How Ring scales world buyer help with Amazon Bedrock Data Bases

This publish is cowritten with David Kim, and Premjit Singh from Ring.

Scaling self-service help globally presents challenges past translation. On this publish, we present you the way Ring, Amazon’s house safety subsidiary, constructed a production-ready, multi-locale Retrieval-Augmented Era (RAG)-based help chatbot utilizing Amazon Bedrock Data Bases. By eliminating per-Area infrastructure deployments, Ring diminished the price of scaling to every extra locale by 21%. On the similar time, Ring maintained constant buyer experiences throughout 10 worldwide Areas.

On this publish, you’ll find out how Ring carried out metadata-driven filtering for Area-specific content material, separated content material administration into ingestion, analysis and promotion workflows, and achieved value financial savings whereas scaling up. The structure described on this publish makes use of Amazon Bedrock Data Bases, Amazon Bedrock, AWS Lambda, AWS Step Features, and Amazon Easy Storage Service (Amazon S3). Whether or not you’re increasing help operations internationally or seeking to optimize your current RAG structure, this implementation offers sensible patterns you’ll be able to apply to your individual multi-locale help programs.

The help evolution journey for Ring

Buyer help at Ring initially relied on a rule-based chatbot constructed with Amazon Lex. Whereas practical, the system had limitations with predefined dialog patterns that couldn’t deal with the varied vary of buyer inquiries. Throughout peak durations, 16% of interactions escalated to human brokers, and help engineers spent 10% of their time sustaining the rule-based system. As Ring expanded throughout worldwide locales, this strategy turned unsustainable.

Necessities for a RAG-based help system

Ring confronted a problem: find out how to present correct, contextually related help throughout a number of worldwide locales with out creating separate infrastructure for every Area. The staff recognized 4 necessities that might inform their architectural strategy.

1. World content material localization

The worldwide presence of Ring required greater than translation. Every territory wanted Area-specific product info, from voltage specs to regulatory compliance particulars, offered via a unified system. Throughout the UK, Germany, and eight different locales, Ring wanted to deal with distinct product configurations and help situations for every Area.

2. Serverless, managed structure

Ring wished their engineering staff targeted on bettering buyer expertise, not managing infrastructure. The staff wanted a totally managed, serverless resolution.

3. Scalable information administration

With a whole lot of product guides, troubleshooting paperwork, and help articles continually being up to date, Ring wanted vector search know-how that would retrieve exact info from a unified repository. The system needed to help automated content material ingestion pipelines in order that the Ring content material staff may publish updates that might turn out to be obtainable throughout a number of locales with out handbook intervention.

4. Efficiency and value optimization

The common end-to-end latency requirement for Ring was 7–8 seconds and efficiency evaluation revealed that cross-Area latency accounted for lower than 10% of complete response time. This discovering allowed Ring to undertake a centralized structure reasonably than deploying separate infrastructure in every Area, which diminished operational complexity and prices.

To handle these necessities, Ring carried out metadata-driven filtering with content material locale tags. This strategy serves Area-specific content material from a single centralized system. For his or her serverless necessities, Ring selected Amazon Bedrock Data Bases and Lambda, which eliminated the necessity for infrastructure administration whereas offering automated scaling.

Overview of resolution

Ring designed their RAG-based chatbot structure to separate content material administration into two core processes: Ingestion & Analysis and Promotion. This two-phase strategy permits Ring to keep up steady content material enchancment whereas conserving manufacturing programs secure.

Ingestion and analysis workflow

Determine 1: Structure diagram exhibiting the Ring ingestion and analysis workflow with Step Features orchestrating every day information base creation, analysis, and high quality validation utilizing Data Bases and S3 storage.

1. Content material add – The Ring content material staff uploads help documentation, troubleshooting guides, and product info to Amazon S3. The staff structured the S3 objects with content material in encoded format and metadata attributes. For instance, a file for the content material “Steps to Change the doorbell battery” has the next construction:

