Arjun Kannan

I'm building ResiDesk.

Residents text their property manager about rent, repairs, renewals, parking, and everything else happening in their building. We help the team answer and keep track of what still needs to happen.

About ResiDesk Say hello

I co-founded the company and spend most of my time on product and data. The rest is customer calls, sales, hiring, bugs, and whatever else is holding us up that day.

ResiDesk

Renting is expensive and weirdly bad.

Your landlord usually talks to you when they need rent. You talk to them when the sink breaks. In between, residents say a lot about what works and what does not. Most property teams never get to see all of it together.

A broken lock can start in a text, continue on a call, and end up in a review. The property team may be switching between an inbox, the property system, a spreadsheet, a vendor, and a policy document just to work out what happened.

ResiDesk gives the property team one history of those conversations. They can answer the resident, see what is still open, and find problems that keep coming back.

AI can read more of this history than a person has time to. It can also be confidently wrong. Every answer needs the lease, the building policy, past messages, and a way to hand the conversation to a person.

See ResiDesk

Read what they wrote

Keep the resident's words and the messages that came before.

Pull the history

Find the lease, building policy, open work order, and the last thing the team promised.

Do the next thing

Answer, send a vendor, explain a charge, or ask someone to decide.

Fix the recurring problem

If twenty residents misunderstand the move-in instructions, write better instructions.

How I got here

I thought I was going to get a PhD.

Both my parents have PhDs. I went to Cornell for applied physics and expected to get one too.

Physics

I wrote an iPhone app for the lab.

The microscopes were sensitive to magnetic noise, and checking the room took a lot of manual work. An iPhone already had a magnetometer, so I wrote an app that cut the setup time roughly in half.

BlackRock

BlackRock rejected me, then hired me six months later.

I liked the people I met in the first interview, so I thanked them and stayed in touch. When another role opened, I tried again. I got the job in Delaware and asked if they could move it to New York. They did. Once there, I won the internal hackathon three times. One prototype won and still went nowhere because a spreadsheet was easier. Later I helped build software that financial advisors used to explain portfolios to clients.

Climb Credit

I ran product, engineering, and data at Climb Credit.

We financed career-focused education. A credit score could tell us whether someone had repaid a loan before. It could not tell us whether the school would help them finish, find work, or earn more. We brought completion rates, graduate earnings, and employment outcomes into the way we made loans.

ResiDesk

Then I started ResiDesk.

Residents were already telling property teams about repairs, renewals, rent, parking, and everything else happening in their buildings. We started by putting those conversations in one place.

30,000+

Homes in a four-month study of what residents said about Wi-Fi and whether they renewed. Read the study

$1M → $300M

Climb's annual loan volume grew from roughly $1 million to $300 million while I ran product, engineering, and data. The early Climb story

$40M ARR

Annual recurring revenue reached by the advisor product I helped build from its first version at BlackRock.

How I work

I ask a lot of questions.

Start with the person doing the work.

What are they trying to finish? What do they have to remember, copy, chase, or explain again?

Take the complaint seriously. Then check the data.

If someone says a problem is everywhere, I want to know how many cases there are and why they happened. That does not make the complaint less real.

Try it with a full queue.

I want to see what happens when the policy is unclear and nobody who built the software is in the room.

Do not fix the same mistake twice.

When someone corrects the software, use the correction the next time.

Writing

I write about AI, housing, and software.

I usually start writing after I say something in a meeting and realize I am not sure I believe it.

I write about the information a model sees, the tools it can use, how we test it, and when a person takes over.

I am also working on questions around AI and fair housing. If software helps decide who gets an answer or which problem gets attention, we should test whether it treats people fairly.

Read my notes

Talks and interviews

Three conversations