Which AI models can I connect?

OpenAI, Anthropic Claude, Google Gemini, xAI Grok, DeepSeek, Groq, Mistral, Ollama (self-hosted), Perplexity and any OpenAI-compatible endpoint — 14+ providers and hundreds of models via live fetch. You bring your own API keys (BYOK) and AgenTorQ routes each task to the right model with cost control and fallback.

Connectors Connect anything

Connect AI to Salesforce, enterprise apps, databases, and every model.

Q: What can AgenTorQ connect to?

AgenTorQ connects to CRM (Salesforce, Dynamics 365, HubSpot), productivity suites (Microsoft 365, Google Workspace, Slack, Teams, Zoom), storage (Google Drive, OneDrive, Dropbox, SharePoint, Amazon S3), developer tools (GitHub, GitLab, Bitbucket, Azure DevOps, Jira), enterprise systems (SAP, ServiceNow, Oracle, Workday), databases (SQL Server, Oracle, PostgreSQL, MySQL, Snowflake), 14+ AI providers, and any REST or GraphQL API, MCP server or webhook. In total, 600+ apps are reachable through the connector layer plus REST, databases and MCP.

Q: How do I connect a database or a custom API?

Databases (SQL Server, Oracle, PostgreSQL, MySQL, Snowflake) connect through the encrypted credential vault with read-first, parameterized SQL and DML guards; writes require approval. Any HTTP service connects as a REST or GraphQL connector with a typed schema, and you can register custom connectors and custom actions with role-based access.

AgenTorQ is an enterprise AI platform with one governed connector layer. Ground AI Workmates in Salesforce, Microsoft 365, Google Workspace, Slack, GitHub, Jira, SharePoint, your databases, REST and GraphQL APIs and MCP servers — then route each task across 14+ AI providers on your own keys. 600+ apps, per-user OAuth, every action auditable.

One connector layer — the same connectors run in the Cloud workspace and inside Salesforce.

Book Demo Explore Platform

600+ connectors Per-user OAuth REST & GraphQL Databases MCP servers 14+ AI models, BYOK

Salesforce Microsoft 365 Google Slack GitHub Jira Databases REST / MCP

Connector layer OAuth · vault · governance

AI Workmate Intent Router

Route to the right model — 14+ AI providers, your keys

Governed action approved · audited Cited answer grounded · sourced

How AgenTorQ connects

Four honest ways to reach your systems

Every connector goes through the same governance: per-user OAuth or a vaulted credential, scoped access, action approvals for writes, and an immutable audit trail. The model only ever sees a masked, grounded prompt — never your raw keys.

Connector layer (Nango)

600+ SaaS apps behind one framework. Each user connects with their own OAuth 2.0 grant, so Workmates act with that user’s permissions. Tokens are stored and refreshed for you.

REST & GraphQL

Any HTTP/JSON service becomes a connector with a typed schema. API-key, token or OAuth credentials live in the vault; in the Salesforce edition Remote Site Settings are wired automatically.

Databases

Connect SQL Server, Oracle, PostgreSQL, MySQL and Snowflake through the credential vault with read-first, parameterized SQL. A DML guard holds back writes until they are approved.

MCP, webhooks & custom

Add any Model Context Protocol server with OAuth 2.1 PKCE and automatic tool discovery, receive signed webhooks, or register a custom connector with role-based actions.

Straight talk on “600+”. That figure counts apps reachable through the connector layer plus REST, GraphQL, databases, MCP and webhooks. Flagship apps — Salesforce, Microsoft 365, Google Workspace, Slack, GitHub — are first-class OAuth connectors. Long-tail and enterprise systems (SAP, ServiceNow, Oracle, Workday) are reached through the connector layer, REST/OData or MCP, not bespoke certified native integrations.

Business & enterprise apps

The apps your teams already run

CRM, productivity, storage, developer tools and enterprise systems — connected once, grounded on live data, and acted on behind approvals. Each connector explains its purpose, authentication, supported actions and typical enterprise use.

CRM

3 connectors

Ground Workmates in accounts, pipeline, cases and contacts — then let them draft, update and route work with your rules enforced.

Salesforce

Purpose: Read and act on CRM records — accounts, opportunities, cases, contacts.
Auth: Per-user OAuth 2.0; USER_MODE RBAC, FLS & sharing in the Salesforce edition.
Actions: Guarded SOQL reads, create/update records, run Flows, send email.
Use cases: Pipeline risk, case triage, account 360, quote drafting.

