Construct AI-powered worker onboarding brokers with Amazon Fast

Enterprises usually battle to onboard new group members at scale. Human assets (HR) groups spend time on guide duties that delay productiveness, corresponding to processing paperwork to answering repeated questions on advantages and insurance policies. For organizations with many new hires, these steps make it tougher to maintain onboarding constant and compliant. Organizations lose substantial quantities of time per day per new rent throughout onboarding, with new workers usually reaching solely a fraction of their potential productiveness within the first month. Amazon Fast is a totally managed agentic service. With it, HR departments can create no-code onboarding brokers that reply new-hire questions, observe compliance throughout current instruments, and clear tickets robotically in order that new hires can ramp quicker with much less guide work.

On this submit, we stroll by constructing a customized HR onboarding agent with Fast. We present the best way to configure an agent that understands your group’s processes, connects to your HR methods, and automates widespread duties, corresponding to answering new-hire questions and monitoring doc completion. You possibly can adapt this resolution to your onboarding workflow so new hires get constant solutions and HR groups reclaim time beforehand spent on routine inquiries.

Key parts of Amazon Fast

Fast transforms worker onboarding from scattered paperwork and guide processes into an clever, linked expertise by the next built-in parts:

• Data bases – Listed content material from exterior sources like SharePoint, OneDrive, and Confluence, in addition to inner content material together with inner web sites, file uploads, and Amazon Easy Storage Service (Amazon S3) buckets. A data base serves as a single searchable repository, so new hires get complete solutions from a number of sources as an alternative of looking by disconnected information.
• Actions (motion connectors) – Safe, permission-aware integrations that allow AI brokers to take actual motion in HR onboarding situations—creating ServiceNow IT gear requests, sending Slack welcome messages to group channels, or updating onboarding workflows in undertaking administration instruments—somewhat than simply offering hyperlinks to varieties.
• Areas – Centered environments that set up team-centered property together with information, enterprise intelligence artifacts (corresponding to dashboards and subjects), data bases, and actions with sharing controls for group collaboration.

Fast will help HR groups create specialised onboarding assistants that mix data entry with automated duties. You should use the built-in system agent (“My assistant”) for quick assist or create customized chat brokers tailor-made to your group’s particular onboarding wants, corresponding to a devoted HR onboarding assistant that is aware of your organization insurance policies and may robotically deal with widespread requests like IT setup or advantages enrollment.

Resolution overview

This resolution makes use of a customized chat agent in Fast for worker onboarding. With out an agent, HR may swap between wikis, SharePoint, ticketing, chat, and e-mail to coordinate every step. With Fast, the agent presents the most recent guidelines from the HR house, solutions with permitted language, opens requests by actions, notifies stakeholders, and factors the worker to the following step. Confirmations and standing stay within the HR instruments, and the agent reads or updates them by actions or flows. The next diagram illustrates the answer structure.

Implementing the answer consists of the next high-level steps:

1. Create the chat agent in Fast.
2. Connect the HR house and hyperlink data sources.
3. Add actions.
4. Take a look at with actual questions and duties, then share with workers.

Fast gives two forms of chat brokers that facilitate this onboarding resolution: the system chat agent (“My assistant”) and customized chat brokers. The system chat agent (“My assistant”) – “My assistant” seems on the Amazon Fast console by default and helps customers ask questions and full duties utilizing assets they’re allowed to entry. Customers can work together with the system agent in a number of methods:

• Ask normal questions utilizing the agent’s built-in data by selecting Normal data.
• Add their very own information immediately in chat (as much as 20 information per dialog) for evaluation and questions.
• Management the dialog scope by selecting from three modes: All information & apps (searches throughout all accessible assets), Normal data (makes use of solely built-in data), or Particular information & apps (targets explicit areas, dashboards, subjects, data bases, or actions). For instance, a consumer may add their worker handbook and ask, “What’s our distant work coverage?” or choose the HR house and ask, “How do I enroll within the medical health insurance plan?” The system agent is offered instantly with no configuration required and adapts its responses primarily based on the chosen scope and out there assets.

