AI agents with real tool access

Write your workflow. AI runs it.

Describe what you want to automate in plain English. Dugong hands your instructions to an AI agent — it reads them, reasons through each step, and calls your apps directly. No builder. No blocks. No code.

Trusted by teams at
VercelNotionLinearRampResend VercelNotionLinearRampResend
Live example

You write this. The agent does this.

A real workflow — written in plain English on the left, executed by the AI agent on the right.

Your workflow description plain english.txt
▸ Trigger — when this happens
A deal is marked "Closed Won" in Salesforce
▸ Steps — in your own words
1
Look up the contact in Clearbit — get company size, industry, and LinkedIn
2
Create an onboarding project in Notion using our "New Client" template
3
If the deal is over $10k, post to Slack #csm-team and assign a CSM. Otherwise skip.
4
Send a personalised welcome email from the AE's Gmail
5
Log the win in our Google Sheets revenue tracker
Agent executing now runtime · live
Acme Corp · $24,000 Done ✓
09:41:02
Trigger received → Salesforce webhook
opportunity.stage = "Closed Won" contact.email = "sarah@acmecorp.com"
09:41:03
Step 1 → tool call: clearbit.enrich()
clearbit.enrich({ email: "sarah@acmecorp.com" }) ← company_size: 340 · industry: "FinTech"
09:41:05
⟳ reasoning: is $24k > $10k?
Yes → taking the CSM path slack.postMessage({ channel: "#csm-team" }) ← assigned CSM: Jordan Lee ✓
09:41:08
Step 5 → tool call: sheets.append()
sheets.append({ sheet: "Revenue 2026" }) ← row added ✓
All steps completed · 7 tool calls · 6.2s
How it works

No builder. Just language.

Traditional automation tools make you draw a diagram. Dugong just reads what you write and does the work.

01

Describe it in plain English

Write your trigger and steps the way you'd explain them to a colleague. Name any app. No special syntax — just plain language.

02

An AI agent reads and reasons

Your description is passed to an AI agent. It understands intent, handles conditions, and decides what to pass between apps — live, at runtime.

03

It calls your apps directly

The agent executes each step with direct tool calls to Slack, Gmail, Salesforce, Notion and more — in the right order, with the right data.

Use cases

Real workflows, described in seconds.

Every team has repetitive processes that eat hours every week. Just write them down.

Qualify and route every inbound lead automatically

The moment a form lands, the agent enriches the contact, scores the lead, routes it to the right rep, and creates the Salesforce deal — all before your rep has opened their laptop.

Try this workflow →
Form submitted
Typeform, HubSpot, or any webhook
Enrich with Clearbit
Company size, industry, tech stack
Tool call
Agent scores and routes
AE for high-fit, nurture for low
AI reasons
Create Salesforce deal
Owner assigned, stage set, notes added
Tool call
Notify rep in Slack
Card with lead score and key context
Tool call

Full client onboarding the moment a deal closes

Write "when a deal closes, create a Notion project, send a welcome email from the AE, and book a kickoff call." The agent does exactly that — every time, tailored to the account size.

Try this workflow →
Deal marked Closed Won
Salesforce trigger fires instantly
Create Notion onboarding
From template, client details pre-filled
Tool call
Send personalised welcome email
From the AE's Gmail, not a generic address
Tool call
Book kickoff call
Calendly invite, CSM auto-assigned by size
AI reasons

Publish once — distributed everywhere, per platform

Drop a blog post in Notion and write "when this is published, write a Twitter thread, a LinkedIn post, and an email excerpt, then schedule them." The agent reads the post and does it.

Try this workflow →
Post published in Notion
Status change triggers the agent
Agent reads and understands it
Tone, key points, audience inferred
AI reads
Writes channel variants
Twitter thread, LinkedIn post, email snippet
AI writes
Schedules in Buffer + SendGrid
Right format, right time, per platform
Tool call

Critical issue raised — team paged in seconds

Write "when a GitHub issue is labelled critical, page the on-call in Slack, open a Linear incident, and create a Notion post-mortem." Done. Every single time.

Try this workflow →
GitHub issue labelled "critical"
Webhook fires instantly
Page on-call in Slack
@mention with issue title and link
Tool call
Open Linear incident
Linked to GitHub issue, team assigned
Tool call
Create Notion post-mortem
Template with timeline pre-filled
Tool call
Under the hood

What makes it actually different.

