Think about a typical Monday morning: you’re manually copying knowledge from a number of totally different programs to create a weekly report, then formatting it for various stakeholders. This single activity can eat a number of hours that may very well be spent on extra strategic work. Multiply this throughout your workforce, and these repetitive duties add up shortly.
Amazon Fast Flows automates these duties utilizing AI workflows. With Fast Flows, you create clever workflows utilizing pure language—no coding or machine studying (ML) experience required. You describe what you need automated, and Fast Flows builds it for you. This publish exhibits you construct your first AI-powered workflow, beginning with a monetary evaluation software and progressing to a sophisticated worker onboarding automation.
What’s Amazon Fast Flows?
Amazon Fast Flows is a part of Amazon Fast, a group of AI-powered options that work collectively that can assist you analyze knowledge, automate duties, and get insights via pure language conversations. This publish focuses particularly on Fast Flows for activity automation.
With Fast Flows, you flip your on a regular basis duties into automated workflows for particular person and workforce productiveness. You create, customise, and share purpose-built AI workflows utilizing your knowledge, insights, and actions out there inside Amazon Fast.
Stipulations
Earlier than constructing your first movement, guarantee that you’ve got an energetic AWS account with Amazon Fast enabled with permissions to entry Fast Flows. For setup directions, see the Amazon Fast Person Information.
Word: Amazon Fast makes use of generative AI. The particular outputs, responses, and generated content material may differ from the examples proven. These variations are regular and anticipated. Deal with understanding the ideas and advantages relatively than matching precise outputs.
Construct your first movement
On this instance, you construct a Monetary Efficiency Analyzer that gathers real-time market knowledge from the online, analyzes key metrics, and compiles knowledgeable abstract primarily based on the knowledge.
1. Navigate to Fast Flows
Open your browser, and log in to Fast. Then go to Fast Flows. The movement creation interface seems with a textual content space to explain your workflow and pattern prompts to get you began.
Determine 1: The Fast Flows creation interface the place you describe your workflow in pure language
2. Enter your immediate
Within the textual content space, enter the next immediate:
Create a movement that gathers complete firm monetary analysis by designing a software with 4 key parts: (1) Actual-Time Market Knowledge gathering present inventory costs and each day modifications, (2) Monetary Metrics Evaluation retrieving key ratios like P/E, market cap, and income, (3) Information Intelligence gathering current monetary headlines and market-moving occasions, and (4) Skilled Evaluation compiling analyst suggestions and scores, every triggered by an organization title or ticker image enter.
Fast Flows now is aware of precisely what you need: a workflow that takes an organization title as enter and returns a whole monetary image.
Fast Tip: You too can create flows instantly out of your conversations with chat brokers in Fast. In case you’re already discussing a activity or course of with an agent, you’ll be able to convert that dialog right into a movement with out ranging from scratch.
3. Generate your movement
Choose the Generate button. Fast Flows analyzes your immediate and maps your necessities to out there capabilities. It identifies that it should collect knowledge from net search, extract key monetary metrics utilizing AI evaluation, and format the outcomes right into a structured report. The AI works via your necessities, identifies the precise steps wanted, and assembles them right into a linked movement.
Determine 2: Fast Flows producing your workflow from the pure language immediate
4. Run your movement
After the technology completes, you’ll be able to take a look at your movement instantly. Enter an organization title or ticker image (like Amazon or AMZN) and choose Run. Watch because the movement executes every step in topological order, gathering market knowledge, analyzing metrics, gathering information, and compiling the ultimate evaluation.
After reviewing the outcomes, you’ll be able to chat instantly with the movement to refine the output. For instance, you may ask it to concentrate on particular metrics, regulate the evaluation depth, or format the outcomes in another way. The movement understands your requests and may modify its output accordingly.
5. Refine the movement within the editor
After your first run, go into the editor view to see how Fast Flows constructed your workflow from a single immediate. Every step is laid out visually, exhibiting the connections between them and the way knowledge strikes out of your enter via every element to the ultimate output.
Your movement begins with a textual content enter step the place you enter an organization title or ticker image.
Determine 3: The Circulate editor exhibiting every step within the workflow with the enter step chosen.
As a result of this movement works with market knowledge, Fast Flows robotically configured every analysis step with the built-in net search performance of Fast. It gathers real-time data from the online relatively than counting on static knowledge.
