A market that talks on every channel
Sales and support teams talk to customers over WhatsApp, Instagram, phone and email — but managed it all in separate tools, losing context, history and opportunities along the way.
The proposal: an omnichannel CRM SaaS with centralized support, workflow automation and marketing/sales funnel management — one place where conversations become revenue.
From Figma to architecture
The product was designed before it was coded: journeys, screens and a design system in Figma, tested against the question that guides everything — does this increase the user's perceived value?
In parallel, the technical foundation was chosen to sustain growth without a rewrite: multi-tenant architecture, asynchronous messaging and a lean, productive stack.
- Journeys and design system in Figma before the first line of code
- Multi-tenant architecture designed to scale per customer
- Stack: React, Node.js, TypeScript, Redis and Express
A multi-channel messaging architecture
The heart of the product is a single message pipeline: every channel — official WhatsApp API, Instagram, VoIP via Twilio — flows through the same path, with queues and workers absorbing spikes without dropping events.
Every external integration was treated as a first-class citizen: idempotent webhooks, retries with backoff and clear contracts between services.
- Official WhatsApp API, Instagram and VoIP telephony (Twilio) in one pipeline
- Redis-backed queues to absorb spikes and guarantee reprocessing
- Idempotent webhooks and backoff retries across all integrations
Scaling without losing a single message
In a customer-service product, a lost message is a lost customer. Concurrency was handled with per-conversation ordering guarantees and parallel processing across conversations.
Observability became a discipline: metrics, structured logs and end-to-end alerts — to find the root cause before the customer feels the symptom.
- Guaranteed ordering per conversation, parallelism across conversations
- Metrics, structured logs and end-to-end alerts
- Analytics dashboards that stay fast under heavy volume
Perceived value in every detail
Acting as PO and developer at once, every feature starts from the user journey — not the ticket. The most complete example: a marketing tracking module conceived and implemented end to end.
From concept to deploy: visitor and lead monitoring, conversion journeys and campaign attribution inside the CRM, integrated with Meta Ads and Google.
- Marketing attribution module integrated with Meta Ads and Google
- Conversion journeys and lead monitoring inside the CRM
- Dashboards and reports that drive decisions, not decoration
$1M ARR — and the lessons that remain
The product crossed seven figures in annual recurring revenue, with a growing customer base and thousands of conversations flowing through the pipeline every day.
Three lessons I carry into any product:
- 1Product is perceived value — users don't buy architecture, they buy the experience it sustains.
- 2Scale is discipline, not accident: queues, idempotency and observability are day-one decisions.
- 3The detail is the product: UX, latency and reliability are the brand that lasts.