The default stack for a modern tech-company recruiting team in 2026 — built around the assumption that AI handles the high-volume, low-judgment work (sourcing, scheduling, screening synthesis) while humans focus on the high-judgment work (relationships, evaluation, closing). Optimized for the 100-1,000 employee tech company hiring at meaningful but not enterprise scale.
How the pieces fit
- Ashby is the ATS backbone. Best-in-class native analytics, modern API, structured interview scorecards as a first-class primitive. Where the funnel data lives and where every stage transition gets recorded.
- Gem is the recruiting CRM. Long-cycle candidate pipelining, outbound sequence management, talent community building. Pairs natively with Ashby — sourcers work in Gem, recruiters work in Ashby, the data syncs.
- juicebox (PeopleGPT) is the AI sourcing layer. Natural-language candidate discovery against 800M+ candidates. Replaces hours of Boolean LinkedIn search with conversational queries.
- BrightHire is the interview intelligence layer. Records and AI-analyzes every interview; produces decision-grade evidence for debriefs; flags interviewer-behavior patterns that hurt candidate experience and quality.
- ModernLoop is the interview-scheduling layer. Automates multi-person multi-stage loop coordination; eliminates the recruiting-coordinator scheduling tax that otherwise consumes 1-2 FTE at this scale.
- Claude is the horizontal AI surface. JD writing, interview loop design, debrief synthesis, rejection feedback, offer prep, weekly recruiting digest. The horizontal AI that connects everything else.
Why this combination
Pre-AI recruiting stacks at this scale were Greenhouse + LinkedIn Recruiter + Calendly + Zoom + maybe Gem — five tools that did their individual jobs but didn’t compound. The AI layer (juicebox + BrightHire + Claude) inverts the model: each piece extends the recruiter’s reach, and the data flows between them produce insights no individual tool surfaces.
The combination accepts six tools (~$200K-$500K annual all-in for a 100-500 employee company hiring 50-150 roles/year) because each does one thing well. Trying to do this with one all-in-one platform (Greenhouse Enterprise, SmartRecruiters Talent Cloud) gives up the AI workbench depth that the specialist tools uniquely provide.
Common variations
- Earlier-stage variant. Replace Ashby with Workable; drop ModernLoop (use Ashby’s native scheduling); use Dover instead of Gem for the smaller candidate-CRM scope. Cuts cost by 60-70% for the under-100-employee stage.
- Enterprise variant. Replace Ashby with Greenhouse Enterprise or SmartRecruiters; add Beamery or Phenom for talent CRM at enterprise scale; add HireVue for high-volume video screening.
- Engineering-heavy hiring. Add HackerRank or CodeSignal for technical assessment; add Holly for autonomous AI engineering sourcing if recruiter capacity is constrained.
- High-volume hourly-hiring variant. Different category entirely — see Fountain plus Harver plus a corporate ATS for the salaried/professional layer above the hourly-hiring stack.
What this stack does NOT replace
- A real recruitment marketing function — career site, content, employer brand work that drives inbound application flow
- A formal structured interviewing discipline — the tools enable it but don’t enforce it; manager and interviewer training still required
- A diversity recruiting program — the tools surface candidates and signals, but the strategic and operational diversity work remains human
- A bias audit infrastructure — required by NYC Local Law 144, EU AI Act, and other regulatory frameworks for AI-augmented hiring at scale