Projects

NebuFit

Challenge

Fitness apps lean on generic templates instead of adapting to a user's real body data, bloodwork, and progress, and users want conversational coaching rather than static charts.

Solution

Built computer-vision body analysis, AI-generated workout and nutrition plans, and bloodwork processing on a RAG system for contextual guidance, orchestrating multiple LLM providers (Claude + GPT-4o + ElevenLabs) with SSE streaming and automatic failover for resilient voice coaching, backed by full Stripe subscription and entitlement management.

Impact

Shipped a production-ready AI coaching platform that delivers personalized, real-time guidance with resilient uptime and monetization built in from day one.

Case Study

Creative OS

Challenge

Writers, filmmakers, musicians, and other creators wanted AI assistance across screenplays, lyrics, and notes, but typical AI chatbots can't reach into local files or hold onto creative context over time — and sending unfinished creative work to the cloud, or being locked into one fixed set of AI tools, wasn't an option either.

Solution

Built Creative OS as a local-first Electron/React desktop app that works less like a chatbot and more like an operating system for creative work — indexing Final Draft screenplays, PDFs, lyrics, notes, and research through a RAG pipeline (ChromaDB + SQLite) for semantic search, auto-tagging, and Claude-powered rewriting with persistent context, including native .FDX editing that preserves screenplay formatting, on an extensible MCP server architecture.

Impact

Gave creators AI-powered editing, search, and organization across their entire local creative library that persists across sessions instead of resetting, with a tool ecosystem that grows without touching the core app.

Case Study

Orient Champions

Challenge

Orient Electronic needed a nationwide field-sales incentive program spanning a four-level hierarchy (National → Zone → Branch → FSM), with real-time sync into their existing SAP ERP — while staying resilient to SAP's own downtime and response-time swings.

Solution

Inherited an already-live but poorly-built implementation — tightly coupled, dead-code-heavy, no real state management — and refactored it: decoupled the React Native frontend and Nest.js backend into clean modules, introduced Redux, rebuilt and retimed the cron-job pipeline for accurate pre-business-hours results, replaced hardcoded screens with real APIs, fixed push notifications and the App Updater, and redesigned the UI.

Impact

Eliminated roughly 90% of the bugs the client used to report daily — the client now calls to discuss new features rather than fix issues — and gave field agents a faster, redesigned app they use without friction.

Case Study

UNITY Desktop App & Admin Portal

Challenge

A healthcare organisation needed one desktop workspace that unified browsing, credentials, notifications, and patient call handling — capabilities otherwise spread across several disconnected tools — with an admin layer that could reconfigure each role's navigation without a new build.

Solution

Built an Electron + Nest.js desktop app whose role-based sidebar, embedded browser with credential auto-fill, socket-driven notifications, and incoming-call handling are all fetched per user at login, paired with a Next.js + Nest.js admin portal where staff drag-and-drop each role's navigation and watch live usage stats.

Impact

Delivered one unified, role-aware desktop tool whose navigation and permissions are configured centrally, replacing what would otherwise be several separate applications for organisations and staff spanning dozens of logins.

Case Study

Houzi

Challenge

Real estate businesses running the Houzez WordPress theme had a fully functional website but no native mobile presence — building a separate app from scratch for each client meant duplicating business logic, API integrations, and UI work, with no easy way to customize branding and features per agency without forking the codebase.

Solution

Built Houzi, a Flutter app that talks to any Houzez-powered WordPress site over its REST API using Dio, Hive, and Provider — with a thin white-label app shell layered over a shared package (via customization hooks and a remote JSON config) so branding and features can be tailored per client without forking the codebase — plus the HouziBuilder desktop configurator and the Houzi-REST-API WordPress plugin.

Impact

A single Flutter codebase now powers native iOS and Android apps for real estate clients across the Houzez ecosystem, reaching 53,000+ theme owners with a 4.84★ rating, 500+ CodeCanyon sales, and 5,000+ Play Store downloads.

Case Study

Linked Golf

Challenge

Golfers had no streamlined way to match availability with friends, connect with new playing partners, and subscribe to golf-course events — and golf-course owners had no way to create or monetize those events themselves.

Solution

Inherited an already-live but poorly-integrated implementation — tightly coupled Bloc state across every core screen, dead backend code, non-functional chat — and refactored it: decoupled the state layer and backend, fixed chat/applink bugs, and added event creation, Stripe-billed subscriptions, team management, and OTP-verified team joins.

