I build software, lead engineering teams, and turn legal experience into working products.
Shipping software since 2004.Called to the Ontario bar, 2012.
Work and projects
Work
Savvica, 2004
E-learning. J2EE, then Rails. Acquired by Educomp.
I co-founded an e-learning company in 2004. We wrote it in J2EE, rewrote it in Rails, and sold it to Educomp.
Barclays Capital, 2006
Regulation W compliance. Credit pricing on the grid. Eleven engineers.
Embedded at Barclays Capital. I built a Regulation W compliance system and a structured-credit pricing engine, and moved the quants' Excel toolkits onto grid services. The team grew to eleven.
The Ontario bar, 2012
Articled at Davies. Called to the bar in 2012.
I articled at Davies and have practised law since 2012.
Tim Hortons, 2017
3,500 restaurants. One espresso machine each.
Tim Hortons put espresso in every Canadian restaurant. I owned the program — vendor negotiation, product approval, installation — until 3,500 of them had a machine.
Milk, five years
Roughly 45 million litres a year.
I led procurement and negotiated a five-year national dairy supply for the Canadian restaurants. Roughly 45 million litres of milk and cream a year.
Lobo Genetics, 2018
A swab kit through Health Canada, and the lab that read it.
I wrote the entire backend in Rails, took a buccal-swab test kit through Health Canada, and stood up the certified lab that processed it. Entheon bought us.
Caravel, fractional
AI governance, privacy, licensing.
Technology counsel, fractional, alongside the operating jobs. AI governance, privacy under PIPEDA and GDPR, and SaaS and IP licensing from seed stage to enterprise.
Fishtail, 2022
Cross-border trade collateral. Rails on GKE.
I architected and built the trade-collateral platform — Rails, PostgreSQL and Hotwire on GKE, with real-time document processing — and ran the first pilots with smaller exporters.
City Lifestyle, 2024
180 magazines to 350. Twenty-one engineers.
CTO and general counsel. Twenty-one engineers through three leads. I wrote the sales qualifier in Symfony myself. The network went from 180 magazines to 350 while I was there.
AI Growth Labs, now
Founding engineer.
After the executive seats I went back to writing code full time. Hands-on is where the work is, and how I learn.
Built alone
Blackline
Canadian employment and contract research, free.
Coding agents read a treatise corpus in a sandbox of your own and answer with the passage attached. Waitlist-gated early trial, free.
Upload the brief and the record, then argue out loud. The jurist interrupts with questions tied to a passage, and keeps the transcript. Invite-only beta.
Ontario real estate, from the APS to the reporting letter.
AI real estate clerk and receptionist for Ontario lawyers. Send the APS; it carries the file to the reporting letter. The lawyer reviews, signs, registers and controls funds.
Most companies try to adopt AI by buying new software and asking people to learn it. It rarely sticks.
The models learned from centuries of people organising work: letters, memos, handoffs, sign-off. So treat agents as new hires. Give them the tools your people already use, a clear role, limits on what they may do and someone to report to. Then manage them.
I build AI agents that work the way people already work: by email, phone, chat, calendar and to-do list, managed like people.
Your team doesn't learn a new tool. They reply to an email.
Building now
Lintel
An AI clerk for Ontario real estate lawyers, with no interface. The lawyer emails, texts or faxes the agreement. Lintel runs the searches, chases the lender and the other side, and drafts everything up to the reporting letter. The lawyer reviews, signs, registers and controls the funds.
I give my AI agents an email address, a calendar and a to-do list, and I manage them the way I'd manage people. Here is why it works, what it costs and where it breaks.
I'm glad to compare notes with people adopting AI in their teams, building agents, or hiring for this work. Connect on LinkedIn, find a time, or email me.
Ajay Krishnan
AI-native lawyer and technology executive.
Called to the Ontario bar, 2012.
A few detours.
0:00 / 0:32
Read the film
A 32-second film about Ajay: software, law, kitchens, a lab, and what he makes now. Graphic scenes are illustrations, not archival footage. The costume and fire-truck photographs are real.
I started with code. Savvica, 2004.
Then law. Called to the Ontario bar in 2012.
3,500 kitchens. Tim Hortons, 2017.
A lab, too. Lobo Genetics, 2018.
Law. Code. Both. Blackline, Greenroom, Slate and Lintel. Earlier work at Caravel, Fishtail and City Lifestyle.
Not always serious. The costume and fire-truck photographs.
