Product leadership
Fractional or interim. For leadership teams with engineers and budget but nobody owning what gets built and why. Fewer things, shipped on time.
what to stop, what to keep, what to shipone page, then weekly
Janaka Ediriweera · Product and AI leader · Sri Lanka, remote worldwide
AI made building cheap. Knowing what to build is now the expensive part.
Fifteen years, forty products, and a company I now run with three people and a fleet of agents. I ask why before anything gets built, and I cut the roadmap before I add to it.
Haputale · 1,431 m · dawn
the house, the garden, the stream
the house, the garden, the stream
The point of view · morning
Most AI products being built right now don't need to exist. Ask why, and most of the roadmap falls away.
For fifteen years the expensive part of a product was building it. AI changed that. Now the expensive part is knowing what to build, for whom, and why. That is the question I ask first, and the one most teams skip.
- The wrapperA brand on a model and a workflow. Over half have no substance underneath, and a business model that won't hold.
- The long roadmapToo much on it, and nobody able to say why each thing is there. Stopping is where the return is.
- Building before the roadThe AI feature before the data and the process exist to feed it. It costs five times what it should. I learned that on the acre.
The proof · late morning
Three from forty. One I run, one I shipped, one I stopped.
Humanised and LankanTax · own products
Two products, run by three people and a fleet of agents
GivenHumanised Payroll and LankanTax, two live SaaS products, and every function that keeps a company alive, after a forty-person agency closed.
PlantedAgents and AI workflows across marketing, sales, finance, operations, engineering and customer success. Humans on the decisions.
3 humans · 11 accountancy firms
CAL.lk · fintech
A broker's back office, turned into retail investing
GivenA licensed broker, a back office built for professionals, and a founder's conviction to democratise investing.
PlantedResearch into the underserved investor, then landing pages, onboarding and the investing portal built from scratch around one thing: simplicity.
50,000 digital sign-ups in six months
Teamchain · the one I stopped
The product I stopped. The why wasn't there.
GivenA product I believed in, a team ready to build it, the wrong year, and not enough runway.
PlantedNothing. I asked why, saw the timing and the runway, and stopped before the money did.
Lived to fight another day
He led with empathy, rigor, and an obsession with simplicity. 50,000 customers rapidly onboarded, and a product people still describe as just so easy to use, years later. Janaka doesn't just design interfaces; he builds trust.
CAL.lk, on the retail investing productHis leadership of a small development team got our product to a stable production v1 which enabled us to go to market. When COVID put five hours of timezone between us, he maintained the communications and the roadmap.
SaaS founder, six years as development leadGrit, dedication, and determination. Highly ethical, always putting the interests of his customers and colleagues first.
RMIT Activator, Melbourne
Also: Colombo Stock Exchange · metjip · Humanised Payroll · takso · Hendry Group · Common Code · RMIT · Beta Launch, forty people, fully remote, ISO 27001 · and thirty more on the Portfolio page.
Where I learned it · the acre
One acre of steep ground in Haputale, bought with everything I had. I failed on it twice.
In Sri Lanka one acre is less than what a normal person can afford, always rural, never the city. It was the size of plot I could buy after Melbourne, with the savings and the comforts gone.
First I built before there was a road, and everything cost five times what it should. Then I went organic with nobody who knew how, me included. Both times I had stopped asking why I wanted it. Every team I meet is making one of those two mistakes with AI.
- One acre
- 4,047 m²
- Bought
- 2022
- Failed
- Twice
- Now growing
- Lettuce, beans, fruit
- In the plan
- Cows. Cheese.
Don't spend on the buildings. Plant the trees first.
Don't build the feature. Build the road it needs.
Don't build the feature. Build the road it needs.
field note · Haputale
The method · a field note
Know your why. Then the plan. Then the numbers.
Every engagement runs in that order. Read everything, talk to everyone separately, come back with one page: what to stop, what to keep, what to ship next. Then weekly. I stopped my own product, Teamchain, this way. The timing was off and the runway was short, so we lived to fight another day.
What I do · afternoon
Three ways to work with me. Each one starts with why.
AI delivery
One workflow at a time, where AI removes real hours. Built in your stack, measured against the manual way, handed over with the code. No wrappers.
shipped, not deckedI still open the code
Training
Founders and leadership teams, in the room or remote. 120 people trained with RMIT Activator, USAID and ICTA, and the one sentence I keep having to say.
if it isn't your why, you won't have the gasthe sentence
15+Years running product
40+Products shipped
120+Founders and leaders trained
3People running two products, with agents
Dusk · on the ground
Ask me why.
Three lines: what you're building, what you have to build it with, and what's not working. I reply within two working days, and the first question I ask is why.
janaka@oneacrefields.coma fire or a blanket, most nightsit gets cold up here