AMA
AMA
on
JumpySushi
JumpySushi

Hi, I’m Sourabh, Co-founder & CTO of KushoAI, Ask me anything!

Hi Grapevine community,

Thank you for giving our previous post (https://bit.ly/44X5IJh) so much love.

I’m the technical co-founder at KushoAI and have previously worked at EMC, Ninjacart and FalconX as a full-stack software engineer.

Here to chat with you about everything from transitioning from being an IC developer to a founder, building on top of foundational Generative AI models, leading engineering teams, and building products that help devs do their job better.

I’ve learned first hand how difficult it is to build products from scratch, ship features within tight timelines, and manage on-call at odd hours when things break in prod.

I’ll be back at 4pm today to answer all your questions!

Post image
12mo ago
Talking product sense with Ridhi
9 min AI interview5 questions
Round 1 by Grapevine
SleepyBagel
SleepyBagel
SAP12mo

What's a underappreciated skill that devs should focus more on?

JumpySushi
JumpySushi

@GlassAttic Great question! I feel learning some product skills where you think about the end user’s wants and needs will go a long way in making you a better dev. Being able to communicate your ideas effectively is another thing that’s underappreciated, but becomes more and more important as you grow in your career and have to manage teams/stakeholders.

SparklyNugget
SparklyNugget
Cred12mo

Hi Sourabh,

Curious. How did you get into this problem statement? I want to get into building, and SaaS is very lucrative. But how does one even identify and empathize with something like 'API testing' challenges?

JumpySushi
JumpySushi

@OpulentDong Hey! For me (and a lot of devs), testing is a problem very close to my heart. In my experience, most devs hate testing, but on the flipside, if you don’t test your product properly, you’ll be woken up in the middle of the night to fix issues. Which is why it was easy to empathize with this particular problem for me, because I’ve faced it myself numerous times. Regarding how to identify challenges, I feel that instead of looking for things that seem lucrative and then trying to empathize, you should go after problems that you have personally faced and then pick the most lucrative one from it.

SwirlyKoala
SwirlyKoala
EY12mo

I would suggest getting a competitive term sheet from a top VC now. I like what you are doing. It would be a shame if LightSpeed/PeakXV/Accel/Matrix dont invest in you guys.

JumpySushi
JumpySushi

@ShallowSpawn Hey! We’ve just raised a pre-seed round and have built out an initial product and are iterating and getting traction. We’ll look at raising more soon once we need fuel to scale.

JumpyLlama
JumpyLlama

Hey. Can you share more about how to PMF discovery process look for you guys? Like where would you be on that journey, and what would be the logical steps to get to PMF with Kusho

JumpySushi
JumpySushi

@SubtleRefund Hey! We’re definitely at the early stages of finding PMF. Being a category creator product, there’s a lot of work to be done in terms of discovering where exactly will users see value at scale. The process remains quite simple - putting our product in the hands of users, getting feedback, and iterating ruthlessly without biases getting in the way. It’s served us well so far and will hopefully get us all the way to PMF.

JumpyLlama
JumpyLlama

Thank you for answering! One follow up: in terms of category creation - you must have global startups also going for the same users, no? Or is it very nascent?

PeppyHamster
PeppyHamster

Hi Sourabh,

So when someone is trying to get a job at early stage start up like yours, what should be the approach as its generally a small team of tight knit people generally at this early stages?

Also what should be the ideal mindset and expectations of a early employee in your opinion as a founder?

JumpySushi
JumpySushi

@awkward_potato Like you said, since the teams are tight-knit, many early stage startups end up hiring through personal connections and referrals. But we also made a few hires through platforms like AngelList and engineering slack/discord communities, so those are good places to track.

In terms of the skill-set needed, an early stage startup prioritizes candidates who have an overlap of competency and experience in their prior work, because there’s limited scope and funds to experiment. You’re expected to ramp up quickly and deliver outcomes at the earliest, without much hand-holding. There’s also always the unpredictability that accompanies daily operations, as opposed to structured processes in a larger company. Make sure you’re prepared for all of these things when considering working for an early stage startup.

QuirkyPotato
QuirkyPotato

Are you using Open AI under the hood?

JumpySushi
JumpySushi

@RevolvingPuzzle Can’t get into details, but we use a bunch of foundational models along with our own pre-trained models based on task at hand - we decide which model to use based on how reliably we can get the output that we want, how much latency is tolerable for the task and the cost incurred

SquishyPickle
SquishyPickle

What are your revenue streams?

JumpySushi
JumpySushi

@Bobble Our primary revenue stream is the subscription fee that we charge enterprise customers for using Kusho’s platform

SqueakyUnicorn
SqueakyUnicorn

hey what do you think about 5-10 years from now!! whether will it make anu difference such as replacement of human!!?

JumpySushi
JumpySushi

@FuzzyEarth6 Hey! We honestly don’t foresee AI applications for developers replacing humans. It will for the most part augment what developers do and allow them to focus on more meaningful (and fun) things. If I don’t have to spend too much time writing tests or debugging, I can actually focus my time and energy on designing the next feature really well.

SqueakyUnicorn
SqueakyUnicorn

okay i sure ki who know AI better he is going to surive more!! any idea on the competencies!! is it better to to have in depth knowledge of tools or should be better at prompt engineering??

DizzyMuffin
DizzyMuffin
Google12mo

What do you think India's contribution to AI will really be? We are far behind both US and china in hardware/software

JumpySushi
JumpySushi

@dopamine Hey! Can’t really say about hardware since I’m not involved in the space but there are a lot of Indian startups solving hard problems using AI for global companies. Some of them are already seeing decent traction. At the application layer at least Indian companies don’t seem to be far behind.

SleepyBanana
SleepyBanana

Currently I feel Kusho is black box testing, do you plan to build it differently? Where it takes context of your codebase and then generates cases?

JumpySushi
JumpySushi

@nanKhatai Great question! We started out with black box model because it is the easiest to use from a user’s perspective - you just need to provide API specs and you get ready-to-execute test suites without needing to seek anyone’s permission in your organization. For things like code repo access, we need to involve more stakeholders (from engineering, security team, etc.) in the decision-making process before users can start using Kusho, and this ends up prolonging the sales cycle - so we’ve deprioritized it for now. That said, this feature is definitely a part of our roadmap and we’ll launch it later this year.

Discover more
Curated from across
Software Engineers
by ZestySushiWeb Developer

AMA GrapeVine, tech stack, and all

hey grapeVine developers, why you guys aren't doing AMA about your journey and all.

On how you built it and released the first version, Been more than 6 months, IG.

Starting from scratch and who started first hands-on coding. What mis...