Ananth Nayak

Datasets

Enhancing search by organizing the web using embeddings

Checkout Datasets ↗

Exa dataset can provide a comprehensive list of any arbitrary entities that match any arbitrary set of criteria, by utilizing Exa’s neural search, high quality and massive index, and AI agents.

Specific project details are under NDA but I'd be happy to discuss my design approach !

Role

Product Designer
Front end Prototyper

2024 / (2 Months)

Team

Tools

Figma, Framer motion, React, GSAP, Three.js

BACKGROUND

Search today is an advertisement driven business. Often times your search results are polluted with advertisements from different services not giving you exactly what you are looking for. Current search engines are good at certain specific search queries.

Imagine having access to a perfect search engine that understands exactly what you are looking for, has everything on the internet, and is as intelligent as an analyst. It's not made for casual internet searches. But to help people gain deeper insights from the internet!

Sales Motion

Finding a list of leads and qualifying them with detailed data during a sales motion.

Recruitment

Finding a list of employment prospects and qualifying them during recruitment.

Competitor Analysis

Finding a table of companies in a niche market, plus arbitrary market data (e.g., market size, revenue, strengths) to inform a competitor landscape evaluation

Search today is advertisement driven

Solutions

Product Landing page : Pick your data type of interest
Dataset performs great for queries that include finding specific companies, people and articles/papers. The goal was to harness these data categories that the user is looking for early on in the user's journey of creating a table of results.

Final landing page design

Prototypes

Step by Step process of setting up an Async Query - Just like setting up a job on Vercel!
Dashboard UI for landing page. Gives user's access to jobs that are currently under process and the ones that have finished.
Search on landing page without any walk through.

Step by Step process of setting up an Async Query - Just like setting up a job on Vercel!
Dashboard UI for landing page. Gives user's access to jobs that are currently under process and the ones that have finished.
Search on landing page without any walk through.

Criteria and Query Breakdown : Better query results in a better search
Query breakdown gives users granular control over their search by letting them define criteria, select result types, and customize output. It ensures precision, especially when current day language models hallucinate.

Final Query & Breakdown modal

Prototypes

After user types query and breaks it down, the Model will come up with Criteria's that can later be edited by the user.
Real time query breakdown. As the user types query, we break it down to come up with potential criteria's
As users type, the query is broken into criteria, and 3-5 sample outputs are generated instantly to preview the final dataset.

After user types query and breaks it down, the Model will come up with Criteria's that can later be edited by the user.
Real time query breakdown. As the user types query, we break it down to come up with potential criteria's
As users type, the query is broken into criteria, and 3-5 sample outputs are generated instantly to preview the final dataset.

Final Dataset
On fine tuning their criteria's users then get to either view samples of their final dataset or generate the entire dataset at a go.

Final Dataset Design

MISCELLANIOUS

Re-designing website landing page:
Apart from working on datasets, I also had the opportunity to work on some interactive designs and prototypes for the website redesign of Exa.ai. Here are some iterations: