Back office ops · Production

Cato Networks uses Amazon Bedrock to transform free text search into structured GraphQL queries

The problem

Filtering events on Cato's SASE management console required manually adding filters, which was time-consuming and demanded in-depth familiarity with the product glossary—creating a steep barrier for new users.

Workflow diagram · grounded in source
1
User enters free text query
trigger
“users can also use free text query mode”
2
NLS service builds FM prompt
integration
“Natural language search (NLS) service – An Amazon Elastic Kubernetes Service (Amazon EKS) hosted service to bridge between Cato's management console and Amazon Bedrock. This service is responsible for creating the complete prompt for the FM”
3
FM converts text to structured JSON
ai_action
“we selected anthropic.claude-3-5-sonnet-20241022-v2:0, which met the error rate criterion and achieved the highest success rate while maintaining reasonable costs and latency”
4
JSON schema validation
validation
“Validate the JSON schema on the response. This step is crucial, because model behavior is inherently non-deterministic, and responses that don't comply with our API will break the product functionality”
5
Translate to GraphQL API request
output
“After the FM successfully translates the free text into structured output, converting it into an API request—such as GraphQL—is a straightforward and deterministic process”
Reported outcome

The free text search feature received positive customer feedback, with query time reduced from minutes of manual filtering to near-instant results; account admins reported near-zero time to value with a minimal learning curve, and non-native English speakers benefited from native multi-language support.

Reported metrics
Query timesignificant reduction in query time—cut down from minutes of manual filtering to near-instant results
Time to valuenear-zero time to value
Translation error rate threshold (release criterion)below 0.05
Reported stack
Amazon Bedrockanthropic.claude-3-5-sonnet-20241022-v2:0GraphQL
Source
https://aws.amazon.com/blogs/machine-learning/how-cato-networks-uses-amazon-bedrock-to-transform-free-text-search-into-structured-graphql-queries?tag=soumet-20
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

The free text search feature received positive customer feedback, with query time reduced from minutes of manual filtering to near-instant results; account admins reported near-zero time to value with a minimal learni…

What tools did this team use?

Amazon Bedrock, anthropic.claude-3-5-sonnet-20241022-v2:0, GraphQL.

What results were reported?

Query time: significant reduction in query time—cut down from minutes of manual filtering to near-instant results; Time to value: near-zero time to value; Translation error rate threshold (release criterion): below 0.05 (source-reported, not independently verified).

How is this back office ops AI workflow structured?

User enters free text query → NLS service builds FM prompt → FM converts text to structured JSON → JSON schema validation → Translate to GraphQL API request.

Related back office ops cases