AI builds

LLM features wired into real workflows.

With permissions, usage caps and a person in the loop. Most of the interesting work is in what the model is not allowed to decide.

At XtendOps

Agents that work alongside customer-support teams.

Internal Smart Agents Assistants that handle customer interactions and automate repetitive support workflows. Claude + OpenAI APIs, Node.js
Internal Agent Assist Suggests replies and next steps to support agents inside the tools they already use. SvelteKit, adapters per support platform, Redis queues, Chrome extension
Internal Agent evaluations Scores agent answers so prompt and model changes are measured, not guessed. Evaluation runs in the customer portal

AI that saves landlords and teachers real hours, metered per plan.

Live Ask BahaySuite Answers questions about bookings, tenants and money in English or Taglish, using read-only tools gated by role. OpenAI tool calling, NestJS
Live Payment proof check Reads an uploaded GCash or bank screenshot. Code, not the model, compares the amount and reference. OpenAI vision, BullMQ job
Live Message drafts Drafts reminders, notices and guest replies that the landlord edits before sending. OpenAI, per-org monthly cap
Live Lesson plans & notes GuroSuite drafts lesson plans and per-learner progress notes from recorded ratings. OpenAI, Next.js
Live Guest reply drafts Drafts answers to guest messages for The B's Humble Abode. LLM endpoint, SvelteKit

Models that run in the browser, on the device.

Live ArnisAI pose scoring Tracks joints from the webcam and scores martial-arts strikes and stances in real time. TensorFlow.js MoveNet, SvelteKit