The Hapee team at AWS Chile

The cloud Hapee runs on is not an accessory: it's the difference between an AI call that responds in 480ms and one that responds in 3 seconds. That's why this week we traveled to the AWS offices in Santiago to sit down with the regional team and formalize Hapee's entry into the AWS Partner Network.

Why AWS, and not another cloud

Hapee has run on AWS since day 1, but this changes the level of integration. Key decisions:

• Bedrock as a model hub · access to Claude 4 (Anthropic), Llama 4, Titan, Mistral, all through the same API. Hapee no longer manages N credentials
• Lambda + Step Functions for workflows · cold start under 200ms in LATAM regions
• Multi-region EKS · automatic failover between SAO and QRO if a region has incidents
• DynamoDB Global Tables · cross-region replication for conversations, with no perceptible lag
• S3 + CloudFront · assets served from the edge closest to the end customer

Who was there

A 50-person meeting in a technical roundtable format: 3 from the Hapee core team, 12 from the AWS Chile regional team, and ~35 founders and CTOs from the partner ecosystem. A 2-hour roundtable on how to scale AI startups in LATAM without falling into the trap of runaway cloud costs.

What's coming operationally

1. June 2026 · Migration to Bedrock Anthropic, Llama and Mistral models will be invoked via Bedrock. Nothing changes for the customer — the names stay the same in the Hapee UI. But billing is unified and throughput goes up.
2. July 2026 · Mexico region (QRO) Hapee gets native deployment in QRO in addition to SAO. Customers in Mexico, Central America and the Caribbean gain ~60ms in latency.
3. August 2026 · AWS Marketplace Hapee will be listed on the AWS LATAM marketplace. Companies with an AWS contract can pay for Hapee with their existing EDP.
4. Q4 2026 · Active co-selling The AWS team starts recommending Hapee to enterprise customers who need a conversational AI stack with compliance + scale.

What this means for Hapee customers

• Lower latency · all LATAM regions with <100ms
• Enterprise compliance · SOC 2 Type II + ISO 27001 (inherited from AWS) ready for enterprise deals
• Predictable pricing · committed use discounts passed on to Tier Pro+ customers
• No downtime · multi-AZ + multi-region means a real 99.95% SLA

Thanks to the team

Thanks to the AWS Chile team for their openness, especially the Partner Development team that accompanied us through the process over the last 3 months. And to the 35 founders who joined the roundtable — the cloud + AI ecosystem in Chile is much more alive than it looks from the outside.