Dugong isn't a visual builder with an "AI" button bolted on. The AI is the engine — it reads, reasons, and acts.

Language is the interface

Your description is the workflow. No canvas, no blocks, no config panels. The agent reads what you wrote and figures out the rest.

Conditional logic for free

Write "if the deal is over $10k" or "unless they've been contacted" — the agent understands and branches accordingly, every run.

Failures handled intelligently

If a tool call fails, the agent doesn't crash — it retries, tries an alternative, or flags the issue with full context.

200+ tools, called directly

Mention Gmail, Slack, Salesforce or Notion — the agent knows how to call them. No manual API setup. No OAuth headaches.

Context carried across steps

Each step isn't isolated. The agent carries what it learned in previous steps — so it can personalise and adapt throughout.

Full visibility into every run

Every tool call, every decision, every piece of data — logged in plain language so you always know what the agent did and why.

Why Dugong

Other tools automate. Dugong understands.

94%
Steps executed correctly first run
12×
Faster than building manually
3.2M
Agent-powered runs per month
67%
Fewer failures vs rule-based tools
200+ integrations

If you can name it, the agent can call it.

Just write the app name in your steps. The agent handles auth, API calls, and data mapping automatically.

GmailSlackSalesforce StripeNotionGitHub HubSpotShopifyAirtable SendGridGoogle SheetsAWS LinearGoogle CalendarClearbit + 185 more
Pricing

Simple pricing. Free to start.

No credit card required. Upgrade when your workflows grow.

Free
$0 /mo
For individuals trying out plain-English automation.

  • 25 agent runs per month
  • Plain English workflow builder
  • 20 core integrations
  • Full run logs and history
  • Community support
Get started free
Growth
$25 /mo
For individuals automating personal workflows and side projects.

  • 250 agent runs per month
  • All 200+ integrations
  • Multi-step agent reasoning
  • Automatic retry & failure handling
  • Email support
Upgrade to Growth →
Custom
Let's talk
For organisations with high volume and custom limits.

  • High volume & custom limits
  • Custom AI model tuning
  • SSO and SAML
  • Audit logs and permissions
  • On-prem deployment option
  • Dedicated success manager
Contact us ↗
FAQ

Questions, answered.

Everything you need to know about automating with plain English.

What is Dugong?
Dugong is an AI-native workflow automation platform. You describe a workflow in plain English — a trigger and the steps you want — and an AI agent reads it, reasons through each step, and executes it by calling your apps directly. There is no visual builder, no blocks, and no code required.
How is Dugong different from Zapier or Make?
Traditional automation tools make you build a rigid diagram of triggers and actions. Dugong replaces that with plain language: the AI agent interprets your intent at runtime, handles conditional logic like "if the deal is over $10k", carries context between steps, and recovers from failures instead of breaking. The description is the workflow.
Do I need to know how to code?
No. If you can describe a process to a colleague in writing, you can build it in Dugong. You name the apps and the steps in plain English, and the agent handles the API calls, authentication, and data mapping automatically.
Which apps and integrations does Dugong support?
Dugong connects to 200+ tools including Gmail, Slack, Salesforce, Notion, Stripe, GitHub, HubSpot, Shopify, Airtable, Google Sheets, Linear, Google Calendar and more. You simply mention an app in your steps and the agent knows how to call it.
How does the AI agent handle errors and conditional logic?
The agent understands conditions written in plain English and branches accordingly on every run. If a tool call fails, it does not crash — it retries, attempts an alternative, or flags the issue with full context. Every decision and tool call is logged in plain language so you always know what happened and why.
Is there a free plan?
Yes. The Free plan is free forever and includes 25 agent runs per month, with no credit card required. Paid plans start at $25/month for 250 runs and all 200+ integrations, with Pro at $100/month for 1,000 runs.
Is my data secure with Dugong?
Dugong runs each workflow with scoped access to only the apps you connect, and gives you a full audit log of every run. Custom plans add SSO, SAML, granular permissions, and an on-premise deployment option.

Stop configuring.
Start describing.

Join 4,000+ teams who shipped their first automation in under five minutes — just by writing it down.

Get started free — no credit card →