Customer support · Virtual assistant · AI workflows
I'm Kent, a customer support and virtual assistant in the Philippines. Over the past four years I've run a paid community's inbox, coordinated patients for a healthcare company and managed my family's rental units. I'm AI powered and quick to adapt to new tech, so I move through high volume fast without letting accuracy slip.
350
emails a week I handled for a paid community, plus DMs and posts, with every reply checked
90+ WPM
typing speed, so replies don't keep customers waiting
GWA 1.38
Magna Cum Laude, earned while working online since my second year
4 years
of remote work, starting in my second year of college
About
When our family's rentals got hard to keep track of, I built an app that holds tenants, contracts, rent and repairs in one place. When collecting payments for my pickleball group got messy, I built Paddl. When the support inbox reached 350 emails a week, I set up a Claude workflow so every reply started from our real policies.
Well rounded
In my second year of college, I was assistant project manager at a startup, running copywriting, design and web development. I learned SEO so our sites would actually get found. I know how writing, design, tech and marketing fit together because I've had to make them fit.
My degree
Multimedia Studies took me through aesthetics, photography, videography, scriptwriting, philosophy, web development and programming, with a thesis in game development. I finished Magna Cum Laude, fully online, so working through technology has been normal for me for years.
Fast learner
I learned Unity, DaVinci Resolve, Base44 and Lovable from scratch and taught them to complete beginners. I've used AI since the GPT 2 days, so your team's tools won't take me long. If a stranger can follow my tutorials, your customers can follow my replies.
Outside work
I'm into psychology, philosophy, technology and astronomy, and I spend a lot of my free time learning things. That interest in people is also why I read a customer's message twice: once for what it says, and once for what they actually need.
Customer support
In each of these, real people were waiting on me, and money or health was on the line.
Sharan YouTube Media Group · Feb to Sep 2026
Peptide therapy healthcare company · 2024 to 2025
Family rental business · 2023 to 2026
Experience
Feb 2026 to Sep 2026
Community Manager and YouTube Researcher, Scriptwriter and Producer
Dec 2025
Freelance Academic Writer
2024 to 2025
Medical Records and Patient Coordinator
2023 to 2026
Property Manager
Jan 2023 to Jan 2024
Copywriter and Assistant Project Manager
Article Writer
Education
2021 to 2025
Magna Cum Laude, GWA 1.38. Team leader in group projects, thesis in game development.
Apps I built
Both are live. I built them with AI in Base44, then tested them the slow way: using them, finding what broke and fixing it.
Collecting payments for my pickleball group was a mess, so I built Paddl. Players schedule sessions, add each other by scanning a QR code and upload payment proof per session, and anyone can see who still owes. The court directory links straight to each court's booking page.
Open Paddl
Good fact checks exist, but they're buried in long articles nobody opens in the middle of an online argument. OpenSagot puts a claim, a short rebuttal and its sources on one screen, so anyone can paste an argument and get a sourced answer on their phone.
Open OpenSagotAlso built: a web app that tracks tenants, contracts, rent and maintenance for my family's rentals.
AI workflow case study
Around 350 emails a week, plus community posts and coaching DMs. At first I pasted each message into Claude by hand. It worked but it was slow, so I turned it into a system where AI writes the first draft and I check every one. I've built this with Claude's scheduled automations, and I've also worked hands on in Re:amaze for shared inboxes, macros and automated workflows.
I researched recurring questions once, condensed the answers and saved them as project files, so every draft starts from our real policies.
A scheduled Claude automation opened new emails, sorted them, for example a real invoice versus a screenshot, and drafted replies.
Self check prompts made the model review its own draft for wrong or outdated details first.
I read every draft, checked facts with a web search when needed, and only then sent it.
Replies went out a lot faster than pasting messages in by hand, and nothing went out unchecked. The drafts that need catching are rarely obviously wrong. They sound confident and are slightly out of date, which is why I never skip the review.
350 emails a week
Plus community posts and coaching DMs
Policy files
Researched once, saved for every draft
Claude drafts
Sorted and drafted on a schedule
Self check
Model reviews its own draft
Kent reviews
Every draft read and fact checked
Reply sent
Faster, and never unchecked
Design and UI/UX
Good support and good design solve the same problem: making the next step obvious. I studied UI/UX in my degree and have worked as a graphic artist, so the FAQs, price lists and help pages I write are laid out for people to actually follow.
Our UI/UX course final project, designed in Figma.
View in FigmaTypography, colors and reusable components for the same project.
View in FigmaA high-fidelity prototype for tracking carbon emissions and buying renewable energy and its services.
View in FigmaIntro video
A short video, so you can hear how I'd sound talking to your customers.
Skills and tools
Contact
If you're hiring for customer support, virtual assistance or operations, send me a message. I'm happy to start with a trial week, so you can see how I work before you commit.