From Chat to Workflow: How Quickly Turns Conversations Into Automations

The Playbook Problem

Every team has processes that live in people's heads. The way you triage an incident. The steps to onboard a new vendor. The checklist for closing out a sprint. These processes get repeated dozens of times a week, and every time, someone has to remember the steps, open the right tools, and do the work manually.

Traditional automation platforms try to solve this by giving you a workflow builder. Open the canvas, drag some nodes around, connect them, configure each one, test it, debug it, deploy it. By the time you have something working, you have spent more time building the automation than you would have spent doing the task manually for the next six months.

We wanted a different path. What if the automation could emerge from the way you already work?

The Journey: From Ad-Hoc to Scheduled

Here is how a typical Quickly user goes from zero to a fully automated workflow, without ever opening a workflow builder.

Week 1: Ad-hoc requests. A property manager types @Quickly create a maintenance ticket from this thread and assign it to the building super in Slack. Quickly reads the thread, creates the Jira issue, and assigns it. Done in 10 seconds. The manager does this a few times a week. Week 2: Pattern recognition. Quickly notices the manager is doing the same thing repeatedly: reading a tenant message, creating a ticket with specific fields, assigning it based on the building address, and notifying the super. Quickly suggests: "Want me to save this as a playbook?" Week 3: Saved playbook. Now the manager types @Quickly run maintenance triage and Quickly walks through a short conversation: "What's the building? What's the issue? Who should handle it?" Then it executes all the steps automatically. The conversational playbook collects information through natural dialogue, extracts structured data, asks for confirmation, and executes. Week 4: Scheduled automation. The manager sets up a trigger: whenever a message is posted in #maintenance-requests, run the maintenance triage playbook automatically. No one needs to type anything. The playbook reads the message, determines the building, creates the ticket, assigns it, and notifies the super. What started as a chat message is now a fully autonomous workflow.

What Makes a Playbook

A playbook in Quickly is a reusable sequence of steps that can be triggered by conversation, schedule, or event. Under the hood, every playbook follows a five-phase state machine:

COLLECT - Gather the information needed to complete the task. In a conversational playbook, this means asking the user questions. In an automated playbook, this means reading from the trigger event or connected data sources. EXTRACT - Parse the collected information into structured fields. If a tenant writes "the kitchen faucet is leaking in unit 302 at 45 Elm Street," the extract phase pulls out the issue type (plumbing), the unit (302), and the building address (45 Elm Street). REVIEW - Show the user what is about to happen and ask for confirmation. This is the human-in-the-loop gate. In Conservative mode, every playbook pauses here. In Autonomous mode, only high-risk actions pause. EXECUTE - Perform the actual work: create the ticket, send the notification, update the spreadsheet, fill the form. This is where Quickly calls your connected tools. COMPLETE - Report the results back to the user or the trigger channel. Include links to everything that was created.

The beauty of this model is that it works identically whether the playbook is driven by a conversation or by a scheduled trigger. The phases are the same. The tools are the same. The safety gates are the same.

Visual Workflows

For teams that want more control over branching logic, we also built a visual workflow builder with 8 node types:

• Trigger - What starts the workflow (message, schedule, webhook, manual)
• Action - A tool call like "create Jira issue" or "send Slack message"
• Condition - Branch based on data: if priority is P1, take this path; otherwise, take that one
• Loop - Repeat a set of steps for each item in a list
• Delay - Wait for a specified duration before continuing
• Human Input - Pause and ask someone for a decision or additional information
• Conversation - Hand off to the AI for a natural language interaction within the workflow
• End - Mark the workflow as complete

You can build visual workflows from scratch on the canvas, or you can convert a conversational playbook into a visual one. The system generates the flow graph automatically from the playbook's phase structure. Going the other direction works too: if you have a visual workflow and want a simpler conversational interface, the converter handles that.

A Real Example: Maintenance Triage

Here is how a property management company uses Quickly to handle maintenance requests end-to-end.

A tenant sends a message to the building's shared Slack channel: "Hi, the hot water in unit 508 isn't working. It's been cold since this morning."

The maintenance triage playbook triggers automatically. It reads the message and runs through the phases:

COLLECT: The playbook reads the tenant message and pulls in the building's metadata from the knowledge base (building address, super contact, property manager).

EXTRACT: Quickly identifies the issue as "hot water not working," the unit as 508, and the urgency as moderate (not an emergency like a burst pipe, but uncomfortable enough to need same-day attention).

REVIEW: In Balanced mode, the playbook shows the property manager what it is about to do: "Create a P2 maintenance ticket for unit 508 (no hot water), assign to Mike (building superintendent), and notify the tenant that help is on the way. Proceed?" The manager clicks Approve.

EXECUTE: Quickly creates the Jira ticket with all details populated, sends a Slack DM to Mike with the ticket link and unit number, and replies to the tenant's message confirming that a work order has been created.

COMPLETE: The playbook logs the action, updates the ROI dashboard (estimated 12 minutes saved), and closes out.

Total elapsed time: about 30 seconds from tenant message to work order, most of which was waiting for the manager's approval.

Why This Matters

The gap between "I wish this was automated" and "I have a working automation" is where most teams give up. Traditional workflow builders require you to think in terms of nodes and connections before you even know what the process looks like. Quickly lets you start with the process you already have and gradually make it more automated.

You do not need to be a developer. You do not need to understand APIs or webhooks. You start by talking to Quickly in Slack, and over time, those conversations become playbooks that run themselves.

That is the path from chat to workflow. And it starts with a single message.