CelcomDigi launches an AI agent for SMEs. The price is still missing

CelcomDigi has moved its Sophia AI proposition from an internal automation story to a commercial service for Malaysian businesses. It describes the product as a configurable digital workforce that can work across existing systems instead of answering questions in a separate chat window.[1]

Its practical example is invoice-to-payment. CelcomDigi names six process verbs: read invoices, validate them, flag exceptions, route issues, trigger approvals and update systems. It says it deployed more than 400 automations internally, but that figure is not evidence that an SME will achieve the same outcome or return on investment.[1]

Sophia AI is positioned as a configurable digital workforce for Malaysian enterprises and SMEs, not a single off-the-shelf chatbot. CelcomDigi says it can be adapted to workflows across finance, retail, healthcare, manufacturing, logistics and the public sector.[1]

The company points to more than 400 automations already used internally. That operating history is relevant because enterprise automation usually fails at the handoffs between systems, policies and people, not at the moment a model produces a fluent answer.[1]

Invoice processing is a useful test case because the work is repetitive but not simple. A system may need to read different layouts, validate supplier details, compare an invoice with an order, detect an exception, request approval and update a finance platform without releasing the wrong payment.[1]

An AI agent can reduce manual copying, but the buyer still needs rules for uncertain fields, duplicate invoices, unusual amounts and failed integrations. A demo that handles the happy path is not enough. The value appears only when the exception queue becomes faster and safer for the finance team.[1]

CelcomDigi did not publish a standard price, deployment time or minimum contract in the announcement. That makes it impossible to judge affordability from the release alone. Integration work, data preparation, security reviews and ongoing support may matter more than the model fee.[1]

An SME should therefore start with one workflow that already has a measurable baseline: time per case, error rate, backlog and cost. The buying decision becomes clearer when Sophia AI can be compared with the current process over a fixed pilot, including the cases that still require a person.[1]