From Idea to Client Deliverable in 15 Minutes
Follow along as we build a real workflow end-to-end. By the end, you'll have a reusable mission that produces client-ready output — and the skills to build dozens more.
What You'll Learn
- Plan a workflow on paper before building it in FlowManner
- Configure a 3-stage agent pipeline (gather → analyze → write)
- Run a mission and interpret the live execution stream
- Iterate on agent prompts based on real output quality
- Export a polished deliverable and schedule recurring runs
This hands-on tutorial takes you from zero to a working client deliverable in 15 minutes. You'll plan a workflow, configure agents, run the mission, and export a polished result — learning the full FlowManner lifecycle in one sitting.
Prerequisites
- FlowManner account with BYOK configured (see Setting Up BYOK)
- A specific deliverable in mind (report, analysis, summary, etc.)
- 20 minutes of uninterrupted time
What We're Building
Imagine a freelance consultant who needs to deliver a weekly competitive analysis report for a client. The report pulls data from multiple sources, analyzes trends, and produces a formatted document with actionable recommendations. Today, you'll build exactly that.
The FlowManner Workflow Pattern
Every FlowManner workflow follows the same pattern: Input → Process → Output. You define what data flows in, which agents handle each processing stage, and what format the final deliverable takes. The Visual Builder makes this tangible — you literally draw the pipeline.
From Tutorial to Production
Once you've built this workflow, you can schedule it to run automatically, share it with your team, or publish it to the marketplace. The skills you learn here transfer directly to any automation task — the pattern is always the same.
Exporting Your Deliverable
FlowManner supports multiple export formats: Markdown for internal docs, PDF for client-facing deliverables, JSON for API integrations, and CSV for data-heavy outputs. Choose the format that matches your client's expectations and your delivery channel.
Sketch first, build second
Two minutes of planning saves twenty minutes of rework. Draw your pipeline on paper or a whiteboard before touching FlowManner.
Test with real data early
Placeholder data hides quality issues. Use a small sample of real data from the start — even 2-3 items reveal problems that synthetic data misses.
Most workflows need 2-3 iterations
Don't expect perfect output on the first run. Adjust prompts, swap models, or add a review step. Each iteration typically improves quality significantly.
Schedule once it's reliable
Only schedule a workflow after you've seen consistent quality across 3+ manual runs. Scheduled runs have no human in the loop — quality must be proven first.
Step-by-Step
Sketch the Workflow
Before touching FlowManner, spend two minutes sketching your pipeline on paper. For our example: (1) gather competitor URLs, (2) scrape and summarize each, (3) analyze trends, (4) write the report, (5) format as PDF. This blueprint guides everything that follows.
Create the Mission
Go to Missions → New Mission. Name it 'Weekly Competitive Analysis'. This creates a blank canvas in the Visual Builder where you'll assemble your pipeline.
Configure the Data Gathering Agent
Add an Agent node and configure it as your data gatherer. Set the model to GPT-4o (good at structured extraction) and write a system prompt: 'You are a competitive intelligence researcher. Given a list of competitor URLs, extract the latest product updates, pricing changes, and key announcements.'
Add the Analysis Agent
Connect a second Agent node to the first. This agent receives the gathered data and identifies patterns. System prompt: 'You are a strategic analyst. Given competitive intelligence data, identify 3-5 key trends, rate their significance (high/medium/low), and explain the business impact.'
Add the Report Writer
Connect a third Agent node. This takes the analysis and produces a formatted report. System prompt: 'You are a business report writer. Given a competitive analysis, produce a polished markdown report with: Executive Summary, Key Findings, Trend Analysis, and Recommendations sections.'
Run with Test Data
Click Run Mission and provide sample input — a few competitor URLs or placeholder data. Watch each agent execute in the live stream. Verify that data flows correctly from one stage to the next.
Review and Iterate
Examine the final output. Is the report detailed enough? Are the recommendations actionable? Adjust the system prompts based on what you see. Most workflows need 2-3 iterations before the output is client-ready.
Export and Deliver
Once satisfied, use the Export button to download the report as Markdown or PDF. Schedule the mission to run weekly using a cron trigger, and your client deliverable is now fully automated.
Related guides
Ready to start building?
Practical tutorials for getting the most out of FlowManner — from your first workflow to advanced multi-agent orchestration.
Start building — it's free