in graduate school about six or seven years ago at this point. We launched AllSpice in early 2022, and originally when AllSpice launched, we were very focused on small teams and startups, and we had this kind of GitHub-like platform for electronics and for EE teams, and the product has really matured and evolved since then, so nowadays we work primarily with large companies and enterprises and doing a lot more of that large-scale automation and workflow and collaboration for electronics teams, still our same audience and the same engineers that we're serving, but it's been great to see. We've gotten to grow alongside some of our customers, grow the company and the product as well. There's a lot of AI hardware startups out there. How would you characterize specifically for our listeners what parts of the workflow that you're trying to make better? JUDY WARNERI believe you'd said when you first started that compared to software engineering, just hardware was so clunky and that you were trying to emulate that, but could you articulate that in your own way, Valentina? VALENTINA RATNERYeah. Whenever I meet a new company, engineers will often come with some sort of challenge or problem that they're trying to solve. It's like, oh, we recently had a respin because X-ray C happened, and one of the questions we get a lot is, can your AI agent do X? I always start with understanding what the actual use case is and what's the actual problem that people are trying to solve, so the way we think about it at AllSpice is we think about it in jobs in three buckets or categories. There are jobs for AI, and AI is really, really good at certain areas. There are also jobs for deterministic systems, right? Sometimes if you need something that is very consistent, 100% accuracy, repeatability, compliance, stuff like that, sometimes traditional deterministic systems are the way to go, right? There are real jobs for humans, and we don't want to give AI things it's not ready for or not the best at, right? VALENTINA RATNER It's only as good as the data underneath it, right? It's only as good as a lot of these things, so if you can't outsource critical thinking, right? You can't outsource zero-to-one research and discovery, and all of these things are, by definition, if something has never been done before, it's going to be really hard to have that, but so we think about it in that way; it's like, okay, is this use case something that what is the best tool for the job, right? What is the right approach? And we try to put them in, like, okay, what is a team where engineers want to spend their time? And there are a lot of areas where there's really high value, and engineers' roles are evolving as we speak with the advent of AI and agents. I don't think that there's a question that agents are going to be part of the workflow; a lot of what customers and teams are going through is, where do these belong, right? VALENTINA RATNER What's the right thing to give them and what's not, and what's the right sequence, and are these agents ready, right? Sometimes they're ready on day one, sometimes they're not, and you have to kind of get this up to speed, so we think about it in those three categories. Okay, that's helpful. JUDY WARNER So lately, because I've been tracking sort of AI and hardware just out of genuine interest, and you and I known each other for a while, and what's really interesting right now is that the conversation and the way founders like you are thinking is evolving, and I think getting a tiny bit more clear, and I hear people being, engineers being less concerned about the hype. So in your mind, you know, you mentioned agentic AI, which I've been hearing more and more about, like the AI actually owning the parts of the grind job that no engineer loves doing, but where is AI or agentic AI really adding value right now practically, because when I put these videos up on the ecosystem YouTube channel, there's always like an engineer going, yeah, whatever, it takes you three times as long to do the job I could do by myself, so you let me know when you have something real. VALENTINA RATNERYeah, that's, and that's totally fair, and I agree with that. A lot of the promise of AI, and we've seen it in other fields, is, oh, you just tell the agent what to do, and then independently we do everything 100%, zero to finish, then release it, and I don't think that that's where electronics is today. You can't just say agent code design, a rocket controller, and just expect it to work, and do all these things on the other side with zero input, right? But there's a lot of areas where AI can be truly, truly powerful, and we also, you don't want to see the teams on the other extreme, which is basically rejecting the technology and kind of falling behind the peers that do adopt it. So, one of the areas that we find AI to be very powerful today is in tasks that are easy to trace, easy to supervise, and where humans still retain the final approval. And an example in our space is, for example, design reviews. VALENTINA RATNER So design reviews are a space where you can bring agents to it, they can help engineers do the job, but engineers can still supervise what the agents are doing, right? They can still kind of see anything that's happening, and they retain that final approval, that like accountability, and that final decision-making still remains human, right? Maybe there's a point where you've done enough of those that you kind of taught the agent how to think like you, and what's an acceptable trade-off or not, right? But that's usually a really good use case, especially when teams are getting started with no AI, right? Like introducing AI in places where you can bring the power of it, but still retain that engineering oversight and that final decision-making. So that's an example where we've seen it be very, very powerful. Another example is in areas where agents can do at scale things that will be impossible for humans to accomplish. VALENTINA RATNER And an example of that will be things like reading hundreds