Customized brokers assist you to construct specialised assistants for your small business wants. You configure habits (objective, tone, response format); connect areas with dashboards, subjects, and data bases for grounded solutions; and hyperlink motion connectors so the agent can carry out duties in instruments like Jira, Slack, ServiceNow, Salesforce, Outlook, or Groups. You possibly can share customized brokers with particular customers or teams. Customized brokers provide the next capabilities:

• Use case-specific responses – Outline the agent’s persona and response model tailor-made to particular enterprise workflows and necessities.
• Steerage by reference paperwork – Add particular paperwork that function response templates for constant messaging and course of guides for following particular steps.
• Complete information integration – Hyperlink areas to the agent to provide it entry to several types of searchable content material and data sources, together with dashboards for analytics, subjects for structured datasets, data bases for exterior, unstructured doc repositories, and native information uploaded on to the house for added data. This helps the agent reply questions utilizing totally different related information inside the group’s permission construction.
• Automated actions – Add motion connectors so customers can create Jira tickets, ship Slack messages, replace Salesforce, or open ServiceNow requests immediately from chat.
• Collaboration – Take a look at, refine, and share brokers with teammates. Directors can management who can create and customise brokers by consumer subscriptions and customized permissions.

You should use the system chat agent for normal help throughout Fast, or create a customized agent tailor-made to a workflow corresponding to HR onboarding. In that case, you outline directions, connect the HR house or data base, and allow actions for requests and notifications.

Within the following sections, we stroll by the steps to implement this resolution utilizing two personas: the HR administrator who units up and shares the agent, and the worker who completes onboarding duties with the agent.

Stipulations

Earlier than you start, be sure to have accomplished the next steps:

1. Create an AWS account. For extra data, see Create an AWS account.
2. Affirm you’ve entry to Fast.
3. No less than one Amazon Fast Enterprise subscription to configure actions and create data bases. Customers who solely use the shared agent may be on the Amazon Fast Skilled subscription
4. Go to Get began with Atlassian Cloud and create a free website, choosing each Confluence and Jira on the Free plan (as much as 10 customers).
   1. In Confluence, create an “HR Onboarding” house to retailer your HR content material.
   2. In Jira, create a easy HR onboarding undertaking that the agent can use for entry or gear requests within the Add actions part.
5. Obtain the ZIP file from the HR onboarding workshop supplies web page.
6. From the HR paperwork folder within the ZIP file, add the next information into your HR Onboarding Confluence house:
   1. employee_handbook.pdf
   2. leave_policy.pdf
   3. onboarding_checklist.pdf
   4. performance_review_guidelines.pdf
   5. public_holidays.csv (elective, used later for reporting or analytics)

In case your group already makes use of a company Confluence website, you won’t have permission to create areas or add pattern information until you request further entry out of your Confluence administrator. To expertise the worth of Fast with out ready on admin modifications, use a separate Atlassian Cloud website to comply with this submit.

Implementation Steps

This process makes use of two personas: the HR administrator who units up and shares the agent, and the worker who completes onboarding duties with the agent.

HR administrator

The next sequence diagram exhibits how the HR administrator creates, configures, and shares the HR onboarding agent in Fast.

Create chat agent

First, you create the chat agent itself, which turns into the only place the place new hires ask questions and get guided by onboarding:

1. On the Fast console, select Chat brokers within the navigation pane, then select Create.
2. Enter a easy pure language immediate describing what you need your agent to do (for instance, “Assist new workers with HR onboarding questions and gear requests”).

Fast will robotically broaden your immediate into an in depth persona and response directions and scan your out there assets to hyperlink related areas and motion connectors to the agent.

3. Evaluate the generated agent configuration and refine as wanted, updating the preview to save lots of your variations inside the session.
4. Select Launch chat agent if you find yourself glad.

