BuhoPilot — From Zero to Live SaaS

At a glance

The challenge

Job seekers applying at volume face the same loop on repeat: find a posting, guess fit against an opaque Applicant Tracking System (ATS), rewrite resume and cover letter, submit—and lose track of what they sent where. Most “AI resume” tools stop at a chat box. They do not persist structured career data, enforce business rules, export ATS-friendly PDFs, or connect discovery to application tracking.

What the client needed

• A credible MVP that could ship to real users—not a prototype demo.
• AI that stays grounded in the user’s actual resume and each job description, with reviewable outputs—not disposable chat text.
• Integrations with job sources and public employer career systems, without building another job board from scratch.
• Monetization and abuse controls suitable for a credit-based SaaS (tiers, limits, subscription readiness).
• Operational foundations: auth, email, analytics, content marketing, and infrastructure patterns the team can extend after launch.

Our approach

We followed the same discipline we use for automation and custom builds—applied here to a full product lifecycle.

01 — Audit & product analysis

We mapped the job seeker journey, competitive landscape (generic LLMs, templates, single-purpose “score my resume” widgets), and high-ROI capabilities for v1: structured master resume, JD-grounded analysis, tailored exports, multi-source discovery, and candidate-side pipeline tracking.

Deliverables: Problem narrative, ICP and personas, feature prioritization, technical feasibility for AI and PDF pipelines, integration inventory (job APIs, ATS public boards, payments, analytics).

02 — Design

We specified user flows, data model boundaries, tier/credit rules, and AI safeguards before implementation—especially the flagship full optimization path (job description → scored analysis → user-selected improvements → tailored resume and optional cover letter → PDF export → tracking).

Deliverables: Flow specs, API contracts, rollout plan for phased features (e.g. automated job hunt on employer ATS pages).

03 — Build

We implemented a production-grade Next.js application with PostgreSQL and Prisma, orchestrated Google Gemini for structured analysis and content optimization, and shipped document generation (HTML → PDF via server-side rendering). Integrations include job listing sources, public Greenhouse and Lever career data for proactive hunt, Stripe-oriented subscription modeling, PostHog product analytics, Cloudflare Turnstile on sensitive flows, and a built-in SEO blog with media storage.

Deliverables: Authenticated web app, AI services with persisted artifacts, admin-ready cost and audit visibility, public marketing site and blog.

04 — Launch & maintain

We prepared the product for real traffic: environment configuration, deployment patterns (including cloud-ready API packaging), monitoring hooks, and documentation so the product can evolve without rework.

Deliverables: Live deployment, handoff-friendly codebase (monorepo: web app, API, shared packages), clear extension points for checkout completion and new integrations.

What we built (client-facing summary)

One source of truth for career data

Users maintain a structured master resume (experience, education, skills, languages, projects) with multiple templates and ATS-oriented PDF export—the foundation for every downstream AI action.

AI workflows that persist—not chat that disappears

• Resume vs. job description analysis with stored scores, keyword gaps, and actionable improvements.
• Full optimization flow on a single screen: analyze JD, score fit, apply selected improvements, generate a tailored resume variant, optional cover letter, and export PDFs.
• Per-section helpers for targeted rewrites without rerunning the entire pipeline.
• Prompts informed by onboarding context (career stage, goals, relocation) so outputs match intent—not generic tone.

Discovery without leaving the product

• Multi-source job search merging external APIs and normalized listings, with deduplication and tier-appropriate access.
• Automated job hunt against configured companies—detecting ATS type, fetching public boards, filtering by title and location, and surfacing leads users can save or dismiss.

Pipeline visibility (candidate CRM)

Application stages from saved through accepted, with counts per stage, rich job metadata, and deep links from AI flows—so users always know where each opportunity stands.

SaaS foundations

• BASIC vs PREMIUM tiers with AI credits and enforced daily/weekly analysis limits.
• Subscription model aligned with Stripe; operational LLM cost tracking for margin visibility.
• Contextual in-product help and walkthrough for a dense dashboard.
• Marketing blog (categories, tags, SEO, media uploads) for organic acquisition.
• Security: NextAuth (Google + email), Turnstile, audit logs, server-side validation on AI routes.

Why this matters for your business

This engagement is proof that we do more than wire Zapier triggers—we own outcomes when the problem requires a tailored system.

Service lines demonstrated

• Custom & Advanced Solutions — Full MVP and bespoke integrations.
• Customer Experience Automation — Context-aware AI assistance embedded in product flows (not a bolt-on chatbot).
• Operations & Efficiency Automation — Metering, audit trails, reporting via product analytics.
• Growth & Revenue Automation — Built-in blog/SEO channel and funnel-oriented onboarding (foundation for lifecycle campaigns).

Technical highlights

• Stack: Next.js (App Router), React, TypeScript, PostgreSQL, Prisma, NestJS API (Lambda-ready), pnpm monorepo.
• AI: Google Gemini with multi-step orchestration, strict JSON parsing, persisted analysis and customized resume entities.
• Documents: Server-side PDF generation (Puppeteer / Chromium) with multiple resume templates and multi-locale content.
• Integrations: Jobicy, Hiring.cafe, public Greenhouse/Lever job data, Vercel Blob, PostHog, transactional email, Turnstile.
• Infra: Terraform-oriented AWS patterns; production deployment on modern hosting.

Results & impact

• Shipped: Public, production web application at buhopilot.com (opens in new tab).
• Workflow compression: Replaced a multi-tool candidate loop with a single workspace—master resume → JD analysis → tailored documents → discovery → pipeline tracking.
• Differentiation: Product persists structured AI artifacts and PDFs; generic LLM chats do not.
• Extensibility: Credit/tier system, subscription schema, and integration layer designed for post-launch growth (payments, new job sources, employer workflows).
• Agency milestone: First complete client journey—strategy, analysis, development, and launch—establishing our playbook for SaaS and advanced automation engagements.

Client testimonial

We did not need another AI resume chat — we needed a product where every analysis, tailored CV, and application stage lives in one workspace. Without a partner who could own strategy through launch, we would still be stitching spreadsheets to generic LLM tools that forget everything the next day. What stood out was the discipline: map the flows and credit rules first, then build AI that persists real artifacts users can review, export, and track.

— Founder, BuhoPilot