Determine 4: The Circulate editor exhibiting every step within the workflow with the online search step chosen.
Totally different net searches collect related knowledge that you just requested primarily based on the immediate for the step.
The ultimate step makes use of a normal information step to synthesize every little thing that the movement gathered. It takes the market knowledge, monetary metrics, information headlines, and analyst scores from earlier steps and compiles a monetary report.
Determine 5: The Circulate editor exhibiting every step within the workflow with the overall information step chosen.
From right here, you’ll be able to customise the movement to suit your wants. You may add a step to electronic mail the report back to your workforce, publish it to Slack, reserve it to a file in SharePoint, or export the output as a PDF or Microsoft Phrase doc. Given this data modifications over time, you may need to schedule your movement to run periodically.
Understanding key ideas
Now that you just’ve constructed your first movement, listed here are the constructing blocks that you just simply used. A step is a discrete element that performs a particular operate inside a movement. Steps are the basic models that you just chain collectively to create a whole movement.
Determine 6: Fast Flows affords 5 classes of steps—AI responses, movement logic, knowledge insights, actions, and person enter—that you just mix to construct workflows.
Fast Flows organizes steps into 5 classes:
- AI responses generate outputs, create pictures from textual content, immediate customized brokers, search the online, invoke Fast Analysis, and carry out duties on web sites.
- Circulate logic controls workflow execution with reasoning teams that outline circumstances, loops, or validations.
- Knowledge insights retrieve data out of your firm knowledge in areas and information bases or analytics from dashboards and subjects.
- Actions carry out learn or write operations in linked exterior programs and functions via pre-built or customized integrations.
- Person enter gathers data from customers via textual content fields or file uploads to provoke and supply context to your workflow.
Fast Flows works together with your firm knowledge wherever it lives. You may join spreadsheets and databases utilizing Amazon Fast, doc shops, resembling SharePoint, OneDrive, Google Drive, or Amazon Easy Storage Service (Amazon S3) via integrations, and take motion via objective constructed or customized integrations.
Past the fundamentals: Worker onboarding automation
Now that you just’ve constructed your first movement and perceive the basics, it’s time to sort out a extra complicated situation. This instance demonstrates superior capabilities like conditional logic and system integrations, exhibiting how the identical rules that you just simply discovered can automate multi-step enterprise processes. Think about you’re a human assets (HR) specialist with three new workers becoming a member of this week. For each, you should create worker data in your HR data system, write customized welcome emails that reference firm insurance policies, and coordinate with data expertise (IT) for badge creation, gear ordering, and electronic mail setup. These guide duties add as much as hours every week that take you away from strategic priorities. The next instance walks you thru constructing a movement that automates these onboarding duties.
Stipulations for the worker onboarding instance
This worker onboarding instance demonstrates superior Fast Flows capabilities. You may comply with alongside conceptually proper now to know the method with no setup required. For hands-on expertise with this particular instance, the next optionally available workshop setup directions helps you construct this in your individual AWS account with simulated HR and IT programs. It’s going to take roughly half-hour to finish.
If you wish to construct this movement in your individual AWS account, full the setup steps from the A Full Information to Amazon Fast workshop:
- First, deploy the workshop infrastructure by following the self-paced setup directions. This creates the foundational AWS assets and simulated HR and IT programs that your movement will combine with.
- Second, comply with the directions throughout the Flows module, together with the prerequisite directions.
After you full these setup steps, you’ll be able to construct and execute the worker onboarding movement. The workshop supplies pattern worker knowledge and step-by-step testing directions, so you’ll be able to see precisely how the automation works in follow.
Understanding the worker onboarding workflow
Determine 7: The worker onboarding workflow makes use of reasoning teams to create conditional logic, executing totally different steps primarily based on whether or not an worker already exists within the system
The workflow represents the sequence of steps from enter assortment and knowledge retrieval to executing actions. It begins by gathering worker data via textual content enter fields (first title, final title, electronic mail handle). Subsequent, the movement calls the workshop’s worker data API utilizing an motion step. This step determines if the worker’s electronic mail handle is registered within the system. A reasoning group on this movement works like an if-then assertion in on a regular basis logic. Consider it as a fork within the highway: if the worker already exists, the workflow ends; in the event that they don’t exist, the workflow continues and creates their data. On this movement, the reasoning group evaluates whether or not the e-mail for the brand new worker was discovered within the earlier step, creating two doable execution paths. If the worker exists, the workflow ends as a result of the document has already been created. If the worker doesn’t exist, the movement executes six sequential steps throughout the reasoning group: create the worker document, generate a welcome electronic mail, ship the welcome electronic mail, generate a badge request ticket, create the badge request ticket, and summarize the onboarding outcomes.
