A conversational AI platform, before generative AI
Product adopted by shopping centres, telco and energy operators, couriers and logistics companies. Conversational flows across web, WhatsApp and Messenger orchestrated from one place.
- My roleChief Operating Officer, then Senior Digital Product Owner — product lead
- DurationFrom 2019
- AreasConversational AI · Multichannel orchestration · WhatsApp · Messenger · Web
Before generative language models reached the market, automating a conversation meant designing it: intents, flows, exit conditions, handover to a human agent. The platform was built to do this across several channels at once, without the company having to run a different tool for each.
Companies already had a website, WhatsApp and Messenger, each managed separately and with no shared memory. A customer who wrote on one channel and then another started from scratch. And whoever answered had no way of knowing what had already been said elsewhere.
- 01
Understand
Working through real pre-sales, post-sales and customer support use cases across very different sectors: what counts as a recurring request for a shopping centre does not even exist for a courier.
- 02
Design and build
Product ownership of the platform: feature definition, prioritisation, backlog, coordination of development and design teams, release into production at client sites.
- Platform to orchestrate and manage conversational flows across website, WhatsApp and Messenger from a single point.
- Coverage of pre-sales, post-sales and customer support services.
- Adopted by shopping centres, telco and energy operators, directory services, couriers and logistics companies.
- The company behind the platform raised over €1 million across two funding rounds.
Anyone automating conversations today with a generative model runs into the same problems we ran into then: where automation ends and a person begins, what happens when the system gets it wrong, how context is carried across channels. The model solves language understanding, not service design.
The client is not named, for confidentiality. Sector and scale are real, as is everything else here.
Recognise the situation?
If something here resembles your case, send me a couple of lines. If it is not my ground, I will say so straight away.