Configure habits

Subsequent, you form how the agent ought to reply so its tone, scope, and guardrails match your HR insurance policies and HR model:

• Agent metadata – Replace the agent’s title, description, welcome message, and starter prompts to assist customers uncover and use the chat agent correctly. These parts function the primary impression and information customers on the best way to work together successfully together with your HR assistant.
• Agent directions – Evaluate and replace the robotically generated persona directions, response format, tone, and size settings from the earlier step. The system-generated inputs present a strong basis, however you possibly can fine-tune to match your group’s particular HR communication model and necessities.
• Reference paperwork – Add particular steering paperwork that present the best precedence directions for agent habits. These reference paperwork shall be adopted as prescribed whereas you should utilize the instruction fields to supply high-level steering on habits and targets.

Join HR data

Now you join your HR data sources so the agent solutions from permitted handbooks and insurance policies as an alternative of inventing its personal language:

1. Create or select an current HR house that holds handbooks, insurance policies, and checklists. By configuring the agent’s data scope to focus particularly on HR-related content material, you ensure responses keep inside applicable boundaries and don’t entry unrelated organizational information.
2. Select Add information to add information to the house, together with:
   1. Worker handbooks and coverage paperwork
   2. Advantages data and FAQ paperwork
   3. Coaching supplies and guides
3. Hyperlink data sources corresponding to SharePoint or a wiki.
4. Hyperlink the configured house to your agent so it could possibly entry this permitted searchable content material for grounded responses.

Add actions

After the agent can reply questions, you add actions so it could possibly additionally set off work in your HR instruments, corresponding to tickets, requests, and notifications:

1. Open the Actions card and select Hyperlink actions.
2. Choose from out there motion connectors that you’ve already configured. For the HR onboarding use case, this might embody instruments corresponding to Jira (to create and replace tickets), ServiceNow (to handle incidents), or Microsoft Outlook (to ship emails).

Solely motion connectors configured with the required OAuth particulars may be linked to the agent, so end-users can authenticate individually throughout their chat. Replace your reference paperwork and persona directions to specify when to invoke particular motion connectors. For instance: “When an worker requests gear, use the ServiceNow connector to create a {hardware} request ticket,” or “For entry requests, create a Jira ticket within the IT-Entry undertaking with precedence set to ‘Regular.’”

Customise, check, and share

Lastly, customise the agent with a welcome message and instructed prompts. You possibly can check the agent with life like situations, tune the expertise, and share it with a pilot group so HR can validate the workflow earlier than broad rollout. Take a look at with actual questions and duties utilizing the preview chat.

While you’re prepared, launch the agent, and it will likely be out there in your private library for personal use. To share with others, select Share and add customers and consumer teams as viewers to make use of the agent. You too can choose different customers out of your group to be homeowners to edit and check the agent together with you. HR managers can share the customized agent with new workers through the use of the sharing choices within the navigation pane to grant entry to particular group members or teams.

Worker

The next sequence diagram exhibits how an worker makes use of the onboarding agent to finish required duties and observe their Day 1 progress in a single place.

Use the onboarding agent

After the agent is revealed and shared with workers as viewers, they’ll open it from the hyperlink HR gives (for instance, of their Day 1 e-mail or HR portal) or from the chat brokers listing in Fast, after which use it as follows:

1. The worker opens the shared HR onboarding agent from the hyperlink or from the chat brokers listing and begins a brand new Day 1 dialog.
2. The agent exhibits the most recent onboarding guidelines from the HR Onboarding house and gives hyperlinks to required varieties, coaching, and inner pages so the worker can transfer by the steps so as.
3. The worker asks coverage or advantages questions in plain language, and the agent solutions utilizing content material from the HR Onboarding house and linked HR data sources so responses match HR-approved language.
4. On this setup, when the worker requests gear or utility entry, the agent makes use of a Jira motion connector to create a problem within the HR onboarding undertaking and returns the difficulty key and hyperlink so you possibly can see the request finish to finish with out touching manufacturing HR methods.
5. For delicate steps corresponding to I-9 verification, tax varieties, or direct deposit, the agent directs the worker to the suitable HR system or safe portal as an alternative of amassing paperwork in chat so delicate information stays in the correct place.