Impact

Delivered a stable, noticeably faster app plus a new paid-event capability for golf-course owners, on both the App Store and Google Play.

Case Study

Exodus App

Challenge

Blue Sky needed a full rebrand to its new Exodus identity, and syncing health vitals from a wearable device into the app was slow and often failed.

Solution

Delivered a full rebrand from Blue Sky to Exodus, and packaged the wearable's vitals SDK directly into the app build in place of a failure-prone per-scan API call.

Impact

Relaunched under its new identity with faster, more reliable wearable vitals syncing for the doctors and patients who depend on it.

Case Study

HouziBuilder

Challenge

Houzez theme owners had to hand-edit a structured JSON file to configure the companion Houzi app's branding, layout, and content — with no visual preview and no way to verify licensed access.

Solution

Built HouziBuilder, a licensed Flutter desktop app that replaces manual file editing with a guided, form-driven builder and live device preview, exporting a configurations.json the Houzi app reads at runtime.

Impact

Turned a code-level JSON editing workflow into a guided desktop tool with live previews, cutting configuration errors and setup time while keeping license verification and WordPress integration built in.

Case Study

Techtiz Corporate Website

Challenge

Techtiz needed a large multi-vertical marketing site — services, industries, a separate government-subcontracting audience, and a blog — while migrating off a legacy static HTML/CSS/JS site without duplicating markup or losing SEO ground.

Solution

Built the site in Astro with a constants-first architecture, a two-tier system for repetitive service/industry pages, a fully isolated SLED experience for the government audience, and a WordPress-backed blog that publishes without a redeploy.

Impact

A maintainable, SEO-optimized site where new pages follow a documented pattern instead of one-off work, content editors publish without a code deploy, and the SLED funnel stays fully separated from the commercial site.

Case Study

ATTN Agency Homepage

Challenge

ATTN wanted a fully native, on-brand agency homepage inside Go High Level — not an embedded page or iframe bolted onto the platform — built entirely from the builder's own Row/Column/Container/Text elements, with no custom HTML or CSS to fall back on.

Solution

Built the homepage entirely with Go High Level's native builder elements, then used Google Gemini to generate the custom visual assets — like the hero's themed video panel — that the native component library doesn't provide on its own.

Impact

A homepage that reads like a custom build while staying fully native to Go High Level, easy for ATTN's own team to keep editing with the same builder tools.

Case Study

Lead Response Automation — United Global Vanline

Challenge

United Global Van Line (UGVL) needed its entire lead-to-deposit funnel to run with zero tolerance for slow response or dropped leads, but voice, SMS, and voicemail automation were split across three disconnected outside vendors, and stalled leads or underperforming agents could go unnoticed for a week or more.

Solution

Consolidated all voice, SMS, and voicemail automation onto Aloware's native AloAi bots and Go High Level's Voicemail Drop, added HOT/WARM/LOW lead-temperature scoring that routes each lead to a role-matched Senior/Junior/Fresh closer, and layered in OpenAI call-QA scoring, a daily funnel-leak audit, a 3-minute HOT-lead safety net, and an automated Clover deposit-to-revenue pipeline.

Impact

One unified conversation thread per lead, HOT leads guaranteed a Senior closer, stalled leads and coaching gaps surfacing automatically, and deposit collection, refunds, and agent revenue crediting all reconciling end to end without manual work.

Case Study

AGENTC — Agentic Job Discovery Automation

Challenge

The client's Selenium-based scraper for finding and verifying decision-maker contacts across four industry niches was expensive to run and fragile to maintain, pushing the whole tool stack to $600–800/month with no room to add reliability or cost visibility.

Solution

Replaced Selenium with Apify actors and orchestrated the entire nightly pipeline in n8n — Hunter.io for decision-maker discovery, ZeroBounce and Clearout Phone for verification, NocoDB as the dedupe layer, and an OpenAI agent to summarize and email the results before business hours.

Impact

Brought monthly cost into the client's $200–300 target while covering four full niches a day, with dedupe, rate-limit handling, and credit monitoring that catch problems before a report goes missing.

Case Study

Capital Shack — Document OCR Processing

Challenge

Manually reviewing, categorizing, filing, and cross-referencing every incoming client document into spreadsheets was slow, repetitive, and error-prone at any real volume.

Solution

Built two linked Google Apps Script automations — one files and logs every emailed document into Drive and Sheets, the other runs each attachment through Document AI to classify and extract its data automatically.