Creating the worker onboarding movement with pure language
Subsequent, create your movement utilizing a pure language immediate. You may describe what you need your movement to do in pure language, and Amazon Fast Flows interprets your description and robotically creates the workflow for you.
Navigate to the flows in Amazon Fast and choose Create Circulate. Within the movement creation interface, you see an choice to explain your movement in pure language. Enter a immediate just like the next instance:
Create an worker onboarding movement that collects new rent data, checks in the event that they exist already in our system, creates their worker document in the event that they’re new, generates a customized welcome electronic mail utilizing our firm insurance policies, sends the e-mail with their supervisor CC’d, creates IT tickets for badge and gear setup, and supplies a abstract of all actions accomplished.
Word that this immediate is a simplified model of the particular immediate used within the workshop. For the whole, detailed immediate and guide step-by-step directions, check with the workshop’s automating each day duties information, which supplies each the detailed pure language immediate and guide movement development steps.
The way to write efficient prompts for Fast Flows
Your immediate ought to embody 4 components: what data to gather (“collect worker particulars”), what choices to make (“test in the event that they exist already”), what actions to take (“create worker document”), and what content material to generate (“customized welcome electronic mail”). Trying on the instance immediate, you’ll be able to see how totally different phrases map to workflow parts. The phrase “Create an worker onboarding movement” states the aim. “Collects new rent data” indicators that the movement should collect knowledge from customers, which interprets to enter steps. While you describe operations like “creates their worker document,” “sends the e-mail,” and “creates IT tickets,” you’re describing actions that modify exterior programs. These change into motion steps that combine together with your Fast integrations. Phrases like “generates a customized welcome electronic mail utilizing our firm insurance policies” point out that the movement should produce content material by accessing organizational information. These change into output steps that hook up with areas, information bases, dashboards, or subjects. Branching logic emerges from language like “checks in the event that they exist already” and “in the event that they’re new.” These phrases counsel conditional logic, prompting Fast Flows so as to add reasoning teams that decide if steps ought to execute or what number of instances.
Understanding the constructing blocks
Your movement consists of 4 predominant teams of parts that work collectively to create the automation.
Enter parts are the place your movement gathers the knowledge that it must function. Within the worker onboarding instance, these are textual content enter steps that gather worker particulars like first title, final title, and electronic mail handle.
Logic parts management movement choices or branches. The reasoning group step makes use of pure language to outline when and the way different steps ought to run. Within the onboarding movement, the reasoning group comprises the instruction “Run if @E mail will not be present in @Examine Worker Exists outcomes.” This alleviates duplicate worker creation by controlling the logic of the workflow.
AI response and knowledge insights parts produce content material and responses utilizing your underlying knowledge. The movement contains steps that put together customized welcome emails by combining worker data saved inside an area.
Motion parts join your movement to exterior programs and carry out actions in actual time. Within the onboarding movement, these steps create worker data in HR programs, ship emails via your group’s electronic mail service, and generate tickets in IT programs. Every motion step specifies what system to hook up with, what operation to carry out, and the info required to name these programs.
Every step in your workflow creates a variable—a labeled container that holds data for different components of your movement. Variables are essential as a result of they allow you to move data between steps. For instance, when somebody enters “John Doe” within the First Identify discipline, that data turns into out there as “@First Identify” all through your whole workflow. With out variables, every step would function in isolation, unable to make use of knowledge from earlier steps. Variables function the connective tissue that lets data movement from one step to a different. Within the following picture from the “Create Worker File” step from the workflow created beforehand, discover the blue highlighted textual content within the immediate. These are variables referenced from different steps. The immediate is written as follows: “Create worker: first_name: @First Identify, last_name: @Final Identify, electronic mail: @E mail …”, the place the referenced variables are used as enter to the “CreateEmployee” API.
