Case Studies | SaaS B2B

Serverless AI agents, without a rewrite

Mastra to AgentCore Migration

About

A SaaS B2B business working with Cloud Combinator on AWS. The client is anonymised at their request.

Challenge

The migration centred on three areas: elastic and cost-efficient compute, a zero-disruption cut-over with decision parity, and preserving the existing agent codebase.

Elastic, pay-per-use compute

Always-on ECS Fargate tasks meant paying for capacity even when demand was low. The goal was AgentCore Runtime's scale-to-zero, scale-on-demand model so compute cost would track actual email volume.

Decision parity and zero disruption

The platform makes live security decisions, so the cut-over had to be reversible and provably equivalent, with a shadow-mode target of over 99% verdict match between ECS and AgentCore and a tested rollback path before any real traffic shifted.

Preserving the agent codebase

The client's 14 agents, 18 tools and 2 orchestration workflows in the Mastra framework represented significant investment. The migration had to keep them unchanged, replatforming only where the compute runs, not how the agents reason.

Solution

Cloud Combinator ran the migration across six phases, consolidated below, keeping the existing ECS deployment fully operational until an intentional, validated decommission at the end.

400,000

Emails per day analysed across 20 customers

>99%

Shadow-mode decision-match target (projected)

14

Agents and 18 tools preserved, no rewrite

By the numbers:

  • 400,000 - Emails per day analysed across 20 customers
  • >99% - Shadow-mode decision-match target (projected)
  • 14 - Agents and 18 tools preserved, no rewrite
Changes

Acceptance was structured around sign-off of the success criteria, phase-level acceptance at each of the six stages, and a production cut-over executed with minimal disruption, rollback capability and no data loss. The figures below reflect the SoW volume assumptions and projected cost model.

  • Serverless computeAgentCore Runtime's scale-to-zero, scale-on-demand model replaces always-on ECS Fargate, so compute cost tracks demand.
  • Zero rewriteThe Mastra framework, 14 agents, 18 tools and 2 orchestration workflows migrate unchanged.
  • Validated parityA week of shadow-mode comparison and a staged canary-to-full traffic shift, with a tested rollback to ECS, de-risk the cut-over.
  • Observability built inCloudWatch dashboards and alerts plus X-Ray tracing give the team visibility into the new runtime.
  • HandoverDeployment and rollback runbooks, an operations guide and a workshop covering AgentCore operations, scaling behaviour and incident response were delivered.

With the DLP email security service now running on AgentCore Runtime, the client's compute footprint scales elastically with demand rather than being fixed and always on. The target architecture is designed for full multi-region active/active operation, which can be enabled in a follow-on phase.

AWS Stack

Amazon Bedrock AgentCore Runtime

For serverless, auto-scaling agent compute that scales to zero when idle.

Amazon Bedrock

For GPT-OSS-120B, Nova 2 Lite and Titan Embed Text v2 model inference.

AWS Lambda

For the Entry Lambda that routes API Gateway requests to AgentCore.

Amazon Aurora PostgreSQL

With pgvector and RDS Proxy for semantic memory and pooled database connectivity.

Amazon SQS

For durable, timeout-free asynchronous email processing at production volume.

YOU MIGHT LIKE

Related success stories

View all case studies

Case Studies | Insights

Utilising Language Recognition, Speed, and Enhanced Security to Make Social Media a Force for Good

  • Here, we take a detailed look at how the Cloud Combinator team collaborated with another cutting-edge AI service provider that provides intelligent systems to “make social media more social” for brands and users alike.
  • Arwen AI is a UK-based startup specialising in AI solutions to manage and enhance brands’ social media interactions. Founded in 2020 by Matt McGrory, Dr. David Cole, and Joel Bailey, Arwen. AI focuses on using AI to automatically detect and remove spam, toxic comments, and other unwanted content from social media platforms.
  • The team at Arwen have three core products. ‘Moderate’ is focused on identifying and removing toxic content from social media channels. ‘Engage’ helps brands identify and engage with meaningful conversations on social media, and ‘Customize’ allows brands to apply bespoke algorithms to their channels - creating an even more effective moderation and engagement.
Read more
CONTACT US

Ready to turn AI into impact?

We'll help you spot the highest-value opportunities, reduce risk around your first AI initiative, and define a clear path to results from day one.

Why talk to us:

Outcome-driven recommendations

AWS-recognised delivery expertise

Risk-aware AI adoption

Clear next step, not a sales pitch

Start with a focused 20-minute conversation about your goals — no pressure, no commitment.

This website uses cookies to enhance user experience and to analyze performance and traffic on our website.

See our Privacy Policy for details.