Aspire’s Aju Mathew on Generative AI in Software Development

hi I'm James Maguire and on today's e speaks we're talking about generative Ai and its role in software development it's clear that generative AI has the capacity to completely reshape the software development process to discuss that I'm joined by someone who knows a lot about the topic with me is auu Matthew vice president software engineering at Aspire auu very good to have you with us today nice meeting you here James and I'm happy to be here great uh so we hear so much about generative AI these days so many headlines what is your sense in terms of software development what is the current role of generative AI well generative AI has been around for almost a year and a half to two now in the area of software development uh you know we've seen uh developers or testers having started to use gener generative AI as a coding assistant now basically asking questions by using prompts U into the chat interface that uh some of these gen platforms provide and get answers from that and which is used in some form in your software development activity now this is a kind of a tactical uh usage of J where you're looking for a point solution so here J may you know give a step by-step approach to your problems and or even a a snippet of the code or solution to address the problem so that's how we see uh you know how we see gen in the current role for software development that makes a lot of sense well are companies adopting it rapidly or are they still waiting to to move forward with it well it's it's a mixed bag here um some companies have subscribed to the J platforms uh which also has a chat interface and deployed it in Enterprise wide within their ecosystem and and has given it an face to all of their employees uh with some you know uh dos and Dons and guidelines for usage whereas others have you know not yet made a formal decision for an Enterprise wide roll out and uh is kinding uh kind of in a wait and watch mode or kind of doing poc's on these aspects actually so it's a mixed pack yeah I mean definitely I've seen that myself well I think that a really big question is how can generative AI play a leading role in software development what about in terms of epics and user stories well um from a a leading role of generative a and software development uh we need to start using generative AI for complex use cases the one that I spoke about earlier is like a point solution when we start looking at more complex use cases in software development then it's like using it in different phases of the sdlc by chaining a promps instead of giving single prompts you kind of give a chain of prompts M uh this can as well be done within the chat interface of the J platform or you know for repeated templatized or parameterized usage uh we can uh use the underlying apis to generate appropriate artifacts for each phas so uh that's how you could uh use it in a more slightly more advanced fashion within software development and to your questions on epics and user stories this is this is slightly futuristic let me talk about the current phase to start with uh I've seen people take pieces of the requirement and you know kind of feed that into generative AI platform and translate that into like single epics or a high level user story but in the near future I'm envisioning this to uh allow the generative AI platform to consume the business requirements document as is or even the recordings or transcripts of uh requirements discussions with business stakeholders and uh with that consumed and with the right type of prom chaining and some Advanced Techniques uh we can Envision to generate um the entire epics or high level user stories or even detailed user stories in the near future James that's how I see that it seems it seems like a lot of lot of potential there is is there one thing you talked about this little bit just a follow up is there something else the companies need to do to be able to use generative Ai and epics and user stories well it's it's um see once you have the platform they they need to spend a lot of time in how to chain prompts to be able to consume that uh the requirements so that that will take more effort and and I'm sure in the near future you know there will be tools and platforms built on top of this to enable easer use in the near future gam yeah what about the design and user interface aspects how can generative AI assist with these yeah I I I'd kind of broadly classify this into three different uh um areas one is on the usability Discovery and research phase the second is the wireframe and visual design prototype and the last being the technical design so you know there are small small use cases in each of these that U the usability engineers or the ux engineers can use it for for example in the ux discovery and research phase they can use it for usability audits then you know potentially identifying feature list generation and compar Ison between uh uh potentially Market uh options um ask U questions to generate in terms of what are the market trends and recommendations on features um identify typical user groups or user personas or user Journeys uh for an e-commerce uh you know use case or something along those lines um also looking at um generating insights for you know from user feedback and survey um that's that's something that you know people are you know using you know generative a for right now comparative analysis strengths weaknesses and Market positioning uh generating information about that then uh basic information architecture which is used by the ux engineers so generating that basic architecture uh from a UI standpoint is something um lot of research report generation user if you questionnaire preparation questions so anything that enables you to generate content from a ux standpoint it can be used uh from a point solution perspective CH and that's on the first part and the second was on the wireframe and visual design prototype here I would say like it'll play a companion role um to the visual designer so it can basically generate various structures and layouts um uh acting as a second person to brainstorm with and find different types of solutions and you know the person can pick and choose what they want to take forward so basically acting as a companion um basically for more complex problems it serves as a companion to offer suggestions to the uh usability Engineers the other use case within that is basically um it it or generative a acting as a mode board Creator providing a wide range of uh visual design Styles basically design variations um and you know if you want to start off with some suggestions it it'll give you options that from which you can choose and then basically work on top of that um basically uh a lot of the research work can be done U from by the usability engineers to generative Ai and last but not the least is the design portion the technical design here we're talking about um you know basically uh two areas within technical design one is a high level design and the lowlevel design on the high level design uh how I Invision is uh you allow generative AI to generate the application architecture and the deployment architecture based on technical inputs and parameters so it's not that easy as as the way I just described it I'm sure there will be a lot of uh uh tweaks that need to be done um or thought process that needs to go in generating such information uh but it's definitely possible and doable um we've seen um you know PCS happening in this space uh uh in the market and similarly for the lowlevel design uh where we expect the generative AI to generate class diagrams and sequence diagrams based on technical inputs so that's how uh you know I Envision all the three areas to we get covered through generative AI on the user interface as well as on the design games yeah those are some pretty major you know changes and possibilities if I'm hearing you correctly sounds like generative AI is going to really have a major effect on the design and user interface absolutely absolutely it's just that it's still points Solutions I wouldn't say it'll go end to end but you can use it for the different uses that I mentioned I think the the challenge or the way forward is to see how this will seamlessly flow or integrate so that it becomes a seamless uh experience for the person or the usability engineer or the designer or the architect who uses this interface uh so that's obviously you know still long ways to go before we reach that seamless experience but yes it has a lot of advantages I think the question that that many companies are wondering is what do you see as the future agenda of AI and software development it seems like some very big changes are ahead so um whatever we discussed so far are basically Point Solutions um prob a little more complex use cases as far as chaining of prompts and you know using the API layer to you know address complex Solutions The Way Forward of the future of gen generative Ai and software development is basically I'm looking at endtoend software development life cycle automation right basically one input feeding into the other right from ux to you know technical design to development to testing to devops so basically uh if you would Envision uh the future of generative AI this is how I would Envision it so the entire splc automation happening through generative AI so one phase feeding into the other and it in each Spas having generative AI generate the respective artifacts and all of this appearing in their native work environment for example if somebody's using Azure devops or somebody's using jir or somebody's using a specific IDE for you know writing their code or using a visual interface designer all of this natively integrated with some of those existing tools and providing a se's experience for them to you know feed input and then you know generate an output then which feeds into the next phase so end to endend automation is what we can look forward to in the future and I'm and I'm anticipating both small and large software providers to kind of build Specialized or Integrated Solutions for each of the sdlc phases so that's that's what I that's how I see that and it's going to be an interesting pH so end to end automation would you care to say a time frame for that might that be next year or a couple years how long will we we wait for that couple couple of years from now possibly a year or two uh is how I anticipate we are seeing specialized Solutions coming out right now Point too Solutions but um U but yeah it will take time for end to endend automation possibly two two to three years from now but it's yeah it's in the very near future J it's really fascinating a a lot of good material uh thank you so much for sharing expertise today I really learned a lot I greatly appreciate it thank you thank you James thanks for having me here

This transcript was generated automatically from the video's captions and may contain errors.