Dynamics 365

Purpose: Query and update Dynamics CRM and ERP entities.
Auth: Per-user OAuth 2.0 via the connector layer.
Actions: Read entities, create/update rows, trigger workflows (approved).
Use cases: Lead follow-up, opportunity hygiene, service tickets.

HubSpot

Purpose: Ground on marketing and sales CRM — contacts, deals, tickets.
Auth: Per-user OAuth 2.0 via the connector layer.
Actions: Read/update contacts & deals, log activity, enroll workflows (approved).
Use cases: Lead scoring, deal summaries, campaign context.

Productivity

5 connectors

Reach email, calendars, files, chat and meetings so Workmates draft, summarize and follow up where work already happens.

Microsoft 365

Purpose: Outlook mail, calendar, files and Teams across Microsoft Graph.
Auth: Per-user OAuth 2.0 (Microsoft Graph).
Actions: Read/send mail, read calendar, read/write files, post to Teams (approved).
Use cases: Inbox triage, meeting prep, document drafting.

Google Workspace

Purpose: Gmail, Calendar, Drive and Docs, grounded per user.
Auth: Per-user OAuth 2.0 (Google).
Actions: Read/send mail, read calendar, read/write Drive files (approved).
Use cases: Email drafting, scheduling, document RAG.

Slack

Purpose: Read channels and post updates where the team talks.
Auth: Per-user OAuth 2.0 via the connector layer.
Actions: Read messages, post messages, send DM summaries (approved).
Use cases: Channel summaries, alert routing, standups.

Microsoft Teams

Purpose: Read and post in Teams channels and chats.
Auth: Per-user OAuth 2.0 (Microsoft Graph).
Actions: Read/post messages, share adaptive cards (approved).
Use cases: Team updates, approvals, notifications.

Zoom

Purpose: Pull meetings, recordings and transcripts.
Auth: Per-user OAuth 2.0 via the connector layer.
Actions: List meetings, fetch recordings/transcripts, schedule (approved).
Use cases: Meeting recaps, action-item capture.

Storage

5 connectors

Index files and objects for cited RAG. Workmates retrieve the exact document or record and answer with a source link, never a guess.

Google Drive

Purpose: Index and retrieve Drive files for cited RAG.
Auth: Per-user OAuth 2.0 (Google).
Actions: List, read and write files; folder-scoped retrieval.
Use cases: Knowledge grounding, contract Q&A.

OneDrive

Purpose: Reach personal and shared OneDrive files.
Auth: Per-user OAuth 2.0 (Microsoft Graph).
Actions: List, read and write files.
Use cases: Document RAG, file drafting.

Dropbox

Purpose: Read and write Dropbox folders.
Auth: Per-user OAuth 2.0 via the connector layer.
Actions: List, read and upload files.
Use cases: Asset retrieval, document grounding.

SharePoint

Purpose: Search sites, libraries and lists.
Auth: Per-user OAuth 2.0 (Microsoft Graph).
Actions: Read documents and lists, write list items (approved).
Use cases: Policy RAG, intranet search.

Amazon S3

Purpose: Reach object-storage buckets.
Auth: Encrypted credential vault (keys/roles) via connector layer or REST.
Actions: List and read objects; write (approved).
Use cases: Log/document retrieval, data pipelines.

Development

5 connectors

Browse repos, review and edit code, run real git operations and manage issues — the connectors behind the DevSpace copilot. See DevStudio →

GitHub

Purpose: Browse, edit and ship code with real git operations.
Auth: Per-user OAuth 2.0 / token via the connector layer.
Actions: Read repos/files, commit, branch, open PR, list, merge — over the API, no local clone.
Use cases: DevSpace copilot, code review, PR automation.

GitLab

Purpose: Read and edit GitLab repos and merge requests.
Auth: Per-user OAuth 2.0 / token via the connector layer.
Actions: Read files, commit, branch, open/merge MRs (approved).
Use cases: Repo edits, review, CI context.

Bitbucket

Purpose: Reach Bitbucket repos and pull requests.
Auth: Per-user OAuth 2.0 / token via the connector layer.
Actions: Read files, commit, branch, open PR (approved).
Use cases: Code edits, review.

Azure DevOps

Purpose: Repos, pipelines and work items.
Auth: Per-user OAuth 2.0 / PAT via the connector layer.
Actions: Read repos/work items, commit, PR, query boards.
Use cases: Backlog triage, code review.

Jira

Purpose: Read and manage issues and sprints.
Auth: Per-user OAuth 2.0 via the connector layer.
Actions: Read/create/update issues, transition, comment (approved).
Use cases: Ticket triage, sprint summaries, status roll-ups.

