~/antcodelabs $ ls ./apps
We ship our own
software too.
Ten apps we designed and built. Some you can sign up for, two are open source, and the rest run in-house or in production for the business they were built for. They are where the ideas we bring to client work get proven: agents with real tools, MCP servers, voice and WhatsApp workflows, and AI that sits inside the product instead of beside it.
Partyyy.party
The Party Operating System — for parties that deserve better than a group chat.
Partyyy runs the whole party, from the first invitation to the photos afterwards. Hosts get one guest book that carries across every party they have ever thrown, RSVPs that actually come back, and a control desk with every number live. Each guest gets a personal, password-free party page — and badges, karma and votes that turn a guest list into a community.
- One guest book across every event, with partners linked and CSV import
- Magic-link RSVPs, date polls and custom questions — no accounts, no passwords
- Email, WhatsApp and SMS invitations, with a chase list for the quiet ones
- Badges, karma, voting and polls, plus a live TV wall for the room
- Seating, beds, meals, activities and a budget ledger that adds up
- Thirteen full guest themes, an AI party planner and an MCP server
YouMeBoth.app
Two lives, one plan. Every item is mine, yours or ours.
YouMeBoth is one place for the life two people share: the plans, the calendar, the shopping list, the holiday, the bills and the passport that runs out next spring. Most apps were made for one person and then given a share button; this one starts from two. Each of you has a colour and the things you do together wear a third, so one look at the week shows where things stand without anyone asking.
- “When are we both free?” answered across both diaries, work calendars included
- To-dos that repeat themselves — shopping, chores, the boiler service
- Trips shaped as places and nights, with bookings, checklists and spend
- Money together and apart: imported statements, fair splits, settle-up
- Renewals that never lapse — passports, MOT, insurance, the mortgage fix
- Works with your AI assistant over MCP — plan a trip or sort the spending by asking
WAHA Analyser
Years of your own WhatsApp chats, finally readable — built on WAHA.
WAHA Analyser is a self-hosted analyser for your own WhatsApp history, designed to work with WAHA, the WhatsApp HTTP API. It syncs live chats through your own WAHA session and imports export files, then gives every chat a statistics page and Claude-written reports on who says what, how, and how it has changed. “Who’s this?” finds anyone by name or number and briefs you on how you know them, and the social graph scores every tie from 0 to 100 to find your circles, your bridges and the friendships that are quietly fading.
- Built for WAHA: live sync and sending through your own WhatsApp session
- Chat analytics and AI reports: activity, reply times, personality, sentiment
- Ask a chat anything, or simulate how someone in it would reply
- “Who’s this?” briefings: key facts, shared groups, what to ask next time
- Social graph in 3D, with circles, bridges and rising or fading ties
- MCP server, so an assistant can search and analyse the archive
Ant Air
A self-hosted log of every flight you have ever taken.
Ant Air turns a booking history into maps, charts, an aircraft registry, achievement badges and AI-written insights. Import your flights, let it enrich them with aircraft and route data, and it works out the rest — including which tail numbers you have flown on more than once. It ships with an MCP server so an AI assistant can read and edit the log, and an optional native iOS companion app.
- Dashboard with a boarding-pass next-flight card and a great-circle route map
- Imports from TripIt, Google Flights and BA Flightpath, plus boarding-pass OCR
- Aircraft registry built from the registrations you have flown on
- Trips and timeline across flight, rail, road and sea legs
- Audits for duplicates, missing legs and city pairs, with one-click fixes
- AI insights and an MCP server exposing every flight, trip and statistic
Ant Radar
An air-traffic-control console for the aircraft over your own roof.
A full-screen, ATC-style kiosk for the aircraft your own ADS-B receiver is tracking. It polls the receiver once a second, enriches every contact with registration, type, operator, route and photo, and streams the result to twenty auto-rotating views — while keeping a durable history of everything the station has ever seen. Built to sit on a wall display, and viewable on a phone.
- Radar scopes with range rings, trails, airspace and an orbiting 3D view
- Flight strips, closest-aircraft deep dives and spotter photos
- Emergency Watch that takes the screen when something squawks 7700
- Live instruments: altitude profile, signal vs range, heading rose
- Weather and runways: an animated sky drawn from the nearest METAR
- Durable history: daily trends, first-seen log and a type “life list”
Ant Dash Home Lab Dashboard
One screen for the whole home lab — and a wall display to match.