of pages of data sheets, right? So when you're checking component information and manufacturer specs is like sometimes I've seen 980 pages long data sheets, right? So who is going to go out there and read every single thing, you know, and kind of like truly incorporate everything and do that time, like a couple hundred like critical components, right? Like those things, if a human was to do it thoroughly, it will take a very long time. So reality winds up happening is humans are making a lot of decisions on trade-offs. They're saying, okay, I'm only going to review or I'm only going to verify these three components because these are the ones that bit me last time, were the ones that are highest like priority, and then the rest I'm just going to like let go. Or I'm going to review this set of components, but I'm only going to check the like the diagram, you know, and like that's it. VALENTINA RATNER So like people are kind of making a lot of trade-offs or how much coverage they can get. That's an area where agents can truly bring like that depth of coverage and agent can actually you can set it up like when you like wrapping up for the evening and you can come in the next morning and have a very detailed and thorough analysis of everything kind of down to the pages, right? Which is something that humans would have had a really, really hard time doing. Humans first, all the ones deciding what to do with that information, right? And like today that the judgment is still very engineered driven, but agents can help with things that are impossible for for engineers to do in a reasonable amount of time and really accelerate that and and give them like the depth of like coverage to to have like a hundred percent, like saying that you you verified a hundred percent of your components like every single data sheet and things like that. VALENTINA RATNERIt's something that it would have taken probably a team of a few folks, you know, like a few days to weeks and now with agents you can truly compress that and kind of give JUDY WARNER Give them 70-80% of it, right? And then just focus on the things that are maybe the exceptions rather than the norm. Right, which I think is what you meant when you said superpowers, right? You've cut down, you've compressed that part and even with humans you're you can we get tired and our eyes get fatigued or whatever, right? So did you find that review of 980 page of a data sheet to be successful and how do you as a company make sure that whether AI you're building or agents that you're putting in place are accurate. VALENTINA RATNER So to your question, ,our agent that AllSpice is called DRCY, spell D-R-C-Y, and DRCY operates as it's a for the users it looks like and it feels like an agent, right? And but if you look underneath the hood it's it's more like a system of agents, right? So we will have an agent that goes and fetches the data sheet and you can use models that are really good at parsing PDFs for things like that and then we'll have another agent doing like the analysis component by component and then we'll have a couple of agents doing the analysis and then we'll have a judge evaluating that analysis across those different evaluations, right? To present the results to the user. So to the user it will look like a result, right? VALENTINA RATNER But there's a lot that can it's happening behind the scenes to to give folks like the most like the most value for for those reviews where we we have DRCY in the platform which is our AI agent and then we also have automations which are deterministic checks. So most of the teams will will set up a combination of those two again depending on what the use case is. VALENTINA RATNERSo where agents where can really shine is again vast amount of information unknown unknowns one of the beauties about it is in a deterministic system you're telling the system what to check for so by definition you kind of know that there might be a problem in that area with an agent a lot of our customers come back to us saying actually it caught something that I didn't think to check right that's the hour of it and it's like you may have deterministic checks in place that are going through a list of things but what happens when it's something that's not on that list of things right that's the power of like kind of bringing this other set of eyes from the agent perspective to potentially highlight things that you wouldn't think to check and then our most sophisticated and successful teams will use a combination of those two right to kind of get the most value out of it. I like that idea because like I said with JUDY WARNER As humans we introduce risk right just because the limitations of our fatigue or whatever is going on and I like the idea that there's a check for blind spots which is wonderful and something that really hasn't been available to engineers in the past unless it was a collective physical or virtual team of design reviewers and you're hoping that collectively you'll catch that so I that sounds like a really important piece to me so every engineer though as you know Valentina has a very diverse software stack that they use to get a system designed so how does AllSpice.io sort of tie into such diverse software stacks and work in conjunction with those you know how does that how do you enable that connected workflow with such a wide variety of chosen software stacks? VALENTINA RATNER Yeah we build AllSpice to be tool agnostic and we understand that every company will have a different combination of how those tools come so the most powerful approach that we found is to build in a way that is open and accessible from a perspective of enabling teams to bring whatever other tools they have into into AllSpice. I don't think there is a single vendor out there that can be best in class for absolutely everything it's really really hard to be best in class for design for simulation for analytics for supply chain for compliance for security for specialty like fields so the only way that you can get best in class across