Enterprise systems

4 connectors

Reach ERP, ITSM and HR systems of record. These are connected through the connector layer, REST/OData or MCP — honest reach, not a claim of deep certified native integrations.

SAP

Purpose: Reach ERP data — orders, invoices, materials.
Auth: Connector layer / REST (OData) / MCP.
Actions: Read via APIs; write behind approvals where exposed.
Use cases: Order status, invoice lookup, supply context.

ServiceNow

Purpose: Incidents, requests and CMDB.
Auth: Connector layer / REST / MCP.
Actions: Read/create/update via the Table API (approved).
Use cases: Incident triage, request automation.

Oracle

Purpose: Oracle E-Business / Fusion business objects.
Auth: Connector layer / REST / MCP.
Actions: Read via APIs; write where exposed (approved).
Use cases: Finance/HR lookups, record sync.

Workday

Purpose: HR and finance data.
Auth: Connector layer / REST / MCP.
Actions: Read workers/positions via APIs; write where exposed.
Use cases: People lookups, org context, approvals.

Databases

Query your data warehouse and operational stores

Databases connect through an encrypted credential vault with read-first, parameterized access. A DML guard holds writes behind approval, and every query is logged — so Workmates can answer with real numbers, cited to the row.

SQL Server

Purpose: Query relational data with governed SQL.
Auth: Encrypted vault; parameterized, read-first.
Actions: SELECT reads; guarded writes (approved).
Use cases: Reporting, lookups, enrichment.

Oracle Database

Purpose: Query Oracle schemas.
Auth: Encrypted vault; read-first.
Actions: Reads; guarded writes (approved).
Use cases: Finance queries, record checks.

PostgreSQL

Purpose: Read and analyze Postgres data.
Auth: Encrypted vault; read-first.
Actions: Reads; guarded writes (approved).
Use cases: Product analytics, lookups.

MySQL

Purpose: Query MySQL and MariaDB.
Auth: Encrypted vault; read-first.
Actions: Reads; guarded writes (approved).
Use cases: App data, dashboards.

Snowflake

Purpose: Query the warehouse for analytics.
Auth: Encrypted vault / connector layer; scoped roles.
Actions: SELECT reads with role scoping.
Use cases: KPI answers, cohort analysis, cited metrics.

AI providers

14+ model providers — on your own keys

Models are connectors too. Bring your own API keys (BYOK), and AgenTorQ’s Intent Router sends each task to the right model with cost control and fallback. AI at cost, no per-message metering, no lock-in. See model routing →

OpenAI

Anthropic Claude

Google Gemini

xAI Grok

DeepSeek

Mistral

Groq SambaNova

Cerebras

Perplexity MiniMax

AI21

Hugging Face

Ollama · self-hosted OpenAI-compatible

OpenAI

GPT-class models for reasoning, drafting and general enterprise workloads.

Anthropic Claude

Long-context, safety-focused models for careful reasoning and analysis.

Google Gemini

Multimodal models with strong grounding and large context windows.

xAI Grok

Fast models for real-time reasoning and conversational workloads.

DeepSeek

Cost-efficient reasoning and coding models for high-volume tasks.

Groq

Ultra-low-latency inference for interactive, high-throughput workflows.

Mistral

Open-weight, efficient models for latency-sensitive and private work.

Ollama

Self-hosted open models that keep sensitive workloads inside your network.

Perplexity

Answer models with web grounding and dated, sourced results.

OpenAI-compatible

Any OpenAI-compatible endpoint — connect private or region-locked models.

Keys are stored in the credential vault and used server-side only. Add Azure OpenAI, SambaNova, Cerebras, MiniMax, AI21, Hugging Face and more; hundreds of models are available via live fetch and switchable per Workmate and per task.

Generic & API-first

If it has an API, AgenTorQ can reach it

When there is no packaged connector, build one. REST, GraphQL, MCP servers, webhooks and fully custom connectors extend the platform to anything in your stack — with the same governance and audit trail.

REST APIs

Purpose: Connect any HTTP/JSON service.
Auth: API key, OAuth 2.0 or token in the vault.
Actions: GET/POST/PUT/DELETE with typed schemas; auto Remote Site Settings (SF edition).
Use cases: Internal services, SaaS APIs, custom actions.

GraphQL

Purpose: Query GraphQL endpoints.
Auth: Token / OAuth via the vault.
Actions: Typed queries and mutations (approved).
Use cases: Product/catalog data, composite reads.