Ant Dash puts an entire home lab on one screen: hypervisors, containers, Kubernetes, single-board computers, desktops, network gear, storage, databases, backups and the odd IoT box. Around twenty integrations each get a page of their own, every page updates live, and CPU, memory and network history is kept for a day. On top of that sit alerts, security findings with the command that fixes them, pending updates and every backup job in one list — plus a kiosk mode that rotates the whole estate across a wall panel.
- About twenty integrations: Proxmox, Docker, Kubernetes, Pi-hole, UniFi, Cloudflare, NAS and more
- Live everywhere: pages update over a server-sent event stream, with 24 hours of history
- Alerts for outages, failed or stale backups, findings and expiring certificates
- Security findings across every host, each with a fix command and an ignore switch
- Backups and updates in one place, with one-click upgrades on the small boards
- Kiosk mode and a desktop agent: a rotating wall display, and Macs and PCs reporting in
Jessica
The AI assistant that lives in our WhatsApp — and our Telegram.
Jessica is the assistant we message every day. WhatsApp and Telegram feed one pipeline — the same personas, memory, approval queue and admin console — so she is the same Jessica on either, and replies wherever you wrote to her. She keeps track of what is coming up, sets reminders from a chat message, looks things up in Partyyy with real tools, and remembers what she is told.
- One pipeline, two channels: WhatsApp and Telegram, with history merged per person
- Modes: personas with their own prompt, model and temperature, edited live
- Long-term memory in private and shared pools, one fact at a time
- Approval queue: strangers wait for a yes before she says a word
- Reminders set in conversation; voice notes transcribed, photos described
- Send API with a cold-send gate, carrying Partyyy's WhatsApp invitations
Lily
An AI property manager who picks up the phone — and WhatsApp — at any hour.
Lily is an agentic AI property manager for residential block management. Residents and leaseholders reach her by phone, from a call button in the browser, or on WhatsApp with text, photos, videos and voice notes. She knows each building from a context store the property team edits themselves, logs repairs and quotes the reference back, and hands the conversation to the right person when a human is needed.
- Voice, browser call and WhatsApp — one assistant, around the clock
- Understands voice notes, photos and videos, not just text
- Knows each building: access, bin days, contractors, concierge hours
- Logs repairs with the ticketing system and quotes the reference back
- Hands over to a person or team when needed, then stays out of the way
- Every call transcribed and summarised to the team, recording attached
Block Management Hub
One operations platform for a managing agent's staff, residents and contractors.
The Hub runs the day-to-day of a residential block managing agent. Residents raise maintenance, enquiry and accounts requests through a guided wizard or by email; staff triage them, instruct works and close them out on one threaded timeline; contractors accept jobs, quote and book visits from a portal of their own. Every block also carries a statutory compliance plan and a planned-maintenance plan, and larger projects run through a staged major-works module from Section 20 consultation to the defects period.
- Request lifecycle on one timeline, with resident, internal and contractor threads kept apart
- AI triage and resolution plans: category, priority and next steps, with approval gates
- Contractor portal: job inbox, quotes, visit slots and insurance tracking
- Compliance and planned maintenance per block, generated from a rule library
- Major works: Section 20, tenders with AI bid analysis, budget, snagging, defects
- Resident portal with magic-link sign-in, notices and building-wide visibility
AI Invoice Manager for Block Management
Supplier invoices read, matched and bundled for the accounts system — with a person signing each one off.
The Invoice Manager takes the supplier invoices a managing agent receives for its blocks and prepares them for the block-accounting system. AI reads each document and extracts the supplier, dates, totals and line items, then matches it to the right block, supplier and service with its confidence and reasoning shown. A person reviews the result beside the original PDF and confirms; every correction becomes a rule, so the same mistake is not made twice. Confirmed invoices are bundled and exported as the original PDFs plus one import-ready CSV.
- AI extraction from PDF or Word: totals, dates, line items and VAT
- Automatic matching to block, supplier and service, with confidence shown
- Side-by-side review: the original document next to the editable fields
- Learns from corrections, with an audit view of every learned rule
- Duplicate and chaser detection, and line-items-versus-total checks
- Export bundles: PDFs plus CSV for the accounts system, checked before download
Want something like these, built for you?
Everything on this page started as a rough idea and a working prototype a few weeks later. That is the same loop we run for clients.
Screenshots show made-up demo data — invented guests, couples, flights, chats, servers, buildings, residents and suppliers. The one exception is the radar, which shows live public ADS-B traffic. Each app is shown in its own light or dark theme to match this page, where it has both.