Global Tech Stack
All stacks

Build your everyday AI workflow

Choose an assistant for daily work, then add research, an AI engine, and automation only when you need them.

  1. 1. Set your direction
  2. 2. Choose your tools

Make this stack fit your work

Keep the tools that work for you. Add only what is missing. No account needed.

Choose for the job first, then compare the full cost.

Before you commit

Start with one assistant or one engine. Review usage allowances, data handling, and permissions. Local models need suitable hardware; hosted APIs can incur usage charges. Availability and currency depend on your region; confirm both before purchasing.

How we choose the options

Options come from the products assigned to this workflow, not their commission rates. They are not tested integration or compliance bundles. Published features and integrations need checking against your chosen plans. Prices are not added into a stack total.

One useful task. Then build from there.

An assistant may be all you need. Add automation only when it helps.

  1. Pick one task

    Try summarizing an approved document. Start with an assistant, not a whole system.

  2. Connect only if needed

    For repeatable work, add an AI engine and automation. Set usage limits and protect credentials.

  3. Test safely

    Use non-sensitive data. Check the inputs and permissions before turning the workflow on.

  4. Review the result

    Keep verified sources and a human approval step before publishing or sending messages.

Connecting an AI engine to n8n

Choose an API or local runtime that n8n can access. If you use Ollama, it must be reachable from your n8n instance. Add a trigger, a model step, and a destination, then check the chosen node's inputs and permissions.

A setup plan, not a ready-made integration.

n8n setup guide
1AI assistant
For daily work

Choose one chat assistant for drafting, explaining, and reviewing. You can stop here; an API is not needed for ordinary chat use.

ChatGPT

A conversational AI workspace for drafting, learning, document analysis, and research, with connected apps where available.

What to check

Answers and calculations can be incorrect

  • Conversational drafting and editing
  • Document summarization and analysis
  • Web-assisted research
Claude

An AI assistant for writing, document work, coding questions, and tasks supported by connected workplace tools.

What to check

Connector access can require administrator approval

  • Conversational writing and revision
  • Document analysis
  • Coding assistance
Google Gemini

Google's conversational AI assistant for learning, writing, and tasks using eligible connected Google apps.

What to check

Connected-app availability varies by account and settings

  • Conversational assistance
  • Writing and learning support
  • Multimodal prompts
2AI research
Optional

Add source-linked research if your assistant does not cover it well. Check the original sources before using an answer.

Perplexity

An AI research and search service that synthesizes answers with links to supporting sources.

What to check

A citation does not guarantee a claim is supported

  • Web search with synthesized answers
  • Links to supporting sources
  • Conversational follow-up research
3AI engine
For connected workflows

Choose one hosted model API or a local runtime when you need programmatic access. App subscriptions and API allowances differ; confirm any included credits and additional usage charges.

OpenAI API

Programmatic access to OpenAI models for applications, structured workflows, and tool-enabled AI features.

What to check

Metered usage can grow with repeated requests

  • Programmatic model access
  • Text generation
  • Tool calling
Claude API

Anthropic's developer API for building applications that use Claude models, with tool use and supported client libraries.

What to check

Usage allowances and billing require separate review

  • Programmatic Claude model access
  • Messages API
  • Tool-enabled application workflows
Gemini API

Google's developer API for model-backed applications, multimodal requests, and function-calling workflows.

What to check

Model capabilities and quotas differ

  • Programmatic Gemini model access
  • Model-dependent multimodal inputs
  • Function calling
Ollama

A model runtime for downloading, managing, and running supported AI models locally, with a local API.

What to check

Hardware capacity limits model choice and performance

  • Local model execution
  • Model download and management
  • Local HTTP API
4Workflow automation
Optional

n8n has documented model nodes for OpenAI, Anthropic, Gemini, and Ollama. Connect your chosen engine using its credentials or supported gateway access; selecting tools here does not connect them.

n8n

n8n is a fair-code workflow automation platform combining visual workflows, integrations, code-oriented flexibility, and AI capabilities, with hosted and self-managed options for different operating requirements.

What to check

Self-hosting requires operations work

  • Visual workflow automation
  • Application and API integration
  • AI workflow capabilities
5Shared knowledge
Optional

Keep approved prompts, source links, and reviewed output in a workspace you already use. Add integrations only after checking access permissions.

Notion

Notion combines documents, knowledge bases, and database-driven trackers in a customizable workspace for individuals and teams that want connected information and work.

What to check

Workspace design needs time and ownership

  • Databases
  • Collaboration
  • Templates
Google Docs

Google Docs is a cloud-based document editor that allows individuals and teams to create, edit, and collaborate on documents in real time from any device.

What to check

Limited offline features compared to desktop editors

  • Live co-editing and suggestions
  • Comments and version history
  • Document tabs and smart chips