Animated diagram: nine workflow routes converge on one junction. One amber packet that does not fit arrives, the routes bend around it, and the junction resolves into the Lucenith spark.
Custom CRMs, dashboards, client portals, AI workflows and integrations.
The amber packet is the exception. The system makes room for it.
Intake, rules, result. We start from the message that goes wrong.
Step 1
Intake
A message arrives. Nobody has to be awake for it.
Step 2
Rules
Checks run in order. When one fails, the system asks once, in the same thread.
Step 3
Result
One clean record. A person only hears about it if the system could not decide.
- Order#214, 20 × 19L
- DeliveryThu 10:00, route 3
- InvoiceINV-0392 sent
Pick a failure and watch the order route around it. This illustrates how we design exception handling. It is not a live log.
- Enquiry (waiting)
- Check (waiting)
- Order #214 (waiting)
- Record (waiting)
- Same enquiry, 11:49 pm (waiting)
- The first enquiry arrives at 11:42 pm.
- The same message arrives again seven minutes later.
- The check sees the same number and the same order.
- It merges into the open order. No second order is created.
Each began as an inbox, a spreadsheet or a phone call. The struck line is what it replaced.
01Hiring automation
AI interview automation
Before: Résumés piling up in a Gmail inbox
80% less time screening applicants.
How it works: A job email is scored by OpenAI and ranked in a sheet. A form is then sent and the interview is booked.
Incoming job emails trigger an n8n flow that scores each résumé with OpenAI, ranks applicants in a sheet, sends a personalised form and books the interview.
- n8n
- OpenAI
- Google Sheets
- Gmail API
02Website
Glacier water delivery
Before: Orders taken by phone call
Live in 72 hours.
How it works: The order form replaces phone orders. It sends the order to the owner's WhatsApp and the contact form to Appwrite.
Three-page site where orders go straight to the owner's WhatsApp and the contact form lands in Appwrite.
- Next.js
- Appwrite
- Vercel
03CMS platform
Study Times
Before: A developer needed for every edit
One admin panel for everything.
How it works: One admin panel edits destinations, reviews, consultations and the team, with no developer involved.
Visa consultancy site with a full admin panel: destinations, reviews, consultations and team, all editable without a developer.
- Next.js
- Appwrite
- Google Drive API
04Lead generation
Targeted contact scraper
Before: Copying contacts by hand
150+ leads in week one.
How it works: On a schedule, a Python scraper collects emails and phone numbers from niche sites into an outreach sheet. Week one produced 150-plus leads.
Python scraper that collects emails and phone numbers from niche sites and fills an outreach-ready sheet on a schedule.
- Python
- n8n
- Google Sheets
05Enterprise CMS
Corporate services website
Before: A deploy for every page change
Editors publish on their own.
How it works: Editors publish in Directus and the ASP.NET Core site reads it. There is no deploy step.
Directus feeds an ASP.NET Core site with dynamic service and industry sections. Nobody touches code to change content.
- ASP.NET Core
- Directus
- Vite
06AI video pipeline
Egocentric video analysis
Before: Hours of footage watched by hand
Video in, written summary out.
How it works: Video is uploaded, indexed by Azure Video Indexer, summarised with GPT and returned as text. Key Vault holds the secrets throughout.
Azure pipeline that indexes uploaded video, summarises it with GPT and keeps every secret in Key Vault.
- Azure Video Indexer
- Blob Storage
- OpenAI
07Platform rescue
TheamLab 3D order system
Before: A quoting system that kept failing
Quoting back online.
How it works: The customer portal, the admin portal and quoting each went from failing to working.
Found and fixed the bugs across the customer and admin portals of a 3D-model quoting platform built on Cofoundry CMS.
- ASP.NET Core
- Cofoundry CMS
What we build
Six kinds of work. Most projects mix two or three; we choose code, low-code or AI per problem.
- Workflows in n8n and Python that read the inbox, score, sort, schedule and follow up while nobody is watching.Closest case: AI interview automation
- Leads, clients and jobs in one screen the team opens every day. Built on your data, not a template.Closest case: Study Times admin panel
- Client portals and back-office tools shaped around how work already flows through your team.Closest case: TheamLab order system
- Sheets, WhatsApp, Gmail, calendars and CRMs passing records to each other, with no copy-paste step.Closest case: Glacier WhatsApp ordering
- Vercel, Docker, Azure or IIS, with CI/CD so every release runs the same way.Closest case: Egocentric video pipeline
- Interfaces designed around the work your team already does. Prototyped first, so you approve the design before we build it.Closest case: Glacier site, live in 72 hours
Four steps, in order. Each one ends with something you can click or sign.
Step 1
Discover
A working session on your goals, users and the tools you already pay for. It ends with a written scope, not a deck.
Step 2
Design
A clickable prototype and a data model. You approve both before any code exists.
Step 3
Build
Short sprints with a demo on a staging link every week. You can see the work at every step.
Step 4
Run
Launch, handover and training. Then we measure what changed and tune what didn't.
"I brought an idea. Lucenith delivered a complete, scalable platform. From dynamic admin panels to custom integrations, everything was tailored to our workflow. Their clarity-first approach made it stress-free."
"Lucenith helped us launch a professional website that looked great and delivered exactly what our business needed. The WhatsApp integration made order handling effortless."
What doesn't fit your current software?
Request a free 30 minute call. Bring the part of your week that never fits a tool; we map it on the call and send a written scope within two days.
Or write it down
Fields marked with an asterisk are required.
Good first projects
- Replacing a spreadsheet or an inbox with one system (Study Times, AI interview automation).
- A site or portal that sends orders and leads straight into your tools (Glacier).
- A platform that keeps failing and needs fixing (TheamLab).