a variety of fields is to to kind of have this open interconnected approach we we don't want to limit users on what tools they can have as part of their flow right and we want to enable them to to bring their best-in-class tools into AllSpice and the kind of the closed ecosystem approach makes it really hard where things will grow at work inside of it but if you like a new tools are coming out like on a weekly daily basis nowadays right so we built AllSpice as this framework where you can connect a lot of input data sources and these will be things like your ECAD tools or your simulation tools or again sometimes these data sheets from from partners and manufacturers or vendors component suppliers and then to other systems that tend to be consumers downstream which will be tools like a PLM system or something that's or something that's custom-built because we work with a lot of enterprises most companies will have their own internal agents right and you want to be this the the power of your data and what you have in AllSpice and in your electronics electronics tends to be and a lot of orgs a little bit disconnected from the rest of the company and they end up getting pulled into oh we have to report it we have this internal tracking system ourselves then we also have to go paste it in Jira ticket so that managers and program and product can can get visibility so can we remove that barrier right can we give engineers electrical engineers an environment that's built for them but also make it really easy to kind of bring in the rest of the company so being able to connect like custom agents that companies have it's with one example of that where you you want to be able to access the data through a couple of different couple of different ways so historically a lot of the tools have been built for humans so they're very UI UX heavy right like nowadays you're building you're building for these three things you're building for humans you're building for other systems and you're building for agents so in AllSpice for example we have like full API's and a headless model because it's like what humans may value on the UX UI right that's great for the engineers but what if you need agents at your company who access that data or pull information or participate right so right having this headless approach enables you to be able to kind of operate as as part of the bigger digital fabric at the company and not as an isolated unit which is being sadly the case for a long time in some of those organizations how many percentage-wise JUDY WARNER How many people want that headless model versus the other way like what are you seeing sort of evolving particularly an enterprise space in the way people are choosing to use and implement the tool yeah it's I would VALENTINA RATNER Say the teams that use headless it's usually a combination of both so we will have and it varies by stakeholder so we see a lot of the design engineers like avionics and sensors like the EE's will be on on the tool itself right like they want it like the ease of of interaction and they want to be able to just kind of like look at their designs the way they're used to like look at their schematics look at their their layouts but headless we see a lot of other types of teams coming in like from a platform or a development perspective and kind of tying it into that bigger picture one example is so we have DRCY and with one of our enterprise customers it was really cool just recently we found out that they had connected headless to their own agents so you had agent to agent interaction instead of AllSpice so DRCY was commenting and then they had their own agent with maybe other context from the company kind of commenting in as well and it looked like a real back-and-forth interaction between humans but it was agents to speaking to agents right and it's still visible so the humans can kind of supervise for now but you can think of a world where if these are like scoped enough and you've done enough of those and you've seen the repeatability and that trustworthiness it's like you could potentially like outsource like that piece completely right and like remove it from from the engineers bucket and give them time to focus on other things so it's I will say like the the companies that are at the forefront of innovation is like all of them and the question will come up differently like not everyone calls a headless headless is kind of a industry term from other kind of comparable tools out there in other industries but we get a lot of questions about like oh can you can you have API endpoints or oh can you can I access my design data or can I ask questions through a plot or through my own enterprise GPT tool that we're using right and all of those things is like yes the use case is like actually the way that you can do that is if you connect this headless but I will say most teams use a combination of both and different people use it in different use cases yeah yeah well it's nice that JUDY WARNER They can choose you know which was my question like the software stack is so diverse you know we've got your EDA tool and then you may have answers or ada's or a compliance tool or whatever it is and everybody has their own stack either by requirement or by preference VALENTINA RATNER so those tools like most of the large Enterprises like sometimes they have optimized tools like special versions of now they've extended a lot so even if you have the same to eCAD or PLM tools it may look different at that company because they take different configurations and then use them for a while so they've kind of set it up like in their own way so being able to that's usually where we see like nowadays like with AI it's become so easy to experiment and to test right and to put up MVPs and to come up with something really cool over the weekend right and what we see at these large enterprises is the the hardest part is is the last 20% you know the last mile of like getting it from cool idea and prototype to production ready compliant security grade