PRODUCT LEADER / AI PLATFORMS / ENTERPRISE SAAS

I ship AI products enterprises trust and buy.

Forecasting at Workday before “foundation model” was a term. The platform beneath Intapp’s agents, with least autonomy as the default. I build the trust before the AI needs it.

See the calls I made Get in touch

CALIX / DIRECTOR, PM, CLOUD / 2014–2019

Built the cloud business from zero.

$0 $30M ARR 2014 2019

85% GROSS MARGIN

A hardware company's first software subscriptions.

WORKDAY / HEAD OF AI/ML PM, PLANNING / 2019–2021

Shipped it before it had a name.

OCT 2020 Pre-trained forecasting, GA AUG 2021 “Foundation model” coined, Stanford 2026 2 US patents granted 2020 2026

10 MONTHS AHEAD OF THE NAME

One model, pre-trained on public economic series. GA to thousands of Adaptive Planning customers on day one.

INTAPP / SR. DIRECTOR, PM, AI PLATFORM / 2021–2026

Built the platform the suite ran on.

PROMOTED SR. DIRECTOR $212M $590M ARR FY21 FY26

THE SUITE ABOVE MY PLATFORM LAYER

Total ARR, Intapp 8-K filings. Cloud net revenue retention 110% to 123% over the span.

BEFORE THE CURVES — 2005 PEROT SYSTEMS, SOFTWARE PRODUCTS · 2008 BT, MANAGED SECURITY · 2011 VERISIGN IDEFENSE, THREAT INTELLIGENCE

WHAT I BRING

I build the platform, not a feature.

Identity, APIs, data contracts and the tool surface agents call. The layer every product team builds on.

I own the number.

ARR, gross margin, retention. I make the product case on where it wins in the market and what it returns to the P&L.

I build senior teams and stay in the work.

Directors and Principal PMs, hired and grown. I stay deep enough in the architecture to argue the trade-offs myself.

I get there early.

A time-series foundation model in 2020. Least autonomy by default for agents, earned task by task.

Three calls

Each one is a decision, and what came of it.

CALIX / 2014–2019 / 0→1

Win the biggest operators first.

PROBLEM

Calix sold broadband equipment. Operators bought boxes, not software on subscription. This was the company's first cloud product, and the first large operators had to be convinced.

WHAT GOT BUILT

Support and marketing cloud products on a microservices platform with telemetry, analytics and automation, running on AWS, Azure and GCP. ML churn prediction and demand forecasting for operators.

RESULT

$30M ARR at 85% gross margin, from zero. One operator cut trouble tickets 36% and earned back its cost in three months.

WORKDAY / 2019–2021 / TURNAROUND

Replace the engine. Don't tune it.

PROBLEM

Anomaly detection was already live in Adaptive Planning, and adoption was low. Training on a customer's data required that customer to opt in under Workday's Innovation Services Addendum, so the models had little to learn from.

WHAT GOT BUILT

One forecasting model, pre-trained on public economic series and applied to each customer's own plans. A foundation model that needed nobody's private data and no signed addendum. Predictive Forecasting and a rebuilt anomaly detection both ran on it.

RESULT

GA October 2020, to the entire Adaptive Planning base — thousands of customers on day one. Two US patents granted in 2026 (12,614,086 and 12,749,008, under my legal name, Narayanan Krishnaswamy), more filings pending. A repeatable release and validation process for shipping ML in the planning unit.

INTAPP / 2021–2026 / PLATFORM AND COMPLIANCE

Build trust into the platform before the agents arrive.

PROBLEM

Intapp's biggest bet is Celeste: agentic AI for the most regulated buyers in software — law firms and private capital, working behind information barriers and MNPI rules. Agents ship there only if every action stays inside a user's entitlements and leaves a trail. That layer had to exist first.

WHAT GOT BUILT

First, compliance. I started the who-saw-what access record and got major UK clients to commit; it shipped under a peer. My team built the DealCloud integration with Intapp Walls.

Then, the layer beneath the agents. Firm admins write playbooks in plain language. My team built what they run on: the MCP gateway and servers, schema contracts for each firm's data model, and unified identity with delegated permissions and audit. Least autonomy by default, earned task by task.

And products on top. My group defined and launched Intapp Signals, a next-best-action engine for professionals, and folded three mobile apps into one.

CONTEXT

Across my five years owning this layer, the suite above it grew from $212M to $590M ARR. Cloud net revenue retention rose from about 110% to 123%.

Building now

I build agents myself. It keeps my judgment about what to delegate to them honest.

DESKTOP APP / DEMO COMING SOON

Personal portfolio terminal

Your money stays on your machine, and the agent sees the shape of your finances, never the raw numbers. It shows net worth, whether it is moving the right way, and what is moving it. It shows you what changed. What to do about it stays your call.

Demo coming soon, with a sample portfolio

AGENT / IN USE

Investment thesis tracker

An agent that re-tests long-horizon investment theses against SEC filings. A monthly check for events, a quarterly deep review. Built for investors, not traders.

NEXT FOR THIS SITE

An agent that answers questions about my work

Grounded in these pages and my résumé. If you would use it, say so.

Tell me when it's live

About

Trained as an engineer; got the MBA to own the business side. I am most useful where the technology and the business case have to be decided together.

I started as a C programmer and Unix admin. The rest of the career is in the curves and calls above. These days I present strategy to CEOs and customers.

In between came six years of security: managed security products at BT, then threat intelligence at VeriSign iDefense, where I launched its threat indicators feed for banks, retailers and federal agencies. That is why I built trust into the Intapp platform before the agents arrived. An agent is a new principal on the network: least privilege first.

MBA, University of Maryland, Robert H. Smith School of Business. MS in Computer Science, George Mason University.

HOW I LEAD

My team never guesses the priority.

Or why it's the priority. I give the conclusion first and repeat it until it sticks.

I argue my team's case in numbers.

The headcount, the platform, the bet, argued in revenue and margin, not roadmap slides.

When someone isn't delivering, I look at what's in their way first.

Ask me about the PM whose real problem was his calendar.

I find the bottleneck before adding hands.

Ask me how we stabilized a failing integration without new headcount.

I take hard problems to customers early.

You can see it in the calls above: UK clients at Intapp, the first operators at Calix.

Building an AI platform? Let's talk.

Open to product leadership roles and interim or advisory work.

US CITIZEN / SAN FRANCISCO BAY AREA

jaykrishnaswamy@gmail.com LinkedIn