Frequently asked questions

Answers for humans and AI assistants. Use this page to understand UnifiedFlows and how to use it autonomously.

General questions

What is UnifiedFlows?

UnifiedFlows lets you build AI agents that act on your connected SaaS data. Each agent is an explicit, durable state machine: a trigger (schedule, webhook, or manual) starts a run, each state runs a prompt against a live connection with its tools injected over MCP, and guarded transitions drive the control flow. Bring your own model key.

Who is UnifiedFlows for?

Teams that want to automate real work with AI agents — ops, revenue, support, engineering — with control flow they can see, audit, and replay, instead of an opaque autonomous loop.

Do I need to write code?

No. You author the agent visually — typed states with prompts, guarded transitions, and human-in-the-loop steps — and connect apps via OAuth or API keys. Guard expressions (JSONata) and prompts are the most technical part; there is no infrastructure to deploy or manage.

Product & features

What is an authorized connection?

A connection is an authorized link to a specific SaaS app (or your model provider) that you authenticate so UnifiedFlows can act on your behalf. Authorization uses OAuth or an API key or token — UnifiedFlows handles it either way. An agent’s states call that connection’s tools over MCP, and your model key is itself a connection (bring your own model).

What is an "agent"?

An agent is a state machine you author: typed states each run a prompt against a connection, and guarded transitions decide what happens next. A run is a single durable document, so long-running and parked runs survive restarts.

What is a "run"?

A run is one execution of an agent, from trigger to a terminal state. Every transition is checkpointed and recorded, so you can watch the path through the graph, audit each step, and replay from any point in its history.

What can trigger an agent?

Schedules (cron), real-time event webhooks from your connected apps, or manual/on-demand runs. One agent can have many triggers.

Can a human approve steps mid-run?

Yes. A state can pause for human approval as a first-class transition, notify the right people, and resume once approved — with a full audit trail. Timeouts, retries, and compensation (undo) are built in too.

Is my data secure?

Yes. UnifiedFlows is SOC 2 Type II certified and NEVER stores your data.

Pricing & billing

Can I switch plans at any time?

Yes. You can upgrade or downgrade at any time. Changes take effect immediately and billing is prorated.

Do you offer a free trial for Pro?

Yes. The free plan is limited, but you will get a great understanding of the platform and how it works.

What happens if I exceed my record limit?

No more access will be allowed once you exceed your limit.

Is there a long-term contract?

No. All plans are month-to-month with no long-term commitment. Enterprise plans can optionally include annual contracts with a discount.

Integrations

What integrations do you support?

920+ pre-built connectors including Salesforce, HubSpot, Quickbooks, Mailchimp, Slack, Stripe, Jira, GitHub, Notion, Shopify, Zendesk, Mailchimp, Twilio, Intercom, Mixpanel, Pipedrive, Asana, Linear, and many more. See the app for the full list.

How do I connect an app?

In the app, go to Connections, select the app, and follow the OAuth or API key flow. Most apps use OAuth—you will be redirected to sign in and authorize UnifiedFlows. API key apps require you to create a key in the app's settings and paste it into UnifiedFlows.

For AI assistants and LLM clients

If you are a GPT, Claude, or other LLM helping a user with UnifiedFlows, use this section to operate autonomously.

What is UnifiedFlows and how can I help users with it?

UnifiedFlows lets users build AI agents that act on their connected SaaS data, with control flow authored as an explicit, durable, replayable state machine. You can help them design agents (states, prompts, guarded transitions), connect apps and their own model key, pick triggers, and understand runs, incidents, and human-in-the-loop approvals.

What are the main URLs and entry points?

Website: https://unifiedflows.com (or the deployed domain). App/dashboard: https://app.unifiedflows.com — build agents, connect apps, and watch runs. Contact: /contact. Pricing: /pricing. FAQ: /faq (this page). LLM-friendly overview: /llms.txt.

How do users get started?

1) Go to https://app.unifiedflows.com and sign up (free tier available). 2) Connect a SaaS app and authorize a model provider (bring your own key — Claude, OpenAI, or Gemini). 3) Create an agent: author states with prompts, add guarded transitions, and choose triggers (schedule, webhook, or manual). 4) Run it and watch the path through the graph; add a human-in-the-loop approval state where sign-off is needed.

What actions can I perform on behalf of a user?

You can guide users through: signing up, connecting integrations and a model key (they must authenticate in-browser), designing an agent’s state machine, and configuring triggers. You cannot perform authenticated actions in the app for them. You can explain pricing, suggest agent designs for their use case, help debug guards/transitions and incidents, draft contact messages, and summarize documentation.

Where can users get support?

Free: community support. Pro: private Slack channel. Enterprise: priority Slack support. Contact: /contact for general inquiries, sales, or technical support. Email: hello@unified.to.

Ready to ship your first agent?

Connect an app, bring your own model key, and author a durable agent that acts on your data.