AWS Managed AI Services — Pre-built AI APIs for developers. No ML expertise required.
Overview
AWS provides 13 core pre-built AI services as APIs for common AI/ML tasks. These services are fully managed — no model training, infrastructure management, or ML expertise required. Just call the API and get results.
Key Insight: Managed AI Services are for developers who want AI capabilities without building models. Think of them as “AI as a Service” — input data, get insights.
Services Overview
| Service | Category | Input | Output | Free Tier |
|---|---|---|---|---|
| Rekognition | Computer Vision | Images/video | Labels, faces, text, moderation | 🟢 5K transactions/mo |
| Textract | Computer Vision | Documents | Structured data (tables, forms) | 🟢 1K pages/mo |
| Transcribe | Language | Audio | Text | 🟢 60 min/mo |
| Polly | Language | Text | Audio | 🟢 5M chars/mo |
| Translate | Language | Text | Translated text | 🟢 2M chars/mo |
| Comprehend | NLP | Text | Sentiment, entities, topics | 🟢 5M units/mo |
| Lex | Conversational | Text/voice | Conversational responses | 🟢 Free tier available |
| Kendra | Search | Documents | Ranked search results | 🟢 750 hrs (30 days) |
| Augmented AI | Human Review | ML predictions | Validated predictions | 🟢 500 reviews/yr |
| Mechanical Turk | Crowdsourcing | Tasks | Human intelligence | 🔴 No free tier |
| Comprehend Medical | Healthcare NLP | Medical text | Medical entities, PHI | 🟢 8.5M chars/mo |
| Transcribe Medical | Healthcare Speech | Medical audio | Medical transcripts | 🔴 No free tier |
| HealthScribe | Clinical AI | Conversations | Clinical notes | 🟢 300 min/mo (2 mo) |
Why Use Managed AI Services?
| Benefit | Description |
|---|---|
| No ML expertise | Pre-trained models, just call APIs |
| Fast integration | Add AI features in minutes, not months |
| Scalable | Handle millions of requests automatically |
| Pay-per-use | Only pay for what you consume |
| Updated models | AWS improves models automatically |
When to Use What?
| Your Goal | Use This |
|---|---|
| Add AI features quickly | Managed AI Services |
| Build custom ML models | SageMaker |
| Use LLMs for GenAI apps | Bedrock |
| No ML team, need AI | Managed AI Services |
| Full control over models | SageMaker |
| Pre-built conversational AI | Lex |
Important Point: Managed AI Services are for “add AI to apps” scenarios. SageMaker is for “build and train models.” Bedrock is for “use foundation models.”
Service Comparison by Category
Computer Vision
| Service | Best For | Key Capability |
|---|---|---|
| Rekognition | Image/video analysis | Face detection, moderation, object recognition |
| Textract | Document data extraction | Forms, invoices, tables, key-value pairs |
Language & Speech
| Service | Best For | Key Capability |
|---|---|---|
| Transcribe | Speech-to-text | 100+ languages, speaker diarization, Call Analytics |
| Polly | Text-to-speech | 100+ voices, 40+ languages, neural TTS |
| Translate | Translation | Neural MT, 75 languages |
NLP & Conversational
| Service | Best For | Key Capability |
|---|---|---|
| Comprehend | Text analytics | Sentiment, entities, PII detection, topics |
| Lex | Chatbots/voice bots | NLU + ASR + TTS, powers Alexa |
Search
| Service | Best For | Key Capability |
|---|---|---|
| Kendra | Enterprise search | Semantic search, natural language queries |
Human Review & Crowdsourcing
| Service | Best For | Key Capability |
|---|---|---|
| Augmented AI | ML prediction validation | Automated human review workflows |
| Mechanical Turk | Crowdsourcing | Human intelligence marketplace |
Healthcare AI
| Service | Best For | Key Capability |
|---|---|---|
| Comprehend Medical | Medical NLP | Extracts medical entities, PHI detection |
| Transcribe Medical | Medical speech-to-text | Clinical terminology, HIPAA-eligible |
| HealthScribe | Clinical documentation | Auto-generates clinical notes with AI |
Common Use Cases Across Services
| Use Case | Services | Example |
|---|---|---|
| Contact Centers | Transcribe, Lex, Comprehend | Speech-to-text + sentiment + chatbot |
| Content Moderation | Rekognition, Comprehend | Image + text moderation |
| Document Processing | Textract, Comprehend | Extract data + analyze content |
| Accessibility | Transcribe, Polly | Subtitles + voice output |
| Customer Support | Lex, Comprehend, Kendra | Chatbot + sentiment + knowledge search |
| Healthcare | Comprehend Medical, Transcribe Medical, HealthScribe | Clinical documentation, PHI extraction |
| ML Data Labeling | Mechanical Turk, Augmented AI | Training data, model validation |
Other Managed AI Services
AWS also offers specialized predictive, anomaly detection, and hardware services for specific use cases:
| Service | Category | What It Does |
|---|---|---|
| Amazon Personalize | Recommendations | Real-time personalized product recommendations (similar to Amazon.com) |
| Amazon Forecast | Time-series | Predict business metrics using historical data and seasonal patterns |
| Amazon Lookout for Vision | Anomaly Detection (Images) | Detect manufacturing defects and product anomalies |
| Amazon Lookout for Equipment | Anomaly Detection (Sensors) | Predict equipment failures from sensor data |
| Amazon Lookout for Metrics | Anomaly Detection (Metrics) | Detect anomalies in business KPIs and metrics |
| AWS Hardware AI | Edge Computing & Education | Panorama (edge CV, support ending), DeepRacer (service closed), DeepComposer (EOL) |
Note: Predictive/anomaly services have no free tier and are typically used for enterprise-scale applications. Hardware devices are one-time purchases for edge AI and education.
Access Requirements
🔴 AWS account required for all services
🟢 Free tier available for most services (first 12 months)
Pricing Notice
⚠️ Pricing Disclaimer: AWS pricing is subject to change. Free tier availability and pricing shown are based on information available as of January 2026. Always verify current pricing at the official AWS AI Services pricing page.
TL;DR
AWS Managed AI Services = Pre-built AI APIs. No ML required. Just call the API.
- 13 core services across 7 categories (Vision, Language, NLP, Conversational, Search, Healthcare, Human Review)
- Free tier for most services (first 12 months, some variations)
- Pay-per-use pricing — only pay for what you consume
- Best for: Quick AI integration, no ML expertise needed
- Alternative: Use SageMaker for custom ML, Bedrock for GenAI
Note: AWS also offers hardware and education programs (Panorama, DeepRacer, DeepComposer; with service lifecycle changes) and predictive services (Personalize, Forecast, Lookout family) for specialized use cases.
Resources
AWS AI Services Official overview of all AWS AI/ML services.
AWS AI Services Pricing Detailed pricing for all managed AI services.
AI Services Documentation Documentation hub for all AI services.