As an worker, the expertise is easy: they open a single chat, see their Day 1 guidelines, ask questions in pure language, and let the agent open requests and level them to the correct methods. As a substitute of juggling emails, portals, and tickets, onboarding looks like a guided dialog the place every subsequent step is evident.

You’ve gotten now arrange the HR Onboarding Confluence house with pattern HR paperwork, created a customized onboarding agent in Fast, configured its habits, linked HR data, and added Jira actions for requests. You should use this setup as a proof of idea with a small group of recent hires or HR companions, then lengthen it by including extra content material, further actions, or new areas for different HR workflows corresponding to efficiency critiques or coverage updates.

Guardrails and security

Fast consists of built-in security and content material controls for chat brokers, so you possibly can comply with together with this submit utilizing the default settings in your account. If you wish to experiment with coverage controls as a part of this proof of idea, you too can add a small listing of blocked phrases or phrases so the agent avoids particular phrases in HR responses (for instance, casual slang or discouraged wording). Blocked phrases are configured on the Fast console and utilized throughout brokers in your account. For step-by-step directions and extra safety choices corresponding to entry management and encryption, see the Amazon Fast Consumer Information.

Fast tiers

Fast affords two consumer subscriptions: Skilled and Enterprise. Skilled helps on a regular basis use of chat brokers and areas, operating Amazon Fast Flows and Amazon Fast Analysis, and viewing Amazon Fast Sight dashboards, with the power to create and share customized brokers and areas. Enterprise consists of every little thing in Skilled plus superior authoring options corresponding to configuring actions, creating data bases, constructing automations in Amazon Fast Automate, and authoring dashboards in Fast Sight, with bigger month-to-month utilization allowances. A 30‑day free trial is offered for as much as 25 customers per account. For particulars, discuss with Amazon Fast pricing.

Conclusion

This submit confirmed the best way to construct an HR onboarding chat agent in Fast, connect HR content material, add actions and elective flows, and share it with workers. Begin with a pilot that covers your most frequent questions and two or three requests, assessment utilization, and refine the agent’s directions and content material. For subsequent steps, broaden the HR house, add further actions as wanted, and assessment the Fast documentation for superior configuration. Past onboarding, HR groups can discover constructing brokers for worker self-service, efficiency administration, expertise acquisition, studying and improvement, analytics, and off-boarding processes to rework their complete HR operations.

Prepared to rework your office productiveness? Get began with Fast, discover pricing choices that suit your wants. Click on right here to start constructing your individual HR agent, discover our official documentation for detailed implementation steering, or contact your AWS account group to debate how Fast can remodel your group’s strategy to data-driven decision-making.

In regards to the authors

Pegah Ojaghi

Pegah Ojaghi is a Generative AI Utilized Architect at AWS with a PhD in Pc Science centered on giant language fashions, generative AI, and reinforcement studying. Her experience and analysis span basis mannequin improvement, RLHF strategies, and novel optimization strategies for LLMs. Her ardour is translating cutting-edge analysis into manufacturing methods throughout healthcare, monetary providers, and insurance coverage industries.

Chinmayee Rane

Chinmayee Rane is a Generative AI Specialist Options Architect at AWS, with a core concentrate on generative AI. She helps ISVs speed up the adoption of generative AI by designing scalable and impactful options. With a robust background in utilized arithmetic and machine studying, she makes a speciality of clever doc processing and AI-driven innovation. Outdoors of labor, she enjoys salsa and bachata dancing.

Ebbey Thomas

Ebbey Thomas is a Senior Generative AI Specialist Options Architect at AWS. He holds a BS in Pc Engineering and an MS in Info Programs from Syracuse College. Outdoors of labor, he enjoys espresso, the outside, exercises, street journeys, and time together with his household.

Sonali Sahu

Sonali Sahu is main the Generative AI Specialist Options Structure group in AWS. She is an writer, thought chief, and passionate technologist. Her core space of focus is AI and ML, and he or she regularly speaks at AI and ML conferences and meetups all over the world. She has each breadth and depth of expertise in expertise and the expertise trade, with trade experience in healthcare, the monetary sector, and insurance coverage.