"I build products users love and founders can point to on a P&L."
I've spent the last few years turning scrappy tools into platforms people actually use — and making sure they generate revenue while doing it. I think in first principles, work closely with engineers, and don't stop at shipping.
Selected Work
These aren't resume bullets — they're the decisions, the wrong turns, and the moments that shaped how I think about product.
The problem
We had 9 paying customers and $50K ARR, and the product was fundamentally unscalable. To power LinkedIn search, we were buying disposable trial accounts at $40 each — good for about 5 searches before LinkedIn's automation detection flagged them. We were one ban wave away from having no product. Every search was a roll of the dice.
The search & the decision
I spent months evaluating over a hundred third-party data solutions. Almost all of them required handing over our own LinkedIn accounts as collateral — a non-starter legally and operationally. I held the line on finding a real API-based integration. Eventually I found one: $3,600 for 250,000 searches. The math was obvious. We went from ~5 searches per $40 to 250,000 searches per $3,600 — a 500x improvement in unit economics. I also found a backup provider by digging through Reddit PM communities, where I connected with a YC founder and became one of their first 10 customers. That vendor now powers one of the largest B2B live-data signals in the market.
What it unlocked
Stable search infrastructure turned a browser extension into a platform. With reliable data, we could build AI-powered vetting, automated outreach, ATS integrations, and workflow automation. Revenue scaled to $1.2M ARR across 200+ customers. In close to two years, the platform had just 6 hours of total downtime.
The problem
A customer asked for something that sounded simple: they wanted to stop juggling Excel sheets for tracking contacts, relationships, next steps, and notes. They wanted it all inside the product. I designed and shipped a mini-CRM collaboration module — and considered it done. A few months later, 80%+ of customers weren't using it.
What I got wrong (and how I fixed it)
I had assumed the CSM team would handle adoption. That was my mistake. I embedded a Hotjar survey directly in the product to ask users if they knew this feature existed. Most didn't. So I changed my approach: I joined weekly customer calls myself, walked through the feature live, connected integrations they cared about (Fireflies notetakers, one-click enrichments), and started writing product newsletters about it. Not delegating — doing. Session time went from 30 to 120 minutes. The module became the backbone for EasyGrowth, our lead-scoring product.
The lesson I carry everywhere now
You can build the best feature in the world. If people don't know it exists, it doesn't exist. Adoption is part of the product. A PM's job doesn't end at launch.
This was my first real lesson in product: I shipped something good and it still failed — because I handed off to CSM and moved on. Joining those calls myself, writing those newsletters, embedding the survey to listen — that's the only reason this feature survived. Now I think about adoption strategy the same day I think about the feature.
The context
Task Mining at ServiceNow captures raw OS-level signals — every click, keystroke, and application interaction — to identify how people actually do their work. The company was going AI-native, and the question was: can we connect these raw signals to ServiceNow's automation products to create autonomous agents that can actually perform those tasks?
The hard part
Raw captured signals aren't instructions. A sequence of clicks tells you what happened, not why or how to replicate it intelligently. We needed to translate this signal noise into structured, step-by-step agent blueprints — including business purpose, task context, and enough detail for an AI agent to execute reliably without human input. I personally tested what signals were being captured at the OS level, identified where the gaps were, and worked with the team on using LLMs to bridge the translation layer.
What shipped
The Automation Center–Task Mining integration now auto-generates agent skeletons and detailed instructions using LLMs. What previously took a domain expert 20–40 minutes to author manually now takes under 2 minutes — with 90% output accuracy on v1. This was showcased at ServiceNow Knowledge 2026, where the VP was impressed that a baseline integration of this scope was completed in such a short timeframe. In Q2 2026, the team is now building on this foundation to power agents with tools that complete the full end-to-end workflow autonomously.
How I Think
A few things I've come to believe from building across recruiting tech, AI automation, and enterprise SaaS — and gotten wrong at least once before I believed them.
Adoption is part of the product
If a feature isn't used, it doesn't exist. I treat GTM and adoption as product problems, not hand-offs. That means writing newsletters, joining customer calls, and embedding surveys — not just shipping to Jira.
Platform economics beat feature velocity
The most impactful thing I've shipped wasn't a feature — it was an infrastructure decision that changed our unit economics by 500×. I think about what unlocks the next 10 things, not just the next one.
Metrics are evidence, not decoration
Every number I put on a slide, I've lived. I know what caused them to move, what I tried that didn't work, and what I'd do differently. I'm skeptical of metrics I can't explain.
Technical fluency changes the conversation
I've written code, debugged production issues, and tested OS-level signals in a Task Mining product. I don't need engineers to translate for me — and that changes how fast we move.
Background
Sept 2025 – Present
Senior Product Manager — ServiceNow
Task Mining · Automation Center · Agentic AI
Nov 2023 – Aug 2025
Product Manager — HireQuotient
Search infrastructure · Outreach · Analytics · Platform
Oct 2022 – Jun 2023
Software Engineer — Labra.io (Ibexlabs)
Azure OAuth2 · API integrations
Aug 2019 – Sep 2021
Associate Analyst — Deloitte Consulting USI
Production ops · Incident management · UAT
Strategy
Execution
Discovery
Go-to-Market
Technical Fluency
Recommendations
From the people who worked with me at ServiceNow — across product, engineering, and leadership.
What impressed me most was the tangible impact he delivered in such a compressed timeframe. Through genuine collaboration and a bias toward action, he drove outcomes that moved the needle on product goals well before anyone would have expected. His ability to connect dots across teams and translate complexity into clear direction made him a force multiplier for everyone around him. Where ambiguity and shifting priorities can stall even experienced PMs, Vishal leaned in with resilience and professionalism — keeping his eye on what mattered. I'd welcome the chance to work with him again in a heartbeat.
Vishal exceeded the initial expectations defined at the start of his engagement and has proven capable of operating independently within a complex product environment. He consistently sought clarity and asked the right questions until fully confident in his understanding of the problem space — reducing ambiguity and supporting informed decision-making. He also demonstrated constructive assertiveness when direction required adjustment, while maintaining professionalism and respect, earning credibility within the team.
Vishal has been one of very few diligent PMs that I have worked with in the recent past. He has the ability to break down complex business asks into engineering-level aspects, helping the team comprehend what is to be built and the purpose of it. He would be a great technical PM to work with, especially in the new world of AI-native product development.
Get in touch
If you're building something and need someone who thinks in first principles, ships fast, and doesn't stop at launch — I'd like to hear what you're working on.