Normalized claim
Time: 75% decrease
Case study generation time was reduced by about 75%, from weeks to minutes.
Skai (Kenshoo) builds Celeste AI agent on Amazon Bedrock Agents to cut report generation time and improve campaign insights
Use case typeCustomer personalizationUpdated Jun 13, 2026
Skai (formerly Kenshoo) is an AI-driven omnichannel advertising and analytics platform for brands and agencies. It unifies data from more than 100 publishers and retail networks to support media planning, optimization, and measurement. Skai built Celeste, a generative AI analytics agent that lets users ask natural-language questions across profiles, campaigns, ads, products, keywords, and search terms, then synthesizes the results into insights, recommendations, and ready-to-present materials.
- Organization
- Skai
- Industry
- Tech & Comms
- Location
- Israel
- Published
- September 2025
Reported outcomes
−75%
timeTime & speed
−50%time−30%quantified impact99.9%time
Strategic outcomes
New product / capabilityBuilt a natural-language analytics agentCustomer experience & trustImproved campaign insights and recommendationsSpeed & agilityAccelerated report generation workflowsSpeed & agilityFaster prototype-to-production delivery
Why do we believe this?Outcome claims, sources, and evidence checks
Normalized claim
Time: 50% decrease
Proof-of-concept to production timeline was reduced by 50%.
Normalized claim
Quantified impact: 30% decrease
Median latency was reduced by 30%.
Normalized claim
Time: 99.9%
The system maintained 99.9% uptime during critical customer demonstrations.
Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks
- Customer
- Skai
- Provider
- AWS
- Maturity
- Unknown
- Linked source
- AWS Machine Learning Blog
No explicit deployment-stage evidence found.
Customer identity supportedSource describes one deploymentMaturity evidence evaluated
Primary read
Use case focus
Showing 3 of 3
- 1Generative AI analytics assistant
- 2Agentic AI
- 3Customer insight generation
- Customers spent days or even weeks preparing static reports.
- Users struggled to query complex advertising datasets without technical expertise.
- Teams needed actionable recommendations and clearer narratives from large, siloed data holdings.
- Skai used Amazon Bedrock Agents to orchestrate tool/API calls against its own data layer.
- The agent joins data across profiles, campaigns, ads, products, keywords, and search terms and generates narrative insights and recommendations.
- The team added custom orchestration, streaming, dynamic session chunking, RAG-based memory retrieval, and monitoring with Amazon CloudWatch Logs Insights.
- Built-in action groups and memory/session management reduced custom integration code and enabled faster iteration from prototype to production.
- Case study generation time was reduced by about 75%, from weeks to minutes.
- Proof-of-concept to production timeline was reduced by 50%.
- Median latency was reduced by 30%.
- The system maintained 99.9% uptime during critical customer demonstrations.
Architecture
A custom customer-experience UI sends requests to a chat manager and chat executor layer, which invokes an Amazon Bedrock agent. The agent orchestrates tool calls through a custom API against Skai's data store. The architecture uses Amazon Bedrock, Amazon Bedrock Agents, Anthropic Claude 3.5 Sonnet V2, session management, streaming, dynamic session chunking, RAG-based memory retrieval, and Amazon CloudWatch Logs Insights for monitoring over 230 agent metrics. Data isolation is enforced so raw customer data is not shared with Bedrock.
Sources & evidence1
Evidence: Low35/100Evidence strength
- Customer explicitly identified
- Quantified outcome available
- Technical implementation details available
Type: Blog PostPublished: Sep 8, 2025Publisher: AWSEvidence: VendorConfidence: Medium
AI-generated summary. Verify important details with the linked sources before relying on this case.
Explore related AI use cases
Customer evidence
Provider evidence
Browse the catalog
Was this useful?
Community
Comments
Loading comments...