Determine 8: The “Create Worker File” step throughout the worker onboarding movement exhibiting how variables from earlier steps are referenced utilizing the @ image.
Fast suggestions when constructing your automations
Now that you’ve got a stable basis for Amazon Fast Flows, you’re prepared to start automating your duties. Listed here are some suggestions that can assist you get began when constructing flows.
Tip 1: Take a look at your prompts first. Earlier than constructing a movement, take a look at your immediate with the chat assistant. For instance, if you wish to automate report technology, first ask: “How do I extract knowledge from a dashboard and format it as a weekly report?” This helps you refine your method earlier than constructing the movement. The chat assistant has built-in information on the product itself. For instance, you’ll be able to ask “What’s Fast Flows and the way do I get began?”
Tip 2: Begin small together with your knowledge. Start with a smaller dataset. You can begin with an area with fewer assets or recordsdata. Fast has a restrict to the dimensions of information that it will possibly match inside its context window. For instance, think about you’ve got an inventory of emails and you should reply to every one. You may initially need to function over all emails directly, however if you happen to use the reasoning group step looping performance, you’ll be able to function over one electronic mail at a time. This breaks down the issue into smaller items. Consult with the documentation for flows limits.
Tip 3: Write an efficient one-shot immediate. Flows is an orchestration service that makes use of the underlying capabilities of Fast. Writing an efficient one-shot immediate will assist help the success of your use case. Conversely, you’ll be able to ask Fast to enhance your immediate. To be taught extra on prompting, learn via the prompting finest practices module within the Amazon Fast workshop.
Tip 4: Ask about motion integrations. When working with motion integrations, you don’t must be an professional to know the APIs or work with them in pure language. For instance, if you happen to’re utilizing the Microsoft Outlook integration, you’ll be able to ask the chat assistant work with it. Your immediate may appear like the next: “What motion do I exploit to ship an electronic mail for the Microsoft Outlook connector and what parameters does that motion have?” This may define the suitable motion and the parameters (enter) used for that connector.
Tip 5: Map out your workflow. Perceive the duty that you just’re attempting to automate, the info concerned, what actions have to be executed, and in what order. Strive writing down or drawing every step of your workflow.
Tip 6: Get assist from a chat agent. You may ask a chat agent to assist design and construct your flows. An instance will be discovered throughout the Amazon Fast workshop.
Clear up assets
To keep away from incurring fees for assets you’re now not utilizing, full the next cleanup steps. Amazon Fast fees are primarily based on utilization, so eradicating unused flows and scheduled runs helps hold prices down.
Delete the flows that you just created for testing by navigating to the flows checklist and choosing Delete. In case you accomplished the workshop, comply with the cleanup directions within the workshop information to take away the deployed infrastructure. Cancel scheduled movement runs by modifying the movement and eradicating the schedule.
Conclusion
On this publish, you constructed two AI-powered workflows utilizing Amazon Fast Flows. You additionally discovered use pure language prompts to generate workflows, how Fast Flows constructing blocks work, and customise flows to your particular wants. With Amazon Fast Flows, you flip repetitive work into clever, AI-powered automations utilizing pure language. You mix knowledge assortment, content material technology, and system integration to cut back time spent on guide duties—no coding required. Listed here are your subsequent steps:
- Log in to Amazon Fast and create your first movement utilizing one of many pattern prompts.
- Full the Amazon Fast workshop to get fingers on expertise.
- Establish a repetitive activity in your each day work and describe it in pure language. Fast Flows can doubtless automate it.
To be taught extra, be a part of the Amazon Fast Neighborhood to ask questions and discover studying assets or occasions.
Concerning the authors
Jed Lechner
Jed Lechner is a Sr. Specialist Options Architect at Amazon Net Providers specializing in Agentic AI options. Previous to his present position, he labored as a Software program Engineer for AWS and different corporations, specializing in vitality and sustainability.
Josh Demuth
Josh Demuth is a GenAI Options Architect with 20 years within the tech business, with a number of years specializing in programs integration. He thrives on creating options that make disparate programs work collectively and discovering modern approaches to enterprise issues. The fast evolution of AI and automation has him excited in regards to the transformative options on the horizon.
