How fast can you start?

A Discovery sprint runs 2 weeks and ends with architecture + execution plan. A pilot in real production lands in 4 to 8 weeks. Full systems in 3 to 6 months. We discuss timelines case by case before committing.

How do engagements work?

Discovery is a fixed 2-week scope. Pilot and Production are scoped to a specific outcome you sign off on. Fractional AI Lead is ongoing thinking-partner work with a 3-month minimum. No open-ended hours, no permanent dependency. Details on the call.

Does my data leave my infrastructure?

Not unless you want it to. On-prem deployments, dedicated instances in your VPC, or sovereign models running in Brazil are all on the table. LGPD / GDPR aware by default.

What stack do you work with?

Whatever your case actually needs. We're stack-agnostic: any leading LLM, any vector store, any orchestration layer, any frontend, any database. If you already have infrastructure, we integrate. If you're starting from zero, we help you decide. The right stack is the one that fits your case, not the one we prefer.

And shipping to production?

Depends on your infrastructure. We work with your existing cloud, CI/CD, and observability. If you don't have those set up, we help build them. The agent you see at /demo is a worked example of a full stack from scratch, ready to reuse or adapt.

What happens after delivery?

Your call: full handoff (documented and we walk away), ongoing support (SLA + on-call), or Fractional AI Lead continuing as a senior thinking partner. No lock-in.

The WhatsApp agent that becomes your operational spine

WhatsApp in Brazil isn't a channel. It's the country's interface.

Your grandmother books a medical appointment on WhatsApp. Your store manager sends photos of invoices on WhatsApp. The trucker unloading at your warehouse sends voice notes on WhatsApp. The complaining customer, the collecting supplier, the internal team resolving an incident at 10pm - all of it flows through there.

Every serious operation in Brazil is already WhatsApp-native de facto. What almost none of them have is the agent on the other side that turns those messages into action, data, ticket, response.

This post is about that agent. It's what Zechim delivers as operational foundation for companies that decided to stop using humans to do work software should do.

What it does

A single, configurable agent, integrated with Meta's official WhatsApp Business API, that:

Reads text messages, photos, audio, PDFs, documents
Understands context (who sent it, which conversation, which subject)
Queries your company's data (ERP, CRM, database, spreadsheet, whatever)
Acts: records a payment, opens a ticket, updates a status, schedules a visit, creates an order, answers a question
Escalates to humans when it involves decisions outside the standard, high cost, or risk
Logs everything in an auditable trail

One agent. Configurable to work as your customer-facing front OR as an internal team assistant OR both simultaneously with different rules.

Two modes of use

Customer-facing mode

Customer adds the company's number on WhatsApp. Sends whatever they want: "where's my order", "want to quote 50 units", "need to exchange", photo of a receipt, voice note explaining a situation. Agent replies. If it involves an out-of-standard decision (large discount, policy exception), it escalates to a human team with the whole conversation pre-contextualized.

Cases we've run in this mode:

Post-sale customer service in retail (exchange, return, status)
Service quoting from a photo (construction, workshop, technical quote)
Scheduling and rescheduling (clinic, salon, workshop)
Friendly collections and debt negotiation
First-line technical support

Internal mode

The company team sends messages to the agent on WhatsApp: "how many orders are still open to deliver today?", "how much revenue came in this week?", "generate the sales-by-rep report for the month", "open a maintenance ticket for supplier X, describe what happened in the note", "send this payment receipt I just photographed to finance".

Cases we've run in this mode:

Internal conversational panel (the owner who hates BI)
Field team sending visit or delivery info
Finance processing receipts sent by store managers
Inventory updating stock in and out via WhatsApp
HR answering internal policy questions from the team

Why WhatsApp and not your own app

You already have an app. It has 2 percent active use. That's how it works in Brazil (and probably in most markets outside Silicon Valley). The user doesn't install anything new. Doesn't open another dashboard. Doesn't learn a new UI.

They already use WhatsApp 80 times a day. That's where attention lives. Finding the user there is the difference between real adoption and "we launched and nobody used it".

Why "agent" and not "chatbot"

Chatbot answers. Agent acts.

Chatbot: "your order was shipped on the 15th, will arrive in 3 business days." Agent: "your order was shipped on the 15th, but the carrier reported a delay today - I opened a request for a proportional refund and I'll let you know when the credit is available. Want me to confirm?"

Chatbot: "to exchange, send the item to address X." Agent: "I opened exchange ticket 88472, generated a shipping label, sent it to your email. Drop off at any post office or request home pickup. Confirm?"

The difference is doing versus saying. Agent does. Agent reads, decides within what you authorized, executes, logs, notifies. Agent is software that replaces an operational human, not an interactive FAQ.

How we deliver it

It's not a shelf SaaS. It's built for your operation:

2-week Discovery - we read your processes, your data, your regulatory limits, understand exactly what the agent can and can't do. We leave with defined pilot scope.
4 to 8-week Pilot - a set of use cases running in small production. Meta's official WhatsApp Business API (no middlemen), integration with your systems, guardrails for human escalation.
3 to 6-month Production - expansion to the remaining cases, infra hardening, automated evals, documented handoff to your team.
Optional ongoing mode - Fractional AI Lead if you want to keep Zechim's seniority involved in the roadmap without hiring an internal AI team.

The agent becomes your property. Repositories handed over. Documentation handed over. No lock-in.

What you need to start

WhatsApp Business API (not Business App). We help set it up directly with Meta if you don't have it yet.
Data accessible via API from your ERP/CRM/database. If you don't have that, we help build the access layer.
Someone on your side with authority to define the agent's limits (what it can do alone, what needs to escalate).

You don't need an internal AI team. You don't need to migrate stacks. You don't need to approve an annual budget before starting.

If your company operates in Brazil, has customers or a team that already lives on WhatsApp, and you notice that 50 percent of the work there is repetitive enough for software to do - worth 30 minutes to talk about agent design for your case.