...pick a product or platform
Jump to navigation
Jump to search
| AI Solver map |
|---|
|
Your current page is in bold
0 · Start
1 · Predict & analyze
tables, numbers, events
Predict values
Predict categories Forecast time series Cluster / segment Find outliers Make recommendations Find key drivers 2 · Understand unstructured data
text, documents, images, audio
3 · Generate & converse
4 · Decide & act
5 · Build it right
applies to every path
|
YouTube ... Quora ...Google search ...Google News ...Bing News
- AI Solver ... Model Search ... ...choose a model ... ...deploy AI ... ...build an agent
Use this page when you know what you want to do and need to know what to buy, use or build on. Product categories last longer than product names, so pick the category first, then shortlist two or three products and test them on your own use case.
Names change often: Google's Vertex AI became the Gemini Enterprise Agent Platform in April 2026, Microsoft's Azure AI Foundry is now Microsoft Foundry, and Microsoft retired Azure AI Anomaly Detector and Personalizer on 1 October 2026. Examples below are as of October 2026.
Build, buy or assemble?
- Buy (ready-made software) when the problem is common and not what makes you different: meeting notes, coding assistance, help-desk deflection.
- Assemble (a platform plus your data, prompts and integrations) for most business use cases: you control quality without running models yourself.
- Build (your own models and infrastructure) when your data, workflow or intellectual property is the advantage and you can staff evaluation and operations.
The landscape by category
| Category | Use it to... | Commercial examples | Open-source / self-hosted |
|---|---|---|---|
| AI assistants (chat apps) | Boost individual productivity: writing, research, analysis | ChatGPT, Claude, Gemini, Microsoft Copilot, Perplexity, Meta AI, Grok | Open WebUI with local models |
| Enterprise assistants and search | Answer questions from company knowledge, with permissions | Microsoft 365 Copilot, Gemini Enterprise, ChatGPT Enterprise, Claude Enterprise, Glean | Self-built RAG (...answer questions from my data) |
| Foundation-model APIs | Build your own AI features | OpenAI, Anthropic, Google Gemini API, Mistral, Cohere, xAI, DeepSeek, Amazon Nova | — |
| Cloud AI platforms | Build, deploy and govern at enterprise scale | Amazon Bedrock and SageMaker AI; Microsoft Foundry and Azure Machine Learning; Google Gemini Enterprise Agent Platform; Databricks; Snowflake Cortex; IBM watsonx | MLflow, Kubeflow |
| Model hubs and inference providers | Find and run open models | Hugging Face, OpenRouter, Together AI, Fireworks, Groq, Cerebras, Replicate | vLLM, SGLang |
| Local and on-device runtimes | Run models on your own hardware | LM Studio, Apple Foundation Models, Android Gemini Nano | Ollama, llama.cpp, MLX, ONNX Runtime, ExecuTorch |
| Agent builders and frameworks | Automate multi-step work with tools | OpenAI AgentKit, Microsoft Copilot Studio and Agent Framework, Google ADK and Agent Platform, Amazon Bedrock AgentCore, Salesforce Agentforce | LangGraph, CrewAI, LlamaIndex, Pydantic AI, smolagents, goose, n8n |
| Agent standards | Connect agents to tools, data and each other | — | MCP, A2A, AGENTS.md, Agent Skills |
| Coding assistants and agents | Write, review, test and fix code | GitHub Copilot, Cursor, Claude Code, OpenAI Codex, Gemini CLI, Windsurf, Amazon Kiro, JetBrains AI | Cline, Aider, OpenHands |
| Computer-use and browser agents | Operate websites and apps without APIs | ChatGPT agent and Atlas browser, Claude in Chrome, Gemini computer use, Perplexity Comet | Browser Use |
| Vector databases and search | Semantic search and RAG | Pinecone, Weaviate Cloud, Qdrant Cloud, Zilliz (Milvus), Elastic, MongoDB Atlas, Azure AI Search | pgvector, Chroma, OpenSearch, FAISS, Milvus, Qdrant |
| Document AI and OCR | Extract data from PDFs, scans and forms | Azure AI Document Intelligence, Google Document AI, Amazon Textract, Mistral OCR, LlamaParse, Reducto | Docling, Unstructured, PaddleOCR, Tesseract |
| Speech and voice | Transcription, text-to-speech, voice agents | OpenAI Realtime, Gemini Live API, ElevenLabs, Deepgram, AssemblyAI, Azure Speech, Amazon Transcribe and Polly | Whisper, Kokoro, Piper, LiveKit Agents, Pipecat |
| Image, video and music generation | Creative assets and media | OpenAI image models and Sora, Google Imagen, Veo and Lyria, Midjourney, Adobe Firefly, Runway, Kling, Luma, Suno | FLUX, Stable Diffusion, ComfyUI |
| AutoML and classic ML | Tabular prediction without heavy coding | DataRobot, H2O Driverless AI, cloud AutoML, Prior Labs (TabPFN) | AutoGluon, scikit-learn, XGBoost, LightGBM, CatBoost, PyCaret |
| MLOps / LLMOps, evals and observability | Track experiments, test prompts, monitor production | Weights & Biases, LangSmith, Arize, Braintrust, Datadog LLM Observability | MLflow, Langfuse, Arize Phoenix, promptfoo, OpenTelemetry |
| Guardrails, security and governance | Filter content, red-team, manage AI risk and compliance | Azure AI Content Safety, Amazon Bedrock Guardrails, Lakera, Credo AI, IBM watsonx.governance | Llama Guard, NeMo Guardrails, Guardrails AI, garak |
| Data labeling and synthetic data | Create training and evaluation data | Scale AI, Labelbox, SuperAnnotate, Snorkel | Label Studio, CVAT, SDV |
| Workflow automation with AI steps | Connect business apps and add AI to processes | Zapier, Make, Microsoft Power Automate, UiPath | n8n |
| Vertical (industry) AI apps | Buy domain expertise ready-made | Legal (Harvey, CoCounsel); clinical notes (Abridge, Microsoft Dragon Copilot); customer service (Sierra, Decagon, Intercom Fin) | — |
Vendor checklist
- Data handling — is your data used for training? Retention period? Where is it stored and processed?
- Security and compliance — certifications (SOC 2, ISO 27001, ISO/IEC 42001), single sign-on, audit logs, admin controls
- Openness — choice of models, MCP and A2A support, data export; how hard is it to leave?
- Quality on your task — run your eval set in a trial, not just a demo (...evaluate and trust AI)
- Pricing model — per seat vs. per use; what happens to cost at ten times the volume?
- Roadmap risk — products are renamed, merged and retired often; prefer standards and portable designs
Learn more
- Choosing a generative AI service | AWS Decision Guides
- Choose a design pattern for your agentic AI system | Google Cloud
- AI agent orchestration patterns | Microsoft
- Agentic AI Foundation (MCP, A2A, AGENTS.md, goose) | Linux Foundation
- Models | Hugging Face ... click on Sort: Trending
- Model Search ... Algorithms
Where next? ◀ Build it right · ...choose a model · ...deploy AI · ...build an agent · ▶ ...evaluate and trust AI