{
	"properties": {
		"slug": "abcde",
		"contentLocale": "en-GB",  # distinctive identifier
		"sourceFormat": "md",      # locale info
		"metadataAttributes": {
			"group": "Service",
			"slug": "abcde",
			"contentLocale": "en-GB"
		},
		"content material": "U3RlcHMgdG8gUmVwbGFjZSB0aGUgZG9vcmJlbGwgYmF0dGVyeTo= 
                VXNlIHRoZSBpbmNsdWRlZCBzZWN1cml0eSBzY3Jld2RyaXZlciB0byByZW1vdmUgd
                GhlIHNlY3VyaXR5IHNjcmV3IGxvY2F0ZWQgb24gdGhlIGJvdHRvbSBvZiB0aGUgZm
                FjZXBsYXRlCgpSZW1vdmUgdGhlIGZhY2VwbGF0ZSBieSBwcmVzc2luZyBpbiBvbiB
                0aGUgc2lkZXMgYW5kIGNhcmVmdWxseSBwdWxsaW5nIGl0IG91dCBhbmQgb2ZmCgpS
                ZW1vdmUgdGhlIGJhdHRlcnkgZnJvbSB0aGUgZG9vcmJlbGwKCkNvbm5lY3QgdGhlI
                GNoYXJnaW5nIGNhYmxlIHRvIHRoZSBiYXR0ZXJ5J3MgY2hhcmdpbmcgcG9ydAoKQ2h
                hcmdlIHVudGlsIG9ubHkgdGhlIGdyZWVuIGxpZ2h0IHJlbWFpbnMgbGl0ICh3aGlsZ
                SBjaGFyZ2luZywgeW91J2xsIHNlZSBib3RoIGEgc29saWQgZ3JlZW4gYW5kIGFtYmV
                yIGxpZ2h0KQoKUmUtaW5zZXJ0IHRoZSBjaGFyZ2VkIGJhdHRlcnkgaW50byB0aGUgZ
                G9vcmJlbGwKCkRlLWF0dGFjaCB0aGUgZmFjZXBsYXRlCgpTZWN1cmUgd2l0aCB0aGU
                gc2VjdXJpdHkgc2NyZXc=    # base64 encoded 
	}
}

2. Content material processing – Ring configured Amazon S3 bucket occasion notifications with Lambda because the goal to routinely course of uploaded content material.

3. Uncooked and processed content material storage
   
   The Lambda operate performs two key operations:
   • Copies the uncooked information to the Data Base Archive Bucket
   • Extracts metadata and content material from uncooked information, storing them as separate recordsdata within the Data Base Supply Bucket with contentLocale classification (for instance, {locale}/Service.Ring.{Upsert/Delete}.{unique_identifier}.json)

   For the doorbell battery instance, the Ring metadata and content material recordsdata have the next construction:

   {locale}/Service.Ring.{Upsert/Delete}.{unique_identifier}.metadata.json

{
	"metadataAttributes" : {
		"group": "Service",
		"slug": "abcde",
		"contentLocale": "en-GB"
	}
}

{locale}/Service.Ring.{Upsert/Delete}.{unique_identifier}.json

{
	"content material": "Steps to Change the doorbell battery:
	Use the included safety screwdriver to take away the safety screw situated on the underside of the faceplate
	Take away the faceplate by urgent in on the perimeters and punctiliously pulling it out and off
	Take away the battery from the doorbell
	Join the charging cable to the battery's charging port
	Cost till solely the inexperienced gentle stays lit (whereas charging, you will see each a strong inexperienced and amber gentle)
	Re-insert the charged battery into the doorbell
	Re-attach the faceplate
	Safe with the safety screw
}

4. Day by day Knowledge Copy and Data Base Creation

Ring makes use of AWS Step Features to orchestrate their every day workflow that:

• Copies content material and metadata from the Data Base Supply Bucket to Knowledge Supply (Model)
• Creates a brand new Data Base (Model) by indexing the every day bucket as information supply for vector embedding

Every model maintains a separate Data Base, giving Ring impartial analysis capabilities and simple rollback choices.

5. Day by day Analysis Course of

The AWS Step Features workflow continues utilizing analysis datasets to:

• Run queries throughout Data Base variations
• Take a look at retrieval accuracy and response high quality to match efficiency between variations
• Publish efficiency metrics to Tableau dashboards with outcomes organized by contentLocale

6. High quality Validation and Golden Dataset Creation

Ring makes use of the Anthropic Claude Sonnet 4 giant language mannequin (LLM)-as-a-judge to:

• Consider metrics throughout Data Base variations to establish the best-performing model
• Examine retrieval accuracy, response high quality, and efficiency metrics organized by contentLocale
• Promote the highest-performing model to Knowledge Supply (Golden) for manufacturing use

This structure helps rollbacks to earlier variations for as much as 30 days. As a result of content material is up to date roughly 200 instances per week, Ring determined to not preserve variations past 30 days.

