WhatsApp Business Operations SaaS

The situation

Conversations, payments and orders live in separate systems.

A business receives customer conversations on WhatsApp, takes payments through M-Pesa, and manages orders across spreadsheets and chat. A single customer query can require checking three systems before anyone can say whether an order was paid or what should happen next.

Building on the existing inbox and customer system is the right starting point rather than replacing everything. The working inbox, contact, conversation and sales features are kept, checked, and wrapped in a reliable business system.

The product's core idea: businesses should be able to hand routine customer work to automated assistants while keeping control over money, customer data and important decisions. The automation organizes real business services — it is not the final word on payments, orders, permissions or money calculations.

Our approach

The product is built in steps:

1. Step one — shared inbox and business records. Dependable WhatsApp messages, security across customers, contacts, orders, permissions and history.
2. Key difference — M-Pesa matching and dispute handling. Checking transactions, matching them to orders, looking into exceptions and managing careful solutions.
3. Automation — sales and operations assistants. Checking leads, reviving stalled sales, answering routine questions and running approved steps.
4. Reports — forward-looking operations and reporting. Finding exceptions, measuring business impact and suggesting or starting approved fixes.

The first release solves one narrow, expensive problem very well: customers contact a business on WhatsApp, the business receives or expects payment through M-Pesa, and the system connects the conversation, payment and order to the right result.

What we built

Step one — shared inbox and business records

A system serving multiple customers that keeps the existing inbox, contact, conversation and sales features. Shared WhatsApp inbox, contacts, orders, permissions and history, with reliable WhatsApp messages and customer security.

M-Pesa matching and dispute handling

Transaction checking, order matching, exception review and careful resolution. Exact matching (receipt, merchant reference, expected amount, time window, customer details, invoice reference) runs before automated reasoning. Matching results are: matched, partly matched, unmatched, unclear, amount mismatch, possible duplicate, or needs review — automated reasoning explains differences and suggests options but cannot confirm an unclear payment.

Automation — sales and operations assistants

Automated assistants check leads, recover stalled sales, answer routine questions and run approved steps. Assistants work within clear tool rules and access levels: level 0 watch, level 1 help, level 2 do routine work, level 3 limited independence. The first release runs money steps at level 1 and low-risk support tasks at level 2.

Reports — forward-looking operations

Exception finding, business-impact measurement and approved-fix suggestions. Revenue and operations dashboards with allowed metrics, customer-separated questions and metric definitions shown with every report.

Engineering ideas

• One source of truth: payments, orders and customer records have one main database version.
• Handle things at least once: automatic updates and jobs never create duplicates — repeat callbacks never make double allocations or order releases.
• Lasting workflows: every workflow's state, what went in, what came out, retries and approval status are saved.
• Clear permissions: the assistant cannot use a tool just because it decides it would help.
• Fail safely: if payment checking or an important connection is unavailable, the system does not pretend it worked.
• Multi-signature support: a manager can use a shared-signature wallet, connected through wallet links.

Outcome

Customers contact a business on WhatsApp, pay through M-Pesa, and the system connects the conversation, payment and order to the right result. Routine work goes to automated assistants while money, customer data and important decisions stay with people.