Work
Three systems in production, built from scratch.
Not a demo, not a mock-up: three systems running in production that people depend on every day.
- 84%
- replies sent by the agent, with no human involved
- 868
- messages sent automatically over 34 days of service
- 66
- properties handled by the same agent
- 19
- languages understood and spoken
Calia Labs
A decision system for finance and insurance.
The context
Finance and insurance, where every decision commits the company and has to be justifiable long afterwards.
The problem
A decision has to be made in real time, and proven months later. Most systems do one or the other.
What we built
An engine that intercepts every decision as it is made, keeps the full trace of it, and treats the regulatory constraint as a value in the system rather than as a check bolted on afterwards.
Tools used
Real-time decisions · Traceability · Built-in compliance
MyKonci
A complete platform, built from scratch and run in production.
The context
Rental management: several property management companies served by the same platform, each with its own clients and its own data.
The problem
A property management company that grows ends up running its properties, its guests, its billing and its compliance across several tools that don't talk to each other.
What we built
A single platform: bookings, multilingual guest messaging, voice agent, billing and subscriptions, client dashboard, GDPR compliance. Nine services in production.
Tools used
Multi-client platform · Voice agent · Dashboard
Nomad's Co
A property manager that answers its guests day and night.
The context
Short-term rentals, sixty-six properties, guests who write in their own language at any hour.
The problem
A question left unanswered in the evening becomes a bad review the next day. And arrival, departure and reminder messages went out by hand, so sometimes late, sometimes twice.
What we built
An agent that reads the history of the stay, answers in the guest's language and hands over to a person as soon as a subject deserves it, paired with a chain of messages that goes out at the right moment from the booking.
The result
84% of the replies sent to guests come from the agent, with no human involved, and 868 messages went out automatically without anyone having to write them. Measured over 34 days of service, 66 properties, 19 languages.
Tools used
Conversational agent · WhatsApp · Booking synchronisation
The next one could be yours.
Describe what you're missing, even in one vague sentence. That's exactly what the scoping is for.