enterprise connected to all of my systems right because again like a lot of this data is a lot of the power of AI it's contingent on the data that you have access models have become incredibly powerful and they're gonna continue to become even more and more powerful right where agents don't have access to the right data or the right context there's only so much they're gonna be able to do and at a large enterprise that usually comes down to connecting tools together and being able to kind of tie all of these pieces so that you can centralize all this information or so that you can give access to all of the information which historically has been a lot of manual uploading from one place to the next right but if you can do it through MCPs through APIs like then you can remove a lot of that barrier and then the analysis gets and the and the quality of the product becomes so much higher how do companies you know from aJUDY WARNERBusiness case scenario think about what this does to their ROI in regards to product development yeah when we talked to companies they think about ROI a couple of different ways and and I love the example that you gave which is like oh it would have taken me less time to just do it myself than to set up the agent and give you feedback and correct it and it's like that's table stakes right it's like if it's not safe it's like if that if that's where you're starting is like it's probably not not super valuable so yeah the way that we think about it is engineering efficiency and giving engineers time back right so what are the tasks on the activities that they're spending time on today that they don't have 20 more and what are the tasks and activities that are taking them a long time that we can compress right what are processes that take two or three weeks so we could get down to a few hours right like really powerful like moving the needle there that's one of them that's like an individual engineering level and that's the reason engineers themselves will choose to use an agent it's like it may take you time to set it up the first time right but the idea is that we I say it with for for DRCY I use this analogy of like it's like when you're you're making pancakes the first one is like always the worst one you know but these things get better over time like they're right themselves it's like the difference with a traditional deterministic program is that the agents will learn and will retain more context and will become better and better over time so the first time you're running it should be the worst time that you see it right and then you should only go upwards from there so individual engineering productivity efficiency right like even engineers are superpowers is number one number two is at a program level like compressing schedules and timelines we we hear we work with a lot of customers that are have seen incredible growth over the last couple of years especially around AI infrastructure right everything that's required building these massive data centers and elevate AI boom people talk a lot about AI boom from a model and infrastructure effective but it's like all of this runs with physical infrastructure and and our customers are building a lot of that and and they have like this incredible demand and they have to go down from 18 months to 12 months to six month programs right so how do you where do you find time to accelerate things is like that's one of the areas that is like okay again let's look at the program like things that were taking two or three weeks it's like how do we get them down to like a few hours and are they measuring that JUDY WARNER So are you or is their company's measuring that so they understand? Or what is that ROI you are measuring that with your customers I'm just again thinking of as a business person because right you you need to you need to have some kind of measurement to validate that this makes sense and of course it does but businesses are driven by metrics VALENTINA RATNER Yeah totally they do and I will say a lot of the the customers that we work with and and I give you the example on on AI for DRCY it's one of the marks the benchmarks is like can we can we outsource 50% of the work right that an engineer would have done throughout the reviews and we see agents like doing like 60 70 percent of it right so you can get to that level that is like okay we can again it's not 10% less work is like can we truly give an agent like 50% of like can they identify on their own 50 60 70 percent of the things that an engineer would have taken their time to do right and those are things that are meaningful that you can measure and then you can kind of compare a lot of the times what we do especially in during evaluation periods is compare it to two programs that they just went through right and it's like okay if you like what happened in the program that just wrapped up right and if this was done through DRCY or through agents right like how much faster would have moved what are the things that you would have caught that maybe you didn't and what was the what was the impact that you would have had so I kind of helps people paint that picture and and then it's like measuring like we we have customers where we have a team of about like 40 50 engineers and we we recently went through some of their their metrics and their usage and we identified that in about like five percent of their releases there was issues that we caught with AllSpice up they would have made it through and it's crazy to think that as engineers we we almost like expect that one of our revisions is just going to be a dud you know it's like okay we're going to do three iterations and one of them is just not going to want to work out and we have that baked in into our timelines because we expect something to go wrong and it's like well if it doesn't what if you can actually prevent all those things like at the design stage right what if what if there are no respins what if there are zero respins right and like you just get it right on the on the first try and we