People & HR

How to onboard a new hire with AI

People & HR3 AI tools7 steps6 friction points

Want an agent to handle this on your live data? See the Starch version →

Onboarding a new hire is one of those workflows that looks simple on paper and sprawls in practice. You're coordinating equipment requests, first-week schedules, account provisioning, introduction emails, documentation access, and role-specific training — often across six or seven different tools — all while running the rest of the business. For small teams, the founder or a single ops person usually owns every piece of it.

The workflow feels like a natural fit for AI because so much of it is templated, repetitive, and text-heavy. Welcome emails, day-one checklists, role-specific training paths, FAQ documents — these are exactly the kinds of structured outputs that a language model can draft faster than any human. The appeal is real: paste in a job title and a few context notes, get a first draft of a 30-60-90 plan in 30 seconds.

ChatGPT, Claude, and Gemini can all do meaningful work here today. They'll draft onboarding checklists tailored to a specific role, write welcome emails with the right tone, turn a messy Notion doc into a structured FAQ, generate first-week meeting agendas, and outline a 30-60-90 plan from a brief description of the role. For one-off drafts, any of these tools get you 80% of the way there fast.

Skip the manual work → See the Starch version

People & HR3 AI tools7 steps6 friction points

AI walkthrough

How to do it with AI today

A practical walkthrough using ChatGPT, Claude, and other off-the-shelf LLMs — what they're good at, what you'll have to do by hand.

Tools that work for this

ChatGPTClaudeGemini

Step-by-step

1 Open Claude or ChatGPT and paste in the job title, team structure, and a few bullet points about the role's first-month priorities. Ask it to generate a day-one and week-one checklist with specific tasks, owners, and deadlines.

2 Copy your existing welcome email (or start from scratch) and prompt the model to rewrite it for this hire's specific role, seniority level, and start date. Include details about the team size and reporting structure so the output doesn't read as generic.

3 Paste in your company's existing documentation — handbook sections, tool access notes, process docs — and ask Claude to identify gaps, flag outdated references, and produce a structured onboarding FAQ the new hire can actually use.

4 Prompt ChatGPT or Gemini to build a 30-60-90 plan for the role. Provide the job description, the team's current OKRs, and any context about what 'good' looks like at 90 days. Review and edit the output before sharing.

5 Use the model to draft introduction emails for each internal stakeholder the new hire should meet in week one. Give it each person's name, role, and one sentence on why the meeting matters — the model handles the prose.

6 Ask the model to generate a scheduling sequence: which meetings should happen in what order during the first two weeks, how long each should be, and what the agenda should cover. Export this as a plain text list you can manually build into calendar invites.

7 Run a final pass: paste all your drafted onboarding materials into one session and ask Claude to identify anything contradictory, missing, or unclear from a new hire's perspective. Use the output as a checklist before the hire's start date.

Prompts you can copy

You're onboarding a new operations manager at a 12-person startup. Generate a detailed day-one and week-one checklist with task owner, estimated time, and priority for each item.

Write a warm, specific welcome email from a founder to a new customer success hire starting Monday. Include what to expect on day one, who to contact with questions, and tone that matches a direct but friendly voice.

Here are three internal process docs [paste text]. Identify what's missing, what's outdated, and produce a structured onboarding FAQ a new hire could use to get up to speed without asking the founder.

Create a 30-60-90 plan for a head of growth joining an early-stage B2B SaaS company. Current ARR is $800K. The role owns demand gen, content, and partnerships. Day 90 success means one repeatable acquisition channel identified.

Draft five short introduction emails for a new engineering hire to send to: the CEO, the product manager, the head of design, the customer success lead, and the data analyst. Each email should explain who they are and suggest a 20-minute intro call.

Reality check

Where this gets hard

The walkthrough above works — until your numbers change, the LLM hallucinates, or you have to re-paste everything next month.

Nothing connects to your actual systems — you're manually copying employee data, tool lists, and calendar availability into the prompt every single time.

Outputs don't persist anywhere useful. The checklist you generated lives in a chat window, not in a shareable, updatable format your new hire can actually track against.

Context resets with every session. The tone, structure, and role-specific details you carefully built into last month's onboarding run don't carry forward to the next hire.

Scheduling coordination still happens manually. The model can draft a first-week meeting sequence, but you're the one creating each calendar invite and chasing confirmations.

Documentation quality depends entirely on what you paste in. If your existing docs are incomplete or outdated, the model confidently generates polished onboarding materials built on bad inputs.

No alerts or follow-through. Once you close the session, there's no mechanism to remind you that the new hire's account access still hasn't been set up or that their first check-in meeting was never booked.

Tired of the friction?

Starch runs the whole workflow on live data — no copy-paste, no hallucinated numbers, no re-prompting next month.

See the Starch version →

Starch alternative

The same workflow on Starch

Starch is an agentic operating system — for onboarding, that means an agent builds the persistent apps, checklists, and automations that run your onboarding process against your live calendar, email, and documentation data, instead of one-off prompts you re-run from scratch each time.

Use the Knowledge Management starter app to build a searchable team wiki. An agent auto-organizes your existing docs, detects when content goes stale, and generates structured onboarding paths for each role — so the new hire has a living resource, not a static PDF.

Connect Google Calendar through Starch's scheduled sync and tell the agent: 'Build a first-week meeting sequence for each new hire, create calendar events for each intro meeting, and Slack me when any are unconfirmed 48 hours before their start date.' That automation runs on its own.

Starch connects to Gmail on a scheduled sync. Describe what you want: 'Draft a personalized welcome email for each new hire, pull their role from our task tracker, and queue it for my review three days before start.' The agent builds that workflow once; it runs for every hire after.

Use the Task Manager app to generate a role-specific onboarding checklist for each new hire — with P1–P4 priorities and due dates — by describing the role in plain English. Tasks stay tracked in one place instead of disappearing into a chat thread.

Connect your HR system (BambooHR, Rippling, Gusto, and others are reachable from Starch's integration catalog) so the agent pulls actual employee data — start dates, roles, reporting lines — instead of relying on whatever you remember to paste in.

Describe the full onboarding system you want in one prompt — 'Build me an onboarding tracker that creates a checklist per hire, schedules intro meetings, tracks documentation access, and alerts me to anything overdue' — and Starch builds it. You don't rebuild it for the next hire.

Get closed-beta access →

Toolkit