Promotion workflow: customer-facing

Determine 2: Structure diagram exhibiting the Ring manufacturing chatbot system the place buyer queries circulate via AWS Lambda to retrieve context from Data Bases and generate responses utilizing basis fashions

1. Buyer interplay – Prospects provoke help queries via the chatbot interface. For instance, a buyer question for the battery substitute state of affairs appears to be like like this:

{
	"textual content": "How can I change the doorbell battery?",
	"market": "en-GB"
}

2. Question orchestration and information retrieval

Ring configured Lambda to course of buyer queries and retrieve related content material from Amazon Bedrock Data Bases. The operate:

• Transforms incoming queries for the RAG system
• Applies metadata filtering with contentLocale tags utilizing equals operator for exact Regional content material concentrating on
• Queries the validated Golden Knowledge Supply to retrieve contextually related content material

Right here’s the pattern code Ring makes use of in AWS Lambda:

## Metadata Filtering for Regional Content material Concentrating on

num_results = 10
market = "en-GB"
knowledge_base_id = "A2BCDEFGHI"
user_text = "How can I change the doorbell battery?"

# Configure Regional content material filtering
vector_search_config = {"numberOfResults": num_results}
vector_search_config["filter"] = {
	"equals": {
		"key": "contentLocale",
		"worth": market
	}
}

# Run Amazon Bedrock Data Base search
response = boto3.shopper("bedrock-agent-runtime").retrieve(
	knowledgeBaseId=knowledge_base_id,
	retrievalQuery={"textual content": user_text},
	retrievalConfiguration={
		"vectorSearchConfiguration": vector_search_config,
	},
)

3. Response era

Within the Lambda operate, the system:

• Types the retrieved content material primarily based on relevance rating and selects the highest-scoring context
• Combines the top-ranked context with the unique buyer question to create an augmented immediate
• Sends the augmented immediate to LLM on Amazon Bedrock
• Configures locale-specific prompts for every contentLocale
• Generates contextually related responses returned via the chatbot interface

Different issues to your implementation

When constructing your individual RAG-based system at scale, take into account these architectural approaches and operational necessities past the core implementation:

Vector retailer choice

The Ring implementation makes use of Amazon OpenSearch Serverless because the vector retailer for his or her information bases. Nevertheless, Amazon Bedrock Data Bases additionally helps Amazon S3 Vectors as a vector retailer possibility. When selecting between these choices, take into account:

• Amazon OpenSearch Serverless: Supplies superior search capabilities, real-time indexing, and versatile querying choices. Finest suited to functions requiring complicated search patterns or whenever you want extra OpenSearch options past vector search.
• Amazon S3 vectors: Gives a cheaper possibility for simple vector search use circumstances. S3 vector shops present automated scaling, built-in sturdiness, and will be extra economical for large-scale deployments with predictable entry patterns.

Along with these two choices, AWS helps integrations with different information retailer choices, together with Amazon Kendra, Amazon Neptune Analytics, and Amazon Aurora PostgreSQL. Consider your particular necessities round question complexity, value optimization, and operational wants when deciding on your vector retailer. The prescriptive steerage offers a superb place to begin to judge vector shops to your RAG use case.

Versioning structure issues

Whereas Ring carried out separate Data Bases for every model, you may take into account another strategy involving separate information sources for every model inside a single information base. This technique leverages the x-amz-bedrock-kb-data-source-id filter parameter to focus on particular information sources throughout retrieval:

vector_search_config["filter"] = {
	"equals": {
		"key": "x-amz-bedrock-kb-data-source-id",
		"worth": '<information supply Id>'
		}
	}
# Execute Bedrock Data Base search
response = boto3.shopper("bedrock-agent-runtime").retrieve(
	knowledgeBaseId=knowledge_base_id,
	retrievalQuery={"textual content": user_text},
	retrievalConfiguration={
		"vectorSearchConfiguration": vector_search_config,
	},
)

When selecting between these approaches, weigh these particular trade-offs:

• Separate information bases per model (the strategy that Ring makes use of): Supplies information supply administration and cleaner rollback capabilities, however requires managing extra information base situations.
• Single information base with a number of information sources: Reduces the variety of information base situations to keep up, however introduces complexity in information supply routing logic and filtering mechanisms, plus requires sustaining separate information shops for every information supply ID.