identify like yeah like dozens of releases that like didn't make it past the the review stage right and like validated in AllSpice and it's like that's like weeks and months of schedule right if you if you think about every one of this like can take you through four weeks to redo right and like we have all those things baked in into our timelines and it's like what if we what if we down what if you actually don't need that anymore it's an exciting prospect JUDY WARNER And with the compression and the innovation that's going it's forcing us there and you know I love all my engineering buddies and i respect their skill and talent and ability so much but i also see ai as absolutely inevitable it's coming whether you want it to or not and so you had said something earlier about you know the ones that learn to adopt and work with agents end up being the ones that do drive career and do see these marked changes but i also see still so much cynicism it's getting better because of agentic but what do you if somebody asked you just point blank, Valentina, "like i'm never going to give away my engineering judgment to machines and that's just not happening I'm cynical and the world's ending--how do you respond to a person like that that maybe invites them to think about it differently VALENTINA RATNER I don't think that there is enough being said about that out there about what's actually happening in hardware we have a privilege and the opportunity to work with some amazing enterprise companies and some of the best teams in the world and ai is happening it's not if it's not when it's how it's here so a lot of it is like it doesn't get flashy news coverage it doesn't make the headlines but this is happening already and we see it with the teams that we work with day in and day i and day in and day out and there is what's getting decided right now it's what's the right way and the wrong way to do it and to go about it and i think like that's where the true value of engineers and judgment is so i don't think it's a matter of like if ai is going to be part of it or not i think it's a matter of thinking about it what is the right way and what's the right way for your team what's the right way for your company and and how do you build a system that compounds and again gets better over time historically software like you will set it up and it will only be if you do nothing to it it will only be that good in perpetuity right with agents that's changing you have a lot of new information and like these agents get better and better and so i will i will kind of encourage people to think about it it's like what is the right way and what is the wrong way right and kind of having an opinion about what are the what are the parts of the job that you can truly give to ai and what are the parts of a job that remain deeply human we are very AI forward at AllSpice but we also we have very strong opinions about where ai doesn't belong right and we use it in our product a lot but we also have plenty of use cases where we say they ask us like can DRCY do like component checks about end of life or availability and we're like a code but that's actually not a good use case for ai we probably should set up some deterministic automation no problem no solution repeatable like point clear data sources past fake criteria you know like end of life is bad not in the life is good it's like it's a very scope problem that you can kind of and we don't have to give it to AI just because ai is the latest available technology where there is other plenty of great resources out there that can do a better cheaper faster more sustainable job but it's also like thinking about where where where the humans are bringing the most value there's typically when when we talk about companies like there's a the way that you can go about it it's like okay let's give ai the most important job you know or like let's give ai like the most valuable job i think about it is like how do we give ai and tools the worst parts of the job how do we give those systems the things that engineers don't want to do right yes not the things not the best part of the job it's like how can we how can you give away the things that you don't like or that are time consuming that are not value add that's not the best is your time so you can spend more and more of your time into the things that truly truly matter the way my JUDY WARNER The marketing side of my brain works is i i came up with this idea of you retain the architecture and engineering judgment everything that makes you the best most skilled excellent engineer and just hand off the grind like that's the part do you really want to read to hundreds of pages of is that a good use of your time and most people hate it but again it's it's interesting times we're all swimming in the fog a little bit but i see it getting a little crisper and you're you're sort of reinforcing that in our discussion so i appreciate the way that you think about it Valentina now because things are evolving so fast how when you're speaking to these amazing engineering teams how do you help them like future proof their software stack so this doesn't completely become a whole different thing a year from now and then they have to re-architect that whole stack because it got outdated or you know how do you speak to that part because engineers love their tools and they love this stability of being able to use that stack reliably especially in places like automotive and and aerospace and defense like there's really critical pieces in there that perfect in a perfect quality wouldn't have to jiggle those around a lot VALENTINA RATNER yeah the way we think about it is very strong and solid foundation with a very flexible and extendible top layer if you wish so i think what's really important for teams is to have a strong data foundation right it's like can you have your data in a place that's accessible in a structure and in a way that agents can understand something that they can actually know what's happening and what's going on and i'll