Generative AI on AWS
Generative AI applications on AWS
Assistants, copilots and analytics built on Amazon Bedrock, connected securely to your data and deployed in your AWS environment.
What we build
- RAG (retrieval-augmented generation)
- Answers grounded in your own documents and data, with sources shown.
- Enterprise knowledge assistants
- One place to ask questions across policies, manuals, wikis and files.
- AI chatbots
- Customer and employee chat that hands over to people when needed.
- AI copilots
- Assistants built into your product or internal tools.
- Document intelligence
- Extract, classify and summarise contracts, invoices, forms and reports.
- Conversational search
- Search that understands what users mean, not just keywords.
- Natural language analytics
- Ask business questions in plain language and get charts and explanations.
- Natural language to SQL
- Turn questions into validated SQL queries on your databases.
- AI recommendations
- Product, content or next-step suggestions from your data.
- AI content generation
- Drafts, summaries and descriptions in your format and tone.
How a Generative AI application fits together
Data stays in your AWS account. Access is controlled with IAM, data is encrypted with AWS KMS, and Guardrails filter harmful or off-topic responses.
- 1Users
- 2Amazon CloudFront
- 3App on Amazon ECS Fargate or AWS Lambda
- 4Amazon Bedrock: Amazon Nova and other models, with Guardrails
- 5Amazon Bedrock Knowledge Bases
- 6Amazon S3, Aurora PostgreSQL with pgvector, Amazon DynamoDB
Technology
- Amazon Bedrock
- Amazon Nova
- Bedrock Knowledge Bases
- Bedrock Guardrails
- Amazon S3
- Aurora PostgreSQL
- pgvector
- Amazon DynamoDB
- AWS Lambda
- Amazon ECS Fargate
- Amazon CloudFront
When an assistant should become an agent
A knowledge assistant answers. An agent can also act: raise a ticket, update a record, run an approved query or start a workflow. Perixope builds both, and helps you decide which one your use case needs.