Catastrophe restoration: Multi-Area deployment

Take into account your catastrophe restoration necessities when designing your RAG structure. Amazon Bedrock Data Bases are Regional sources. To attain strong catastrophe restoration, deploy your full structure throughout a number of Areas:

• Data bases: Create Data Base situations in a number of Areas
• Amazon S3 buckets: Preserve cross-Area copies of your Golden Knowledge Supply
• Lambda capabilities and Step Features workflows: Deploy your orchestration logic in every Area
• Knowledge synchronization: Implement processes to maintain content material synchronized throughout Areas

The centralized structure serves its site visitors from a single Area, prioritizing value optimization over multi-region deployment. Consider your individual Restoration Time Goal (RTO) and Restoration Level Goal (RPO) necessities to find out whether or not a multi-Area deployment is critical to your use case.

Basis mannequin throughput: Cross-Area inference

Amazon Bedrock basis fashions are Regional sources with Regional quotas. To deal with site visitors bursts and scale past single-Area quotas, Amazon Bedrock helps cross-Area inference (CRIS). CRIS routinely routes inference requests throughout a number of AWS Areas to extend throughput:

CRIS: Routes requests solely inside particular geographic boundaries (equivalent to inside the US or inside the EU) to fulfill information residency necessities. This will present as much as double the default in-Area quotas.

World CRIS: Routes requests throughout a number of industrial Areas worldwide, optimizing obtainable sources and offering larger mannequin throughput past geographic CRIS capabilities. World CRIS routinely selects the optimum Area to course of every request.

CRIS operates independently out of your Data Base deployment technique. Even with a single-Area Data Base deployment, you’ll be able to configure CRIS to scale your basis mannequin throughput throughout site visitors bursts. Observe that CRIS applies solely to the inference layer—your Data Bases, S3 buckets, and orchestration logic stay Regional sources that require separate multi-Area deployment for catastrophe restoration.

Embedding mannequin choice and chunking technique

Deciding on the suitable embedding mannequin and chunking technique is necessary for RAG system efficiency as a result of it instantly impacts retrieval accuracy and response high quality. Ring makes use of the Amazon Titan Embeddings mannequin with the default chunking technique, which proved efficient for his or her help documentation.

Amazon Bedrock provides flexibility with a number of choices:

Embedding fashions:

• Amazon Titan embeddings: Optimized for text-based content material
• Amazon Nova multimodal embeddings: Helps “Textual content”, “Picture”, “Audio”, and “Video” modalities

Chunking methods:

When ingesting information, Amazon Bedrock splits paperwork into manageable chunks for environment friendly retrieval utilizing 4 methods:

• Customary chunking: Mounted-size chunks for uniform paperwork
• Hierarchical chunking: For structured paperwork with clear part hierarchies
• Semantic chunking: Splits content material primarily based on subject boundaries
• Multimodal content material chunking: For paperwork with combined content material sorts (textual content, photos, tables)

Consider your content material traits to pick out the optimum mixture to your particular use case.

Conclusion

On this publish, we confirmed how Ring constructed a production-ready, multi-locale RAG-based help chatbot utilizing Amazon Bedrock Data Bases. The structure combines automated content material ingestion, systematic every day analysis utilizing an LLM-as-judge strategy, and metadata-driven content material concentrating on to attain a 21% discount in infrastructure and operational value per extra locale, whereas sustaining constant buyer experiences throughout 10 worldwide Areas.

Past the core RAG structure, we lined key design issues for manufacturing deployments: vector retailer choice, versioning methods, multi-Area deployment for catastrophe restoration, Cross-Area Inference for scaling basis mannequin throughput, embedding mannequin choice and chunking methods. These patterns apply broadly to any staff constructing multi-locale or high-availability RAG programs on AWS.Ring continues to evolve their chatbot structure towards an agentic mannequin with dynamic agent choice and integration of a number of specialised brokers. This agentic strategy will permit Ring to route buyer inquiries to specialised brokers for system troubleshooting, order administration, and product suggestions, demonstrating the extensibility of RAG-based help programs constructed on Amazon Bedrock.

To study extra about Amazon Bedrock Data Bases, go to the Amazon Bedrock documentation.

Concerning the authors