give you an example there we we work with a lot of customers that when they start it's like oh i'm gonna i'm gonna give Claude a pdf of my schematic you know and then analyze it and it's like your code but not all of your information it's in in a pdf you know like you're probably missing out right or oh i'm gonna give it a pdf and and a BOM and a netlist and it's like well then you're starting to like aggregate all of this like different pieces but you're probably you're trying to put together back a picture like you're decomposing something into little chunks and then trying to reassemble it which you're probably gonna get some things around the edges wrong or miss something on there so step one is like can you have like full data access in a way that is like again a hundred percent structure understandable from there you can build a lot right and it's like so you need a solid data foundation and then you need an extensible interface or layer that allows you to do things if you're beholden to a vendor giving you a button to be able to do xyz right you may be waiting 18 months for that to happen right so kind of how can you do it in a way that allows you to quickly move and iterate because there may be tools a year from now that aren't available today and then they become industry standards like things are moving at that speed and hardware is a wall where in the past things will take years and we're going down to months and two weeks yeah so building in a way that allows you to extend things and you can if you have a strong data foundation right it's kind of like your home and then you can go from there to different point solutions but you always come back and you you have like a solid grounding place to kind of go to and from so foundation and then extensibility is the other one it's like if you you want to build in a way that you remove any potential for information loss one of the the reasons we talked about way that hardware is harder than the software to automate or to introduce AI is because software is a world of apis and perfect information flow they have a lot of disjointed they have a lot of tools their tools are connected hardware and the other side has a lot of very very powerful tool a lot more powerful than software in many cases to go from a to b you have to manually export or get a pdf or have someone type things in one place or the other right and that is both time consuming error prone so how can you remove any data or information loss between points is like it's kind of like one of the things where extensibility becomes really really important so removing data loss opportunities having a foundation in a format and languages or structures that agents can kind of understand and parse and then having it also in a way where engineers kind of can explore and experiment one of the powers with today is that a lot of engineers can build their own extensions right with ai themselves and you want to be able to do it in a way that encourages that quick test on iteration but you also have a process to to get them production ready right someone's building something and and you'll like it and you want to make it part of the stack right it's like making sure that we cross the we close the gap between mvp and okay production ready where we're actually going to put our real designs and a live product or something that's going to manufacturing soon on that side as well JUDY WARNER Well this has been a really amazing learning experience for me Valentina I'm sure will for our listeners too. Do you have or have you guys published any of the use cases you said we're not touting it sort of in a journalistic kind of way where people are seeing the changes that are happening at companies like AllSpice so do you do you have published case studies on that and or where can they learn more about your products and go dive in to do a little bit more learning and research so they can maybe be invited into the conversation and upskill as they go VALENTINA RATNER Yeah absolutely we do have a couple of case studies from some customer experiences that talk about what AllSpice and what DRCY has has done for them best places to check out our website AllSpice.io and i can share some links to different parts of the work more about the product and to learn more about our customers and the use cases that we're working with them on JUDY WARNER Valentina before i let you go what's next you've come so far since the last time we talked but what's coming up next and what are you excited about in regards to the current work that you're doing at AllSpice.io right now VALENTINA RATNER Something i'm really excited about coming to AllSpice is the ability to chat with DRCY we talked a lot about today around what's the best place for an agent to potentially start and at AllSpice we've done that through design reviews and even engineers that kind of analysis and and help on that side the next step is making it an interactive experience where you can start steering the agents and giving them more autonomous tasks and have them create and generate things for you so i'm very excited about that coming soon any timeline on that JUDY WARNER Whats the timeline, as far as month of the year or quarter that our users our listeners can take a look at and be looking forward to VALENTINA RATNER This is for this fall so very excited about that and we've been testing with some of our existing customers on a closed beta and it's been really powerful to to learn what are the things that they want DRCY to be able to generate for them for example so very excited to share more about that JUDY WARNER Well we'll definitely keep our eyes open for that and again Valentina thank you so much for your time and sharing the exciting things that you're doing across the industry and congratulations to you and your team For our listeners, thank you so much for for listening in on this conversation between myself and Valentina Ratner of AllSpice.io we will see you next week until then remember to always stay connected to the ecosystem!