Esval Industrial
WhatsApp as a serious B2B channel
Esval Industrial manufactures plastic packaging for cosmetics brands in Colombia, and most of its B2B pipeline starts on WhatsApp. We built EVA, an AI sales assistant that knows the catalog, qualifies the lead, detects promises and payment proofs, and hands the conversation to the right advisor through WhatsApp’s interactive buttons — every step auditable, every setting editable from the panel.
Duration
3 months · delivered August 2026
Completed
August 17, 2026
Reply, lead extraction, commitment, payment, frustration and a shared classifier
Personality, hours, cooldowns, credentials, advisors and legal data — no code deploy
The schema evolved with zero manual database edits
Webhook, panel, scheduled jobs, training, settings and auth
Esval is a Colombian industrial SME selling plastic packaging to cosmetics brands.
WhatsApp is its number-one lead channel, and until 2026 it was handled by three people answering manually, one chat at a time, during business hours. After hours, leads went cold. There was no centralized history, no audit trail and no way to see who was handling what.
- — The main channel had no business tools: advisors answered from their own phones, with no shared history, reassignment or workload view
- — Most questions were about the catalog — collapsible tubes, bottles, caps, materials, minimum order quantities, barrier by material — and nobody knows the full catalog by heart
- — No prior infrastructure: no Meta Business setup, no approved templates, no data-protection policy under Law 1581, no catalog embeddings, no panel
- — Human judgment was non-negotiable: commitments, payment proofs, frustration and complete customer data must reach a specific person, with context
Eight systems in one platform at app.
esval.com.co, delivered with a formal sign-off in August 2026 and operated by Esval’s marketing lead and sales advisors without depending on us.
- + EVA, the AI agent · six AI analyses run on every inbound message: the catalog-grounded reply plus lead extraction and commitment, payment and frustration detection
- + Escalation router · the customer picks their advisor with WhatsApp interactive buttons, or EVA routes by customer type; only the chosen advisor is notified, by WhatsApp and email
- + B2B lead board · five stages built for Esval’s sales process, automatic assignment and a personal view per login
- + Catalog knowledge base · product catalog, technical sheets and sales procedures embedded in pgvector, with vision-based reading of image-only PDFs and self-service CSV catalog updates
- + Operator-editable panel · 95 settings, from the assistant’s personality and sales mode to credentials, holidays, advisors, campaigns and legal data
- + Audit, email and compliance · every escalation, takeover and button tap recorded; branded transactional email; public privacy, terms and data-deletion pages under Law 1581; an embeddable web widget
Inside the build.
Brand, interface and system — shown as they shipped.
The interface is the small part.
EVA is not a chat widget on top of a model. It is 49 migrations, 61 endpoints, six AI analyses per message, 45 versioned operations scripts, an hourly scheduled job and a panel with 95 settings. About 85% of the code is backend, AI pipelines, orchestration and audit.
Strictly idempotent
Meta resends a message if it does not get a fast 200 OK. Every inbound message is keyed by its WhatsApp ID, so a retry never generates a second AI reply.
Advisor choice with a safety net
If the chosen advisor is removed before the customer taps the button, the system falls back to the next active one. If the customer types instead of tapping, EVA closes the prompt and keeps talking.
Independent cooldowns
Payment, commitment and escalation alerts each have their own window, so an active conversation never floods an advisor’s phone.
Credentials checked before they are saved
Gemini, Meta and Resend keys are tested against the real services before saving, with clear messages for an expired key, exhausted credits or a quota limit.
Hourly jobs on GitHub Actions
The hosting plan only allows daily cron jobs, so inactive conversations are closed every hour by a GitHub Actions workflow that calls a secret-protected endpoint.
Hybrid AI Coding
A senior WebMinds engineer steers AI agents (Claude Code) for implementation, with QA on every change and a simulator that runs the exact production pipeline. Architecture and every call with the client stay human.
Technologies & Tools
The tech stack powering this solution
Beyond the numbers, we delivered a system that transformed how this client operates.
A real B2B pipeline on WhatsApp
What used to be a manual, after-hours black hole is now a measurable pipeline: catalog-grounded replies, every lead on a board, every escalation logged, and the right advisor brought in at the right moment.
The business in the driver’s seat
With 95 editable settings, Esval changes the assistant’s personality, rules, cooldowns, credentials, advisors and legal data on its own. Engineering stays with the engineers; the business stays with the business.
Auditable from day one
Every escalation attempt, manual takeover and button tap is recorded with its outcome per channel, and customer data is handled under Colombian data-protection law.
- May 2026Project confirmed: Esval brand, proposal and data model
- Jul 2026Catalog ingestion with vision OCR, branded email per alert type and legal pages under Law 1581
- Jul 31, 2026Customer-driven advisor selection with WhatsApp interactive buttons
- Aug 2026Self-service catalog upload and real-time unread badges
- Aug 17, 2026Formal delivery v1.0, user manual and panel-editable settings
- Since thenIn continuous operation with WebMinds support
Ready to achieve similar results?
Let's discuss how we can transform your business with a tailored digital solution.