Shreyansh Jain

9359723466

HOME / JEEVA AI

Building an AI sales platform

Building an AI sales platform

Building an AI sales platform

Building an AI sales platform

FOUNDING DESIGN LEAD

JEEVA AI ↗

2024 — 2025

SEED STAGE

AI SALES AUTOMATION

Zero → One

Overview

Overview

Jeeva is an AI sales platform — agents that find, enrich, and reach leads on their own. I joined at seed stage: a bold product idea, an outdated Squarespace site, and no design foundation under either.

As founding design lead I owned two things end to end — the product, and the brand and website that sold it — and built a small design team as we grew.

CHAPTER 01

The Product

The Product

The software I designed — an AI sales team that researches, writes, and reaches out on its own.

The software I designed — an AI sales team that researches, writes, and reaches out on its own.

The work that eats the company

The work that eats the company

Imagine you’ve built something people actually want. The demos land. The pipeline should be full. One problem: someone has to fill it.

First, you find who to sell to.

Thousands of companies might fit. A handful actually do — and telling them apart is hours of research per lead.

Then you learn enough to sound human.

The right name, role, email, phone, and a reason to care — dug out of ten browser tabs.

Then you write.

A message worth a reply, personalised, for every person. Or a blast everyone deletes.

Then you do it again tomorrow.

Outbound isn’t a project. It’s a treadmill.

The tools that promised to fix this — Clay, Apollo — still put a human behind every step. They made the treadmill faster. They never got you off it.

What if the sales team wasn’t people at all — but AI agents that research, write, and reach out on their own?

What if the sales team wasn’t people at all — but AI agents that research, write, and reach out on their own?

In 2024, nothing on the market really did this. That was Jeeva’s bet — and the product I was there to design.

In 2024, nothing on the market really did this. That was Jeeva’s bet — and the product I was there to design.

So I stayed close to how people used it

So I stayed close to how people used it

Across ~3,000 customers and 450+ support tickets, I read every recurring complaint — then watched PostHog recordings and heatmaps to see what people actually did, turning patterns into a prioritised backlog with PMs and engineering.

~3,000 CUSTOMERS · 450+ SUPPORT TICKETS · POSTHOG RECORDINGS & HEATMAPS · CLAY & APOLLO TEARDOWNS

How do you find the right 50 — without a filter?

How do you find the right 50 — without a filter?

Most tools make you speak their language: filters, tags, boolean strings. Jeeva flips it — you describe who you want, in plain English, and the agent goes looking.

Find me 50 software CEOs in California working in AI.

One sentence in; a real, matched list out. No query builder in sight.

A name isn’t enough — so how do you learn the rest?

A name isn’t enough — so how do you learn the rest?

A lead is useless until you can reach it. Jeeva enriches each one — verified email, phone, role, company signals — and shows exactly how complete each record is.

ENRICHED — READY

PARTIAL — MISSING FIELDS

VERIFIED

EMPTY

Now enrich them with emails and phone numbers.

How do you say something worth a reply — fifty times?

How do you say something worth a reply — fifty times?

Generic gets ignored; personal doesn’t scale for a human. The agent drafts a message per lead from what it just learned — and you keep control of tone and edits.

How does all of this run without you in the loop?

How does all of this run without you in the loop?

The manual way was six separate decisions. I designed the automation to hold that complexity instead of dumping it on the user.

Audience

→

Enrichment

→

Message

→

Timing

→

Sequence

→

Launch

Each step stays visible, but only asks for a decision when it truly needs one.

AUTOMATION BUILDER — the six-step flow

The moment it clicks

The moment it clicks

Every capability above also lives inside one chat. You don’t configure Jeeva — you talk to it, and the same engine does the work underneath.

CHAT ASSISTANT — conversational execution

The bug that only showed up at the end

The bug that only showed up at the end

One complaint kept returning: a user sets up a whole campaign, generates dozens of emails — then notices the signature is wrong. Every email, fixed by hand.

The fix wasn’t a new setting; it was moving an old one. The signature decision now happens before the work, not after it.

CHAPTER 02

The Brand & Website

The Brand & Website

The front door I designed and built — so the product had somewhere to land.

The front door I designed and built — so the product had somewhere to land.

A product no one sees can’t sell itself

A product no one sees can’t sell itself

The product was moving fast, but the site still looked like an early-stage company — an outdated Squarespace build that undersold everything behind it.

Rebuilding the front door

Rebuilding the front door

I moved the site off Squarespace onto Framer — CMS-driven and component-based — so marketing and content could ship without rebuilding the experience each time. It grew to 160+ pages: product, pricing, demos, resources, blog, and support.

SQUARESPACE → FRAMER — the rebuilt site

A system, not pages

A system, not pages

A 70+ component website system — alongside a matching 80+ component product design system — became the shared language across brand, web, and product, with tokens, states, and a Lottie / Framer-Motion motion layer.

DESIGN SYSTEM — components, tokens, states

DESIGN SYSTEM — components, tokens, states

Never finished

Never finished

The site was a permanent experiment — hero, pricing, and landing-page variants, shipped and compared. Those experiments drove +40% sign-ups and grew the site to ~30,000 users.

EXPERIMENTS — hero / pricing / landing variants

THE RESULT

The Product

+26%

+26%

user retention in two months

~3,000

~3,000

customers

80+

80+

product components

The Website

70+

70+

website components

160+

160+

pages shipped

+40%

+40%

sign-ups

~30,000

~30,000

site users

Closing

Closing

Zero-to-one design isn’t waiting for a clean brief. It’s owning the product and the front door at once — staying close to the users and the team, and turning real signals into decisions, every day.

Thanks for taking the time to explore my work.

© 2026 SHREYANSH JAIN

LINKEDIN

EMAIL

RESUME