Hiring, onboarding, internal knowledge, project management — AI agents that operate inside your org chart, on your servers. Employee data, contracts, salaries — none of it ever leaves your network.
Each new hire is met by a personal AI agent that knows their role, their team, and your processes.
Contracts auto-drafted. Hardware ordered. Account requests opened.
Personal onboarding agent introduces team, calendar, and first-week plan.
Q&A on internal docs, code, processes — answered from your real KB.
Agent queues warm-up tasks scoped to skill level. Mentor flagged for blockers.
Agent runs structured check-in, surfaces feedback, schedules manager 1:1.
Eight specialised agents, all running on your hardware, all handing off to each other through the orchestrator.
Pulls from LinkedIn, job boards, referrals. Screens against the role rubric. Schedules interviews. Drafts feedback summaries.
Personal AI assistant for every new hire. Knows your stack, your slack channels, your processes — and your tone.
Generates employment contracts, addendums, NDAs from templates. Country-specific clauses for CZ/SK/DE/AT.
RAG over your wiki, code, runbooks, past decisions. Cites sources. Stops the “who knows about X?” loop.
Cross-checks timesheets, expenses, and overtime against contracts. Flags anomalies before run, not after.
Pulls 360-feedback, project deliveries, calibration data. Drafts reviews — manager edits, doesn’t start blank.
Aggregates Jira/Linear, surfaces stalled work, flags resource conflicts, drafts weekly status updates.
Front-line for HR, IT, finance asks. Answers from policy docs. Tickets only the cases that genuinely need a human.
Watches for policy drift. Flags missing trainings, expired certifications, GDPR/ISO renewals.
HR data is the most sensitive personal data your company holds. Salaries, performance reviews, medical notes, contracts. Sending it to a US-hosted AI model is not just a GDPR question — it’s a duty-of-care question.
Amplitica’s HR agents query your existing HRIS (Personio, BambooHR, SAP SuccessFactors, Helios HR) but the LLM never sees raw PII it doesn’t need. PII redaction runs before any prompt. The model itself is open-source weights frozen on your GPU.
Open-source models, deployed where your IT can audit them. The DPO sleeps. The works council signs.
No automated decision affects an employee without a documented human-in-the-loop checkpoint.
Every agent recommendation comes with the prompt, the context, and the model that produced it.
LLaMA 4, Mistral Large 3, Gemma 4, Qwen3 — DPO can audit the model, not just the contract.
Pick one HR or Ops process. We’ll deploy a working agent on your stack within two weeks — open-source, on-premise, GDPR-by-design.