click force is likely one of the leaders in digital promoting providers in Taiwan, specializing in data-driven promoting and conversion (D4A – Information for Promoting & Motion). CLICKFORCE’s mission is to supply industry-leading, trend-driven and modern advertising options that assist manufacturers, companies and media companions make smarter promoting choices.
Nonetheless, because the promoting {industry} quickly evolves, conventional analytical strategies and customary AI outputs are not ample to supply actionable insights. To remain aggressive, CLICKFORCE turned to AWS to construct Lumos, a next-generation AI-driven advertising analytics resolution powered by Amazon Bedrock, Amazon SageMaker AI, Amazon OpenSearch, and AWS Glue.
On this submit, we present how CLICKFORCE constructed Lumos utilizing AWS providers to rework promoting {industry} analytics from weeks of guide work to an automatic one-hour course of.
Digital promoting challenges
Earlier than adopting Amazon Bedrock, CLICKFORCE confronted a number of obstacles in constructing actionable intelligence for digital promoting. Giant-scale language fashions (LLMs) have a tendency to provide common suggestions quite than actionable industry-specific intelligence. With out understanding the promoting surroundings, these fashions lacked the {industry} context wanted to align their suggestions with actual {industry} realities.
One other main problem was the shortage of a unified inner dataset, which diminished the reliability of the output and elevated the chance of hallucinations and inaccurate insights. On the identical time, advertising groups did not have a standardized structure or workflow and relied on disconnected instruments and strategies like vibe coding, making processes tough to keep up and scale.
Making a complete {industry} evaluation report was additionally a prolonged course of, usually taking two to 6 weeks. The timeline resulted from a number of labor-intensive steps: 1-3 days to outline targets and arrange the analysis plan, 1-4 weeks to gather and validate information from numerous sources, 1-2 weeks to carry out statistical evaluation and create graphs, 1-2 weeks to extract strategic insights, and eventually 3-7 days to draft and finalize the report. Every stage typically required back-and-forth coordination between groups, which additional prolonged the schedule. Because of this, advertising methods had been continuously delayed and primarily based on instinct quite than insights backed by well timed information.
Resolution overview
CLICKFORCE was constructed to deal with these challenges. Lumosis an built-in AI-powered {industry} evaluation service powered by AWS providers.
The answer is designed round Amazon Bedrock Brokers for contextual inference and Amazon SageMaker AI to fine-tune the accuracy of Textual content-to-SQL. CLICKFORCE selected Amazon Bedrock as a result of it offers managed entry to the underlying fashions with out the necessity to construct or keep infrastructure, whereas additionally offering brokers that may coordinate multi-step duties and combine with enterprise information sources by a information base. This allowed the staff to ascertain insights primarily based on actual, verifiable information, decrease hallucinations, and quickly experiment with totally different fashions, whereas lowering operational overhead and dashing time to market.
Step one was to construct an built-in AI agent utilizing Amazon Bedrock. Finish customers work together with a chatbot interface operating on Amazon ECS. stream light And in entrance of that’s the Utility Load Balancer. When a consumer submits a question, it’s routed to an AWS Lambda perform that calls Amazon Bedrock Agent. The agent retrieves related info from an Amazon Bedrock information base constructed from supply paperwork similar to marketing campaign reviews, product descriptions, and {industry} evaluation information hosted in Amazon S3. These paperwork are robotically transformed to vector embeddings and listed by Amazon OpenSearch Service. By basing the mannequin’s response on this curated set of paperwork, CLICKFORCE ensured that the output was contextualized, much less hallucinatory, and according to real-world promoting information.
Subsequent, CLICKFORCE used Textual content-to-SQL requests to make the workflow extra action-oriented. When a question required information retrieval, Bedrock Agent generated a JSON schema by way of the Agent Motion API schema. These had been handed to a Lambda Executor perform that turned the requests into Textual content-to-SQL queries. AWS Glue crawlers repeatedly up to date the SQL database from CSV information in Amazon S3, permitting analysts to run exact queries about marketing campaign efficiency, viewers conduct, and aggressive benchmarks.
Lastly, the corporate is utilizing Amazon SageMaker and ML flow Incorporate it into your improvement workflow. Initially, CLICKFORCE relied on an underlying mannequin of Textual content to SQL conversion, which proved to be rigid and sometimes imprecise. Utilizing SageMaker, the staff processed information, evaluated totally different approaches, and fine-tuned the complete Textual content-to-SQL pipeline. After validation, the optimized pipeline was deployed by an AWS Lambda perform and reintegrated into the agent, permitting enhancements to be mirrored instantly within the Lumos utility. MLflow offers experiment monitoring and analysis, streamlining information processing, pipeline tuning, and deployment cycles, permitting Lumos to enhance question era accuracy and ship automated, data-driven advertising reporting.
outcome
The influence of implementing Amazon Bedrock Brokers and SageMaker AI has been transformative for CLICKFORCE. Business evaluation that beforehand took two to 6 weeks can now be accomplished in lower than an hour, dramatically accelerating decision-making. The corporate additionally diminished its reliance on third-party {industry} analysis reviews, leading to a 47% discount in operational prices.
Along with time and price financial savings, the Lumos system has expanded scalability throughout roles throughout the advertising surroundings. Model house owners, companies, analysts, entrepreneurs, and media companions can now generate insights independently with out ready for a centralized analyst staff. This autonomy elevated general marketing campaign agility. Moreover, by rooting the output in each inner datasets and industry-specific context, Lumos has considerably diminished the chance of illusions and ensured that insights extra intently align with {industry} actuality.
Customers can generate {industry} evaluation reviews by pure language conversations and iterate and enhance content material as they proceed to work together.
These visible reviews are generated by the Lumos system, powered by Amazon Bedrock Brokers and SageMaker AI, and reveal the platform’s capacity to generate complete market intelligence inside minutes. These graphs illustrate a model’s gross sales distribution, retail and e-commerce efficiency, and reveal how AI-driven analytics can automate information aggregation, visualization, and perception era with precision and effectivity.
conclusion
CLICKFORCE’s Lumos system represents a breakthrough in the best way digital advertising choices are made. By combining Amazon Bedrock Brokers, Amazon SageMaker AI, Amazon OpenSearch Service, and AWS Glue, CLICKFORCE reworked its {industry} analytics workflow from a time-consuming guide course of to a quick, automated, and dependable system. On this submit, we demonstrated how CLICKFORCE constructed Lumos utilizing these AWS providers to rework promoting {industry} analytics from weeks of guide work to an automatic one-hour course of.
In regards to the creator
Ray Wang I am a Senior Options Architect at AWS. With over 12 years of backend and consulting expertise, Ray is concentrated on constructing trendy options within the cloud, particularly NoSQL, huge information, machine studying, and generative AI. A voracious employee, he handed all 12 AWS certifications, rising the breadth and depth of his technical information. He loves studying and watching science fiction films in his free time.
shana chan I am an answer architect at AWS. She focuses on observability in trendy architectures and cloud-native monitoring options. Earlier than becoming a member of AWS, she was a software program engineer.
