Gopinath Perkinian · AI product leader · Chennai

I keep asking why?

The question on my mind right nowWhy would anyone actually use this?

Fifteen years ago I was a programmer who could not stop asking what my code was for. Curiosity and empathy have driven every move since, along a path that was never a straight line: engineering, consulting, presales, sales, and nine years of product. Today I lead the agentic AI charter at Aviso. Before that I built semantic and voice search in 12 Indian languages at PhonePe Appstore. Every week I write down what the asking taught me.

Portrait of Gopinath Perkinian in a navy suit, smiling
15 yrsin software, 9 of them in product
116essays, one every week for 2+ years
730K+impressions on the four most-read essays
300M+users served by products I have worked on
What makes me different

What do you learn by sitting in every seat at the table?

Most product managers have argued with engineering, sales and consulting. I have done each of those jobs. Every move took me one step from executing someone else's specification toward owning the outcome.

Gopinath presenting to a room, gesturing with both hands
  1. 2010

    Engineer

    Data warehouse and BI at Cognizant. How you ask for data decides what it costs.

  2. 2014

    Consultant

    Product owner on a Medicaid platform. Making an ambiguous ask buildable is most of the job.

  3. 2016

    Presales

    Owned RFP responses at Virtusa. Win rate decides what work exists at all.

  4. 2017

    Sales

    Carried a number across Australia, New Zealand and APAC. Owning a target feels nothing like supporting one.

  5. 2018 to now

    Product

    Thryve, PhonePe, Aviso. I own the product I once sold, and I speak all four earlier languages.

Executing the specificationOwning the outcome

A non-linear path left me with three strengths.

1

I understand the technology

Deep enough to argue model choices, latency and cost with engineers and data scientists.

2

I tie product to the business

Every decision gets connected to a commercial outcome a CEO or a buyer would recognise.

3

I translate between the two

When engineering, sales and leadership disagree, I am rarely guessing what the other side needs.

Where I can help

Which of these is slowing you down?

I mentor product managers and engineers who are good at the craft and want to be trusted with bigger calls.

Product judgment

Can't get the room to back your call?

We work on framing the problem, choosing the trade-off, and defending the decision with evidence instead of conviction.

Collaboration

Does every function want something different?

We work on getting engineering, design, sales and leadership to one decision without burning trust.

Executive presence

Do your answers run long when the question is short?

We work on answering first, in one line, and staying steady when the question turns hostile.

Shipped

Does all the asking pay off?

Five results I can explain end to end, from the first piece of evidence to the trade-off I chose.

15 → 300

accounts a sales rep can cover each week with the AI prospecting agents I own.

Aviso AI
$30M → $100M

annual revenue. A five-year target reached in four and a half.

Thryve Digital, now enGen Global
44% → 66%

click-through on generic searches after the semantic search rebuild.

PhonePe Appstore
12

Indian languages of voice search, launched on the same day.

PhonePe Appstore
$5M

new-logo win built on a case-study matrix I assembled across every function.

Virtusa
What I write about

What separates the PM who gets the hard problems?

Frameworks are easy to find. I keep coming back to three skills that are harder to teach.

01

Product judgment

“Constraints are the canvas.”

Which problem deserves the week? I write about trade-offs, product-market fit, and what changes when every team builds on the same AI.

    02

    Cross-functional collaboration

    “Crossing the finish line alone means you probably ran the wrong course.”

    Who holds the part of the answer you are missing? I write about disagreeing well, ending comfortable silence, and getting one decision out of a room of fifteen.

      03

      Executive presence

      “Calmness is not a personality trait. It's a competitive choice.”

      Can you say it in one line when thirty people are watching? I write about conviction, hard questions, and staying clear under pressure.

        Decision log

        What did I get wrong?

        Judgment shows in the record, so here is mine. Each entry has the call, the evidence behind it, what happened, and what I do differently now.

        HeldPhonePe Appstore

        Ask the CEO to fund a semantic search rebuild

        Do not bring a hunch to a CEO.

        Evidence
        Three months hand-classifying about 60,000 search keywords with product ops, which showed that most long-tail searches were colloquial phrases and not app names.
        Outcome
        Funded. Click-through on generic searches went from 44% to 66%.
        HeldPhonePe Appstore

        Launch voice search in 12 languages on the same day

        For regional users, voice was the product and not a feature of it.

        Evidence
        Field research with drivers, delivery riders and domestic workers in tier two and tier three cities, who spoke their searches far more easily than they typed them.
        Outcome
        Regional activation rose 24% and search usage tripled.
        MissedPhonePe Appstore

        Group search results into collections to reduce clutter

        Usability and commercial outcome are two separate hypotheses. Test both.

