Onboarding Coach.

The problem

The first 90 days in a new role decide whether you thrive or quietly drown — and the conditions that make it hard are getting worse. Hiring slowdowns mean new hires are under pressure to prove value faster. Remote and hybrid work removes the casual feedback loops that used to tell you "that meeting went well." "Quiet quitting" fears mean managers expect visible credibility from day one.

The new hire's pain stack: ambiguity ("I don't know what 'good' looks like in this specific role"), overwhelm (too much information, no prioritization), fear / psychological safety (afraid to ask basic questions and look incompetent), isolation (no one to talk to about the small daily uncertainties), and the gnawing impostor syndrome that affects even senior ICs and first-time managers most.

The solution

Onboarding Coach is a personal growth tool, not a company tool. It doesn't sit between you and your manager; it sits between you and your own head. It generates a personalized 90-day plan structured into phases with weekly focus and daily tasks, then runs recurring check-ins that adapt the plan as the reality changes.

The deliberate non-feature is the Daily Standup Bot that auto-reports to your manager — rejected during product design because it breaks psychological safety, and users won't trust a tool that reports to their boss. Onboarding Coach is private by design.

How it works

Three phases: (1) Onboarding — a guided multi-step form collects role, start date, company context, emotional context (excitement, worries, situation tags), and goals (personal goals, manager expectations). (2) Plan Generation — AI synthesizes a 90-day roadmap with phase structure (e.g. Listen / Contribute / Lead), weekly focus areas, and daily tasks; user reviews and commits phase by phase. (3) Check-ins — hub-and-spoke navigation from dashboard, users initiate check-ins, AI adapts the plan as reality shifts.

Model selection was rigorous: 5 models compared across 9 personas × 2 runs, graded by GPT-5.4 as LLM-as-judge, automated via n8n + Airtable. Winner: GPT-5.4 at temperature 0.3. Prompt architecture evolved from a single monolithic instruction set into a pipeline of specialized stages (clarify → generate → coach → update) — the modularity dramatically improved consistency.

Who it's for

Primary persona — The Anxious Achiever: mid-to-senior professional starting a new role, or a first-time manager. Attributes: high stakes (came in via a competitive process), wants to make a strong impression, likely experiencing impostor syndrome, feeling overwhelmed by information density. Required inputs: just role and start date. Optional inputs deepen personalization across job context, emotional context, goals, and concerns.

Business model: B2C primary (the individual new hire buys it for themselves during the offer-acceptance to first-day window). B2B secondary: HR departments offering it as part of the new-hire benefits package — which solves the trust problem (employees won't trust a company-bought tool that reports back; HR can buy it explicitly *not* to receive reports).

Why it matters

The AI coaching and career development market is one of the fastest-shifting categories in the AI wave — traditional employment is moving to more autonomous, results-based work, remote/hybrid is the default, and personalized AI advice is finally good enough to replace the executive coaching that only senior people could previously afford.

Onboarding Coach's strategic bet is on the personal-growth-tool framing: by refusing to report to managers, HR systems, or anyone but the user, it preserves the psychological safety that company-bought tools forfeit by default — and turns "I'm afraid to look incompetent" from a churn driver into the exact reason the product exists.

At a glance

Project

Onboarding Coach

Built by

Dino Sepac

One-liner

Personal AI career coach for the first 90 days — generates a personalized onboarding plan and runs weekly check-ins for the Anxious Achiever.