Smart has published fuller pricing and service details for AI.Smart, the artificial intelligence workspace that first appeared as a free perk when the telco launched LiveSmart on September 18. The browser-based service gives Smart and TNT subscribers access to several AI model families through credit packs starting at PHP 49.
AI.Smart was already live through its website and the LiveSmart app on September 18, according to Amihan Global Strategies. Smart, Avtica, and their partners later presented the service at a September 22 media briefing in Makati. As CTRL+PH previously reported, AI.Smart was among the launch offers included with LiveSmart.
AI.Smart’s own pricing page lists four packs:
| Pack | Price | Validity | Credits | Team seats |
|---|---|---|---|---|
| Solo | PHP 49 | 7 days | 100,000 | 1 |
| Flex | PHP 99 | 14 days | 225,000 | 1 |
| Pro | PHP 149 | 30 days | 350,000 | 5 |
| Max | PHP 499 | 30 days | 1.2 million | 10 |
The site describes Solo as intended for everyday personal use, Flex for heavier personal use, Pro for small teams, and Max for larger teams. It also says credits left when a pack expires are forfeited and cannot be refunded. The pricing page advertises an extended free trial for the first 100,000 sign-ups.
AI.Smart runs on Amazon Web Services, with inference served through Amazon Bedrock. According to Amihan, subscribers can access five model families from the same workspace: Amazon Nova, Anthropic Claude, Google Gemini, OpenAI, and Thinking Machines.
AI.Smart automatically selects a model based on the task rather than asking the user to move between separate AI services. Subscribers can generate text and images, work with uploaded files, create documents and presentations, build spreadsheets, and analyze data. The workspace supports English, Filipino, and Taglish.
Smart and TNT subscribers register using their mobile number and an SMS verification. Packs can be charged directly to the subscriber’s mobile account after confirming a one-time PIN.
Smart and Amihan say subscriber uploads are not used to train the underlying models, adding that data remains encrypted at rest and in transit and is not retained by model providers.