        Evidence
        I validated that grouping made results easier to scan. I never modelled what it would do to revenue.
        Outcome
        Revenue went down, because the new layout moved people away from the surfaces that earned money.
        HeldAviso AI

        Take AI research off the screen and make the user wait

        On AI systems, quality, latency and cost are one decision and not three.

        Evidence
        The multi-agent research pipeline produced better personalization and took too long for a live screen. Accuracy mattered more to reps than speed.
        Outcome
        Six screens became two, reps are notified when output is ready, and weekly coverage went from 15 accounts to 300.
        MissedA generative AI feature

        Measure a content generator on how fast it assembles a draft

        A feature can hit its metric and still fail, if the metric was the wrong one.

        Evidence
        Speed was easy to measure and looked like the pain. The real pain was relevance.
        Outcome
        The feature hit its speed target and adoption stayed modest, because people rewrote the output anyway.
        MissedMy own career, 2013

        Commit a year to an MBA without testing the cheaper route

        Run the cheap experiment before funding the expensive one.

        Evidence
        I assumed consulting firms would not hire me without the degree and never checked. I treated an assumption as a constraint.
        Outcome
        The MBA opened consulting, sales and then product, so the decision worked. Six weeks of testing first would have cost me nothing.
        Writing

        What is on your board this week?

        Pick the one that sounds familiar and I will point you to three essays.

        Or start with the four most read

        Looking for something specific?

        Every essay I have published · taller bar = more readers
        Hover or tab through the bars to see each essay. The four brightest are the most read.
        About

        So who is doing all this asking?

        Portrait of Gopinath Perkinian in a navy suit, smiling

        I started as a programmer after computer science at CEG, Anna University. For three years I delivered exactly what the specification asked for, and kept wondering what business value the work created.

        That question took me to an MBA at Great Lakes, then into consulting, presales and a sales leadership programme, and finally into product in 2018.

        Curiosity and empathy have driven every move since. At Thryve Digital, now enGen Global, I co-created a five-year growth strategy for a US healthcare platform. At PhonePe Appstore I built search for people in tier two and tier three India who would rather speak than type. At Aviso I own the agentic charter for sales engagement, where AI agents do the research and outreach that used to eat a rep's week.

        I started writing every week to think more clearly. The archive has become a record of what I have learned, including the mistakes. Outside work I am a partner in a restaurant venture in Chennai, which teaches me more about customers than any dashboard.

        How I got here

        1. 2025 to now
          Senior Product Manager

          Aviso AI

          Agentic charter for sales and marketing automation: AI prospecting agents, an AI personalizer, and AVA, a conversational AI avatar and sales development rep.

          Part of the product team behind a platform that analysts ranked at the top of its category in 2026.

          • Leader and Customer FavoriteThe Forrester Wave: Revenue Orchestration Platforms for B2B, Q3 2026
          • Leader2026 Gartner Magic Quadrant for Revenue Action Orchestration
        2. 2022 to 2024
          Product Manager, Search

          PhonePe Appstore formerly Indus Appstore

          Semantic machine-learning search, Instant Search, and voice search in 12 Indian languages for India's homegrown Android app store.

        3. 2018 to 2022
          Product Manager, Strategy and Growth

          Thryve Digital now enGen Global

          Enterprise healthcare platforms for Highmark, and a partnership with Google on a preventive health platform. Thryve Elite, top 3% performer.

        4. 2016 to 2018
          Presales, then sales leadership

          Virtusa

          The only person selected into the Global Sales Leadership Programme that year. Carried a number across Australia, New Zealand and APAC.

        5. 2014 to 2016
          Consultant and Product Owner

          Cognizant Business Consulting

          Service authorisation and membership modules on an enterprise Medicaid platform.

        6. 2013 to 2014
          MBA, Marketing and Operations

          Great Lakes Institute of Management

          A year out to move from executing a specification toward owning the outcome.

        7. 2010 to 2013
          Data Warehouse and BI Engineer

          Cognizant

          Built a batch scheduler that removed a recurring manual reporting workload entirely. B.E. Computer Science from CEG, Anna University, just before.

        Gopinath in a maroon polo beside a flower garden display in Las Vegas Gopinath in a pink shirt, smiling at a microphone

        Education. MBA, Great Lakes Institute of Management, Chennai · B.E. Computer Science, College of Engineering, Guindy · Product Management specialization in AI and ML, Duke University.

        Got a why of your own?

        Speaking invitations, mentoring and conversations about AI products are all welcome. LinkedIn is fastest. Email works too.

        gopiperk@gmail.com

        Message on LinkedIn Send an email