Skip to content
Connector Scout
–saved records

Guides

AI connector guides

Guides interpret and compare the documentation-verified connector relationships in this directory — they explain what the records mean, not just what exists.

Every claim in a guide traces back either to a provider’s own documentation or to a relationship record we have verified against it. Where a guide compares platforms, the counts and capability statements are read from the same dataset that powers the relationship directory, so a guide cannot quietly drift from the records behind it.

Start here: What are AI connectors? — it establishes the vocabulary the other guides rely on.

Compatibility reference

Which of ChatGPT, Claude and Gemini can actually connect to a given app, and what each connection is documented to do.

Compatibility reference · Documentation verified

AI connector compatibility matrix: ChatGPT vs Claude vs Gemini

A searchable matrix of 16 widely used business applications across all three platforms: 48 documentation-verified relationships showing the product surface each connection lives on, whether it is retrieval only, and which actions are actually documented. Every cell is generated from the underlying record, so it stays correct as records change.

Updated 9 Sept 2026Documentation verified

Open the compatibility matrix

Practical workflow library

What to actually do once ChatGPT, Claude or Gemini is connected to your work apps.

Practical workflow library · Documentation verified

15 Practical AI Connector Workflows for Work in 2026

Fifteen concrete workflows — morning briefs, meeting follow-up, CRM logging, pipeline risk, incident timelines, lead intake — each with its verified app stack, automation level, approval points, a prompt to start from, and the limitations that apply today.

Updated 4 Sept 2026Documentation verified

Browse the 15 practical connector workflows

Security and governance

What to inspect before an assistant is allowed to reach your systems, and what changes once it can act on them.

Permissions explainer · Documentation verified

Read-only vs. write-capable AI connectors: what changes when AI can take action

The difference between retrieval and side effects, the four layers that decide whether an action can actually run, how ChatGPT, Claude and Gemini handle write actions and approvals across their different surfaces, a staged enablement sequence, and why MCP tool annotations are hints rather than guarantees.

Updated 16 Sept 2026Documentation verified

Open the read-only versus write-capable guide

Security review · Documentation verified

How to evaluate an MCP server before connecting it

Fifteen checks covering publisher provenance, the authority each tool actually holds, OAuth and token handling, tool poisoning and prompt injection, approvals, logging and revocation — plus what ChatGPT, Claude and Gemini Spark do and do not vet on your behalf, and a sandbox pilot test to run before rollout.

Updated 16 Sept 2026Documentation verified

Open the MCP server security checklist

Original research

Connector Scout syntheses of public evidence, with every figure labeled by the strength of the evidence behind it.

Original research · 25 sources reviewed

The Real ROI of AI Connectors: What Enterprise Evidence Actually Shows

How connectors change productivity, automation, cost and ROI compared with stand-alone LLMs — across four controlled studies, 18 named enterprise deployments and three vendor-commissioned economic models, with an illustrative ROI calculator.

Updated 5 Aug 2026Evidence reviewed

Read the real ROI of AI connectors

Evidence reviews

Syntheses of public research on how AI is actually used at work, cited inline from each claim.

Evidence review · Workplace AI adoption

AI at Work in 2026: Adoption Is Broad. Automation Is Still Shallow.

Google’s ATLAS study, Anthropic’s Claude analysis, Microsoft’s Copilot and agent data, Federal Reserve adoption estimates and a 7,137-worker randomised trial, read together: adoption is broad, but end-to-end automation is still rare.

Updated 11 Aug 2026Evidence reviewed

Read why workplace AI automation is still shallow

Connector fundamentals

How the underlying integration layers differ, so you can choose an architecture instead of guessing at labels.

Fundamentals · Connector architecture

AI Connectors Explained: MCP vs. APIs vs. A2A vs. Native Integrations

APIs, the Model Context Protocol, native AI integrations and Agent2Agent solve different problems and increasingly work together. This explainer maps the four layers, cites the primary specifications inline and ends with a decision tree.

Updated 17 Aug 2026Documentation verified

Compare MCP, APIs, A2A and native AI integrations

Flagship workflow guide

Long-form documentation for building a repeatable AI research workflow on official APIs.

AI SEO Workflow Guide

Build a Daily AI SEO Analyst

A step-by-step build for a daily, read-only AI SEO analyst wired to Search Console, Analytics, Core Web Vitals and third-party SEO APIs, with a human reviewing every recommendation.

Updated 3 Aug 2026Documentation verified

Read the Daily AI SEO Analyst build guide

Platform comparisons

Which assistant reaches the systems your workflow depends on.

Documentation verified · v1.0

ChatGPT vs Claude vs Gemini Connectors: Which Is Best for Your Workflow?

Verified app coverage, MCP surfaces, read/write actions and best-fit workflows across all three platforms, derived from the central connection records.

Updated 4 Aug 2026Documentation verified

Read ChatGPT vs Claude vs Gemini Connectors (2026)

Comparison guides

Platform against platform, and connector model against connector model, using verified records.

Fundamentals

What a connector is, what the vendor vocabularies mean, and what they can actually do.

Permissions and governance

Reviewing scopes, authorizing accounts, administrator approval and revocation paths.

Troubleshooting

What to check when a connection is missing, greyed out, or returning nothing.

Research behind the guides