Key Outcomes
A working proof of concept for an automated, multi-agent campaign-creation workbench, running on Amazon Bedrock, designed and built in a three-day sprint.
A validated three-year business case, projecting roughly 429% ROI and a four-month payback.
A prioritized portfolio of agentic use cases and a reference architecture for an “agentic customer-intelligence platform.”
A clear path from concept to production, following Thoughtworks' 3/3/3 method.
A new user experience that replaces manual, code-like tasks with a conversational flow, backed by built-in guardrails and human oversight.
Splio is a European customer-marketing company with deep roots in European retail and telco, now evolving from CRM and marketing automation toward a customer intelligence platform. Its differentiation is predictive: a deep-learning approach to 'Individuation' marketing that forecasts individual customer intent, positioning Splio as a sovereign European alternative to the large US marketing cloud platforms. The company has set an ambitious goal: to generate 50% of annual recurring revenue (ARR) from AI offerings by 2027, and to prove quickly that agentic AI could get it there.
Splio's predictive intelligence ran ahead of its execution. The insight was rich, but acting on it stayed manual. Marketers spent their days on the toil of campaign orchestration: hand-building audiences, copy-pasting translations for each region and working around the platform's template syntax, which kept campaign creation out of reach for non-technical marketing managers. For customer teams that were shrinking as workloads grew, this ‘action gap’ limited how much value the platform's intelligence could deliver.
Splio wanted to close that gap on a single, cloud-native foundation, and to move past AI hype to a validated, costed case before committing to a build.
3/3/3 metholodolgy in action
From concept to a costed business case in days, through the AWS Agentic Catalyst Program.
How the work came together
Splio, AWS and Thoughtworks ran the engagement through the AWS Agentic Catalyst Program (ACP), a jointly funded format for proving both the technical feasibility and the business value of agentic AI quickly. Thoughtworks brought its 3/3/3 method and forward-deployed engineering, with product strategy, AI solution architecture and machine-learning engineering working alongside Splio's team, while AWS provided the program, solution architecture and cloud platform. From the outset the goal was not only to show what agentic AI could do, but to tie it to commercial outcomes Splio's leadership could measure and act on.
The team worked as one unit. A short, focused workshop mapped Splio's marketing value chain end to end, surfaced the moments of highest friction for the marketers who live in the product and scored candidate use cases on impact against complexity. Automated campaign creation rose to the top: the clearest path to removing manual toil while proving tangible business value, and to reinforcing Splio's position as an AI-first platform that turns customer intelligence into action, not just insight.
The AWS Agentic Catalyst Program helps software companies identify the right agentic AI bet and prove it fast. Splio came ready to move quickly and left with a working prototype and business case they could act on immediately. Thoughtworks' engineering rigor turned a sprint deliverable into a production-ready foundation for Splio.
A conversational, multi-agent campaign workbench
Rather than add a single assistant to the product, the team designed a workbench of specialized agents that turns a marketer's plain-language intent into a ready-to-run, cross-channel campaign. A marketer describes what they want; the system interprets the intent, recommends the highest-yielding audience segments and generates brand-aware copy and localized variants, with no template code and no copy-pasting.
Because the agents learn each brand's guidelines from previous campaigns, the output stays on-brand, and compliance and brand-safety checks are applied as content is generated, with a person approving each step.
How Splio's workbench runs on AWS
The proof of concept runs on an AWS-native, serverless foundation:
Amazon Bedrock for managed foundation-model inference, running Anthropic's Claude Sonnet 4.5, so the team could build on frontier models without operating any model infrastructure
A multi-agent design: an intent classifier, a context agent and a content agent, each with its own governed instructions, coordinated by the Strands agent framework
Amazon S3, including S3 Vector storage, for content and the embeddings that ground the agents in Splio's brand and campaign context
Amazon API Gateway to expose the workbench's services cleanly for integration into Splio's platform
The outcomes that matter
In the time many organizations spend scoping a project, Splio came away with three things: a working proof of concept that the workflow is feasible on AWS, a costed business case its leadership can act on and a sequenced roadmap to production. The projected case is strong, a roughly 429% three-year return and a four-month payback, driven by new revenue as customers adopt Splio's new AI offering, lower churn and a marked lift in marketer productivity, with a modest projected cloud run-cost.
The deeper outcome is strategic. Splio now has a validated, reusable pattern for agentic customer-intelligence platform and a de-risked path toward its 2027 AI-revenue goal, proven collaboratively rather than assumed.
A validated 3 - year business case
Projected 429% ROI
4 -month payback
Our ambition is to become the customer intelligence platform for European retail and telco, which means turning our predictive insight into action for every marketer who uses us. Working with Thoughtworks and AWS through the Agentic Catalyst Program, we went from an idea to a working agentic prototype and a costed business case in a matter of days. That gave us the evidence, and the confidence, to make agentic AI core to how our platform works.
What's next
With the concept validated, the path forward follows the same 3/3/3 rhythm: a short prototype tested with beta clients, then a focused build toward a production minimum “lovable” product, consolidating Splio's technology onto a single cloud-native foundation and extending the agent pattern to adjacent use cases such as performance tracking and ROI reporting, on the way to Splio's wider agentic-enabled customer intelligence vision.
What began as a three-day sprint has given Splio a foundation to build on, and a template for how a focused, jointly funded engagement can take agentic AI from ambition to evidence, fast.