Impact

Eliminated manual document triage and data entry entirely, so every incoming document is consistently categorized, filed, and logged without anyone opening it by hand.

Client confidentiality prevents sharing screenshots or product media for this project.

Case Study

Ramirez Legal Group — ClickUp Dashboard Sync

Challenge

ClickUp case-tracking data and the client-facing dashboard on Emergent Labs had to be kept in sync by hand, risking the two drifting apart.

Solution

Built a Zapier pipeline that watches ClickUp continuously and syncs case updates and new client records to the Emergent Labs dashboard automatically.

Impact

Keeps the case tracker and dashboard continuously synchronized in real time, with zero duplicate manual entry.

Case Study

Social Media Posting Automation

Challenge

A client needed AI-drafted captions, minute-precise scheduling, and hands-off publishing across six social platforms — without a person juggling six different upload APIs, six sets of OAuth tokens, or finding out about a failed post days later.

Solution

Built a suite of five linked n8n workflows backed by per-platform Google Cloud Functions: an AI agent that drafts platform-specific captions from a Google Form submission and holds them for human approval in a shared sheet, a minute-precision Google Apps Script scheduler, a webhook-triggered publisher that gates and uploads to YouTube, Facebook, Instagram, LinkedIn, Threads, and TikTok, a scheduled OAuth token-refresh workflow, an auth-callback workflow handling every platform’s handshake (including Threads’ extra compliance requirements), and a centralized error logger.

Impact

Replaced manual caption-writing, scheduling, and OAuth babysitting with a pipeline where the only manual step left is approving AI-drafted copy — everything else, from minute-precise publishing to token refresh to failure alerts, runs on its own.

Case Study

Outreach Email Classification Automation

Challenge

Every reply to an outreach campaign had to be opened and read by hand just to tell an interested prospect apart from a bounce, an out-of-office, or an unsubscribe request — and n8n's native Gmail Trigger gave no way to filter or rate-limit what reached AI classification.

Solution

Moved inbox polling out of n8n into a Google Apps Script that filters replies against allowed recipients, blocked domains/senders, and warm-up/unsubscribe keywords before queuing them to an n8n webhook at a rate-limited pace, where Gemini 2.5 Flash Lite classifies each one's intent and posts a formatted status to Slack.

Impact

Replaced manual inbox triage with automatic, rate-safe classification and instant Slack visibility on every reply, with filtering rules and classification behavior both tunable from a spreadsheet and a prompt instead of code.

Case Study

Zendesk AI Support Automation

Challenge

A support team's agents were spending time piecing together context by hand on every new ticket, every pending call recording, and every ongoing conversation before they could even start drafting a reply.

Solution

Built three linked n8n workflows — OCR-and-GPT-4o analysis of new tickets and their attachments, ElevenLabs transcription and summarization of pending call recordings, and full-history-aware response drafting for ongoing conversations — all logging into a shared Central Intelligence Hub database.

Impact

Every new ticket, pending call, and ongoing conversation now gets an instant AI-drafted summary and reply logged straight into Zendesk and the Intelligence Hub, before an agent even opens it.

Case Study

Kumquat — Talent Match

Challenge

Skill-sharing and consultation-booking apps — coaching, tutoring, telehealth — needed a proven starting point instead of a blank canvas, with auth, search, booking, payments, chat, and live video all designed from scratch each time.

Solution

Published Kumquat, an 81-page FlutterFlow template spanning 11 feature modules — auth, provider search, appointment booking with payments, chat, video calling, sessions, reviews, and a wallet — on a Firestore-ready data schema.

Impact

Published to the FlutterFlow Marketplace as a reusable starting point for skill-sharing and booking apps, cutting weeks of UI/UX design down to configuration and backend integration — 577 downloads on the marketplace to date.

Case Study

Foodii — Recipe Book

Challenge

Developers building recipe-sharing apps needed a credible, ready-made starting point rather than designing every screen from scratch — a nested recipe/chef data model across 57 feature pages, with no live backend to build against yet.

Solution

Published Foodii, a FlutterFlow template with custom Data Types mirroring a future Firestore schema, a shared component library, and App State variables simulating backend data so the full app could be built and demoed before Firestore goes live.

Impact

Published to the FlutterFlow Marketplace as a backend-ready recipe-app template — connecting real Firestore collections later needs minimal rework, since the schema and UI bindings already exist — with 537 downloads to date.

Case Study