ethos
← Back to Projects
🚀

Onboard

Personalized employee onboarding plans

Thesis

Employee onboarding is inconsistent. AI can personalize plans based on role while ensuring nothing is missed.

The Problem

Employee onboarding is chaos: HR has a checklist. The manager has different expectations. The new hire doesn't know who to ask what. Some people get great onboarding, some get thrown in. It takes months to feel productive. The information exists. The personalization doesn't. Everyone gets the same generic checklist regardless of role.

Implementation Approaches

Approach 1

Role-Based Plan Generator (Recommended)

Recommended

Generate personalized onboarding from role + company materials

⏱️2 weeks
📊Medium complexity

Implementation

  • Input: job description, team info, company docs
  • AI generates: week-by-week onboarding plan
  • Mix of: company-wide + team-specific + role-specific
  • Track progress, surface blockers
  • Adapt plan based on new hire feedback

Pros

  • +Personalized without manual work
  • +Consistent baseline with role customization
  • +Measurable: track onboarding metrics
  • +Flywheel: learns what makes good onboarding

Cons

  • Quality depends on source materials
  • Role-specific content needs to exist
  • May miss company culture nuances
Approach 2

Buddy + AI Assist

AI supports assigned onboarding buddy

⏱️1 week
📊Low complexity

Implementation

  • Buddy assigned as usual
  • AI provides: suggested topics, reminders, check-in prompts
  • New hire can ask AI questions first
  • Escalate to buddy when AI can't help

Pros

  • +Keeps human connection central
  • +Reduces buddy burden
  • +AI handles repetitive questions
  • +Works alongside existing programs

Cons

  • Still depends on buddy quality
  • Doesn't fix broken programs
  • Limited impact compared to full solution
Approach 3

Self-Service Knowledge Base

Searchable onboarding knowledge base with AI chat

⏱️1 week
📊Low complexity

Implementation

  • Centralize all onboarding materials
  • AI chat for questions
  • Suggested reading paths by role
  • Track: what do people search for?

Pros

  • +New hires can self-serve
  • +Scales without adding headcount
  • +Shows gaps via search patterns
  • +Foundation for more advanced features

Cons

  • Passive, requires self-motivation
  • Doesn't replace structured program
  • Knowledge base must be created first

Validation Plan

Hypothesis to Test

Companies will pay $99/mo for AI-generated personalized onboarding plans

Validation Phases

1

Manual Plan Generation

1 week
  • Get job descriptions and onboarding materials from 2 companies
  • Manually generate personalized plans using Claude
  • Compare to their existing onboarding
  • Get feedback: better, worse, what's missing?
Generated plans rated better than existing
2

Generator MVP

2 weeks
  • Build role-based plan generator
  • Support 3 common role types
  • Test with 5 upcoming hires across 2 companies
  • Gather feedback from new hires and managers
New hires and managers prefer generated plans
3

Full Program Pilot

4 weeks
  • Run full onboarding program for 10 new hires
  • Track: time to productivity, satisfaction scores
  • Compare to control group with standard onboarding
  • Validate $99/mo pricing
Measurable improvement in time to productivity

Kill Criteria

Stop and move on if any of these become true:

  • Generated plans are too generic to be useful
  • Companies don't have source materials to work with
  • New hires prefer human-led onboarding regardless
  • Too hard to measure ROI