Data residency and privacy
Production customer data is planned to remain in AWS EU Regions, with region selection aligned to customer and legal requirements.
Cortera on AWS
A secure, EU-hosted foundation for trusted AI readiness.
Cortera is building an AWS-native platform that combines conversational AI, deterministic assessment logic, secure customer data handling, and automated report delivery for growing European businesses.
Reference architecture
Interactive layer view of the planned platform. Service names are labels only; no official AWS marks are used.
Edge and delivery
Users
Customers and Cortera reviewers
Amazon CloudFront
Global content delivery
AWS WAF
Request filtering and rate limiting
AWS Amplify Hosting
Next.js portal and landing page
Application
Amazon API Gateway
Authenticated HTTP APIs
AWS Lambda
Assessment and reporting services
Amazon Cognito
Authentication and role-based access
Amazon EventBridge
Domain events and scheduling
Data and AI
Amazon RDS for PostgreSQL
Organizations, answers, scores, audit metadata
Amazon S3 + AWS KMS
Reports, uploads, export packages
Amazon Bedrock
Conversation, extraction, summarisation
AWS Secrets Manager
Secrets and API credentials
Workflow and operations
AWS Step Functions
Report generation orchestration
Amazon SQS
Retries, reminders, webhooks
Amazon SES
Transactional email
CloudWatch + CloudTrail
Monitoring and audit logging
AWS Backup
RDS and protected data recovery
Architecture principles
Secure by design
Encryption in transit and at rest, least-privilege IAM, organization-level tenant isolation, secret management, audit trails, and controlled retention.
AI with guardrails
Amazon Bedrock is used for conversational guidance and structured summarization. Deterministic rules remain the source of truth for scoring, risk flags, workflow state, and report generation decisions.
Human accountability
Higher-risk flags and early reports can be routed to a human quality review queue before delivery.
Cost-efficient scale
Serverless and managed AWS services minimize operational overhead and let Cortera scale with actual customer usage.
AI designed as an assistant, not an uncontrolled decision-maker.
Cortera will use Amazon Bedrock to access foundation models through AWS-native identity, logging, billing, and regional controls.
Use cases
- Conduct a guided, empathetic, one-question-at-a-time discovery conversation.
- Extract structured facts from free-text customer answers.
- Ask focused clarification questions where information is incomplete.
- Generate an executive-friendly narrative from verified assessment data.
- Draft tailored 30-day plans based on deterministic rules and approved recommendation libraries.
Guardrails
- Models do not independently change billing, customer permissions, or report publication status.
- Scores and risk classifications are calculated by versioned, deterministic rules.
- Inputs and outputs are schema-validated where possible.
- Sensitive data minimization and content filtering are planned before model invocation.
- The platform will maintain traceability through prompt, model, and methodology version metadata.
AWS services and planned roles
| AWS service | Planned role in Cortera | Why it is used |
|---|---|---|
| AWS Amplify Hosting | Hosts the Next.js landing page and customer portal. | Managed CI/CD and hosting for a small team with no server maintenance. |
| Amazon CloudFront | Delivers portal assets and API responses at the edge. | Low latency for EU customers and a single TLS entry point. |
| AWS WAF | Filters malicious traffic and applies rate limits. | Baseline protection for a public B2B application. |
| Amazon Cognito | Authenticates users and enforces roles per organization. | Managed identity avoids building custom auth for a security-sensitive product. |
| Amazon API Gateway | Exposes authenticated HTTP APIs for the portal and integrations. | Throttling, authorization, and request validation out of the box. |
| AWS Lambda | Runs assessment logic, scoring, and report services. | Event-driven compute that matches spiky, per-assessment usage. |
| Amazon RDS for PostgreSQL | Stores organizations, users, answers, scores, workflow state, and audit metadata. | Relational integrity and versioned scoring need a proper transactional database. |
| Amazon S3 | Stores generated reports, controlled uploads, and export packages. | Durable object storage with fine-grained access policies and lifecycle rules. |
| AWS KMS | Manages encryption keys for S3 and RDS. | Customer-managed keys and key usage auditing. |
| Amazon Bedrock | Provides model access for conversation, extraction, and narrative drafting. | AWS-native identity, logging, billing, and regional controls for model usage. |
| AWS Step Functions | Orchestrates multi-step report generation and review workflows. | Explicit state, retries, and visibility for long-running processes. |
| Amazon SQS | Buffers background jobs such as reminders and webhook handling. | Decoupling and reliable retries under load. |
| Amazon SES | Sends transactional email: invitations, progress reminders, report delivery. | Reliable EU-region email with deliverability controls. |
| AWS Secrets Manager | Stores API credentials and application secrets. | Rotation and least-privilege access instead of environment sprawl. |
| Amazon CloudWatch | Collects metrics, logs, alarms, and dashboards. | Operational visibility for a small team running a production service. |
| AWS CloudTrail | Records API and administrative activity. | Security traceability and audit evidence. |
| AWS Backup | Automates backup and recovery for RDS and protected data. | Recovery objectives without custom backup tooling. |
| Amazon EventBridge | Routes domain events between services and schedules recurring jobs. | Loose coupling between assessment, reporting, and notification flows. |
| Amazon GuardDuty (future security roadmap) | Continuous threat detection across accounts and workloads. | Planned as the platform handles more customer environments. |
Phased roadmap
Phase 1
Pilot and MVP
- Landing page and secure customer portal
- Guided assessment and rule-based scoring
- Report generation on the AWS-hosted core platform
Phase 2
Paid self-service readiness platform
- Payments and automated onboarding
- Founder review queue
- Deeper analytics and improved report personalisation
Phase 3
Governance and regulated-industry capabilities
- Multi-user workspaces, AI inventory, evidence review
- AI policy support and integrations
- Recurring reassessments and regulated-industry modules
Why AWS Activate support matters to Cortera
AWS Activate credits would accelerate the transition from a validated pilot to a secure, production-ready platform. Credits will be used primarily for Amazon Bedrock model inference, secure data storage and processing, customer portal hosting, managed PostgreSQL, workflow orchestration, monitoring, security controls, and testing in EU AWS Regions.
- 35%Amazon Bedrock experimentation and inference
- 20%Application compute and workflow orchestration
- 15%Secure data storage and database
- 15%Security, monitoring, logging, and backups
- 15%Testing, staging, and product development environments
This allocation is illustrative and will evolve based on validated customer demand and usage.