MCP servers

Purpose: Add any Model Context Protocol server.
Auth: OAuth 2.1 PKCE + Dynamic Client Registration.
Actions: Automatic tool discovery; agent calls tools with approval.
Use cases: Custom tools, internal capabilities, partner MCP.

Webhooks

Purpose: Trigger and receive events.
Auth: Signed (HMAC) delivery with timestamps and idempotency keys.
Actions: Inbound triggers, outbound signed events.
Use cases: Run lifecycle, approvals, SIEM export.

Custom connectors

Purpose: Register bespoke integrations and actions.
Auth: Vault-stored credentials; per-user OAuth where needed.
Actions: Custom actions with typed schema and RBAC; writes gated by approval.
Use cases: Proprietary systems, internal tools.

One platform, two editions

The same connectors, wherever you deploy

Connectors, credential vault, model routing and governance are shared across both editions of AgenTorQ. Only the deployment differs — pick the edition that fits, and connect the same enterprise stack.

AgenTorQ Cloud

Standalone workspace

A web workspace that needs no install. Connect Salesforce alongside Microsoft 365, Google, Slack, GitHub, databases, REST APIs and MCP servers, and start in minutes.

All connector categories, per-user OAuth
Instant start, no deployment

Launch Workspace Explore Deployments

AppExchange AgenTorQ for Salesforce

Native managed package

Runs inside Salesforce as a managed package — LWC, Apex, Flows and record pages. Data stays in Salesforce, reads and writes run in USER_MODE, and it still reaches Microsoft 365, Google, Slack, GitHub, databases and MCP.

Native security, sharing, CRUD/FLS
Admin-managed, installed in your org

View AppExchange Edition View Security

Explore the platform

Connectors are one layer — here is the rest

What a Workmate does with all that connected data, the AI-native IDE it ships with, and the governance that keeps it safe.

Connector FAQ

Questions about connecting your stack

What can AgenTorQ connect to?

CRM (Salesforce, Dynamics 365, HubSpot), productivity (Microsoft 365, Google Workspace, Slack, Teams, Zoom), storage (Google Drive, OneDrive, Dropbox, SharePoint, Amazon S3), developer tools (GitHub, GitLab, Bitbucket, Azure DevOps, Jira), enterprise systems (SAP, ServiceNow, Oracle, Workday), databases (SQL Server, Oracle, PostgreSQL, MySQL, Snowflake), 14+ AI providers, and any REST or GraphQL API, MCP server or webhook. In total, 600+ apps are reachable through the connector layer plus REST, databases and MCP.

How does AgenTorQ authenticate to my apps?

Most SaaS apps connect with per-user OAuth 2.0 through the connector layer, so every user acts with their own permissions and tokens. Databases and HTTP APIs use credentials stored in an encrypted credential vault. Custom MCP servers use OAuth 2.1 with PKCE and Dynamic Client Registration. Tokens are refreshed automatically and are never exposed to the model.

Is “600+” a set of native certified integrations?

No. The 600+ figure reflects apps reachable through the connector layer (Nango) plus REST, GraphQL, databases, MCP servers and webhooks. Flagship apps such as Salesforce, Microsoft 365, Google Workspace, Slack and GitHub are first-class OAuth connectors. Long-tail and enterprise systems such as SAP, ServiceNow, Oracle and Workday are reached through the connector layer, REST/OData or MCP — not bespoke certified native integrations.

Can AgenTorQ take actions, not just read?

Yes. Connectors support governed actions — create and update records, run Salesforce Flows, send email, commit code and open pull requests, transition Jira issues and more. Write actions are shown as approval cards, and every action is recorded in an immutable audit log. In the Salesforce edition, reads and writes run in USER_MODE with CRUD, FLS and sharing enforced.

How do I connect a database or a custom API?

Databases connect through the encrypted credential vault with read-first, parameterized SQL and DML guards; writes require approval. Any HTTP service connects as a REST or GraphQL connector with a typed schema, and you can register custom connectors and custom actions with role-based access.

Do connectors work in both the web and Salesforce editions?

Yes. The same connector layer, credential vault, model routing and governance run in both editions. AgenTorQ Cloud connects your whole stack from a standalone web workspace, and AgenTorQ for Salesforce runs natively inside Salesforce and still reaches Microsoft 365, Google, Slack, GitHub, databases, REST APIs and MCP servers.

Connect your stack

Connect Salesforce, your apps, and every model — governed.

Start in the standalone workspace or install the AppExchange edition. Same connectors, same governance, same audit trail.

Book Demo Launch Workspace Explore Platform