SUPPORT
Triage tickets automatically
New ticket → an AI agent reads it, classifies it, looks up the customer in your CRM, drafts a reply, and posts it to the right Slack channel. A human just clicks "send".
Daisy lets you connect AI to your real work — emails, spreadsheets, tickets, databases, Slack — and run it on a schedule, on a webhook, or when a human approves. No coding marathons. No vendor lock‑in.
If one of these sounds like you, you're in the right place.
You've watched your team copy-paste the same data between five tools every week. With Daisy you stitch those tools together with a few clicks, drop an AI agent in the middle, and reclaim the afternoon.
You don't want to rebuild prompt management, retries, RAG, guardrails, and observability again. Daisy gives you all of that, lets you write workflows as YAML or build them visually, and runs on your own infra.
Your team wants to use AI; you need an audit trail, role-based access, data residency, and proof you're not leaking PII. Daisy ships HIPAA and GDPR modes that enforce it for you — not bolt-on, not theatre.
A few of the most popular patterns. Every one is a real workflow other teams already run.
New ticket → an AI agent reads it, classifies it, looks up the customer in your CRM, drafts a reply, and posts it to the right Slack channel. A human just clicks "send".
A 7am workflow pulls last night's signups from your DB, summarises each one with an AI, and posts a digest to the sales channel with the top three to call.
An invoice arrives by email. Daisy reads the PDF, extracts vendor + amount + due date, files it into Sheets, and posts to your accountant's chat. Pay your bills on time.
Point Daisy at your docs, Notion, or Confluence. Employees ask questions in Slack. Daisy retrieves the right pages and answers — with citations they can click.
Publish a blog post → Daisy reads it, drafts a LinkedIn post, a Twitter thread, and three image prompts. You approve in one tap; Daisy posts everything.
A monitor fires. Daisy checks recent deploys, asks an AI whether this looks routine or serious, and only pages the on-call when it's actually a problem.
On a schedule, when an email arrives, when someone fills a form, when a webhook fires. Pick the trigger and you're half done.
Drag from a palette of 44+ connectors. Need decisions? Drop in an AI agent. Need a human in the loop? Add an approval step. Workflows are saved as plain YAML you can review.
Every run shows you step-by-step what happened, what each step received, what it returned. Something broke? Resume from the failed step in one click.
Non-engineers click around the visual canvas. Engineers diff workflows in git like any other file. Same workflow either way — pick the surface that fits the person.
name: support-triage trigger: webhook nodes: - name: lookup action: sql.select inputs: sql: "SELECT * FROM customers WHERE email=$1" params: [${email}] - name: classify action: agent.classify inputs: agent: triage-bot text: ${body} labels: [refund, technical, billing] - name: notify action: slack.post inputs: channel: "#support-${label}" text: "New ticket from ${customer.name}"
Daisy ships with first-party connectors for the tools you reach for first. Need something else? The plugin SDK lets you wire up anything with an API.
Pre-built AI agents you can drop into any workflow. Pick OpenAI, Anthropic, or run a local model — same interface either way.
Upload your docs. Daisy makes them searchable so your AI answers from your data, not from guesses.
Catch PII before it leaves your network. Block jailbreaks and toxic content. Turn on once; runs on every AI call.
Write tests for your prompts. Daisy runs them on every change so quality doesn't drift.
Try a cheap model first. Escalate to a smarter one only when needed. Routing rules live in one place.
Pause a workflow for approval. Resume when a person clicks OK. Perfect for refunds, deploys, sensitive content.
Conversations remember prior turns. Workflows remember past runs. Build real chatbots, not goldfish.
Every run is replayable. Every token is metered. Every action is in the audit log.
No vendor lock-in. Your prompts, your data, your audit trail. Run on your own infrastructure when compliance demands it.
Don't rebuild retries, RAG, guardrails, evals, observability. Daisy gives you all of it on day one so you can focus on the actual feature.
AI is a first-class citizen — not "send this through a Python operator". Multi-tenant, role-based, audit-ready from the first day.
No surprises. No per-run charges. No "contact sales for our pricing page".
forever
Everything to build and run on a single machine.
unlimited users
For teams shipping AI features to real customers.
priced to your scale
For regulated industries and large rollouts.
No. The visual editor lets you build a working workflow by dragging steps and filling in forms. Pick a template, edit a few fields, hit Run. Engineers on your team can also edit the YAML in git if they prefer — same workflow either way.
Daisy stores your credentials encrypted at rest with a key you control. Self-hosting? They never leave your network. Use the cloud? Bring your own KMS key (AWS / GCP / Azure) and Daisy can't decrypt without it.
OpenAI, Anthropic, Google Gemini, AWS Bedrock, Azure OpenAI, Ollama (local), and a built-in mock provider for development. Switch providers without changing your workflow.
Daisy has HIPAA and GDPR modes that enforce provider allow-lists, PII redaction, and audit retention automatically. Data residency controls let you keep traffic in the EU, US, or APAC region your customers expect.
You see exactly which step failed and what it received. You can fix the data and resume from there, skip the step, or just hit Retry. Long-running workflows survive worker restarts — Daisy picks up where it left off.
Yes. Drop in a small HTTP server that speaks Daisy's plugin protocol and register the manifest URL. It shows up in the editor like any built-in connector. The SDK + examples are on GitHub.
Daisy ships as a Docker Compose stack. Postgres, Redis, the API, the worker, the editor — all bundled, all yours.
# 1. Grab the repo git clone https://github.com/daisy-workflow/daisy.git # 2. Bring up the stack cd daisy docker compose up -d --wait # 3. Open the editor open http://localhost:5173