AI-native HR · human-in-the-loop by design
Every other vendor is racing to make HR AI do more on its own. We built the opposite: an AI that answers your team’s routine questions from your approved policies, and hard-stops on anything sensitive. A human handles every complaint, leave, and termination. Full audit trail included.
Routine work, handled safely. Sensitive work, always human.
Which features are you most interested in?
Human-in-the-loop by default
The agent drafts case emails and HR approves them. For the first 60 days no email sends itself, and after that only routine categories with a verified source do. The date is enforced in code, not in the prompt.
Grounded answers only
Maxwell answers from approved documents, never from thin air.
Answers in 49 languages
Your team writes in their language and the AI replies in it. Interface translated in English, Spanish, and Chinese.
An audit trail you can prove
Every agent and HR action on a case is written to an append-only log the database will not let anyone edit.
HR lives across a payroll portal, spreadsheets, a shared drive, email threads, and chat. Employees ask the same questions again and again. Sensitive matters get lost in the same inbox as PTO requests.
Requests, cases, knowledge, time off, hiring, performance, compensation, and compliance in one place. The routine is answered instantly from your own policies; the sensitive is routed to a person, with an audit trail.
AI help desk & cases
Maxwell answers routine questions from your approved policies. Anything sensitive becomes a case for a human.
Time off, payroll & benefits
Requests, balances, and approvals over a real accrual engine, with timesheets and benefits enrollment alongside.
Hiring & onboarding
An ATS with AI resume parsing and fit scoring, then onboarding workflows that assign the right tasks automatically.
Performance & goals
Full 360 reviews with competencies, calibration, and AI synthesis, plus goals and PIP tracking that follows through.
Compensation & pay equity
Pay bands, compa-ratios, imported BLS OEWS benchmarks, and a pay-equity regression, with compensation views and changes logged.
People data & insights
Explainable retention risk with a suggested gameplan, turnover explorer, headcount, org design, and pulse.
Compliance & documents
I-9 tracking with an assisted E-Verify workflow (deadlines, reminders, and an audit history; HR files the case in E-Verify itself), COBRA and EEO-1 record tracking, policy answers scoped to each employee’s work state, and AI-drafted documents with native e-signature.
And a great deal more: learning, skills, pulse, org design, e-signature, integrations.
The chat answers routine questions from approved documents behind a classifier. Every HR email the agent drafts waits for a person to approve it.
Employee help desk
Maxwell
Streams answers grounded in approved documents, screens every message for sensitive categories, and opens an escalation case the moment one appears.
HR queue
The case agent
Works the case queue on a schedule: triage, grounded answers, drafted replies for approval, SLA chasing, and closing abandoned cases.
Managers
Co-Pilot
Coaches a manager through a people situation with the handbook and the employee’s state in hand, and drafts the documentation.
Leaders
Coaches Corner
An advisory panel plus team personality dashboards, DISC and Big Five, with personalized reports and 1:1 prep.
The platform adapts to who signs in, enforced by authentication and by row-level security in the database.
A clean self-service portal: help desk, requests, time off, tasks, learning, benefits, pulse, and a confidential drop box.
A focused Co-Pilot, team reviews, time approvals, and the people insights that matter for their reports.
The full command center: cases, approvals, knowledge, compliance, compensation, recruiting, analytics, and the audit log.
Row-level security
Employees see only their own records and shared directory data. HR-only tools are gated in the UI and by database row-level security policies on every table.
MFA and audit
TOTP second factor, required for HR and system admins and re-verified each new session once enrolled, plus an append-only audit trail of every agent and HR action on a case, enforced in the database.
A short, named supply chain
Reasoning and drafting run on one model provider whose business terms bar training on your data. Search embeddings go to a second provider and receive only search text and the policy text you index; omit that key and search falls back to keyword and FAQ matching. Email delivery, HR legal research, and notifications use a small set of named services. Every subprocessor, what it receives, and where, is listed in our subprocessor list, available on request.
Your brand, your tenant
White-labeled per deployment: name, logo, colors, and per-company letterhead on generated documents and forms.
Most suites are a bundle of tools that do not talk to each other. Here everything shares one data model: the help desk knows the time-off balance, the case agent knows the policy, the retention model knows the review, and every screen is scoped by role and by row-level security. The AI is built in, not bolted on, and it is designed to hand sensitive matters to a person rather than guess.
Yes. Each deployment carries its own name, logo, colors, and letterhead, and can run several companies under one HR team. Policies, time-off rules, workflows, review templates, and document templates are all yours to define. Plan tiers (Core, Pro, Enterprise) turn features on as you need them.
Load the knowledge base with your approved policies and add your people (or sync them from your payroll system). For the first 60 days the agent is draft-only by default: every email it writes waits in the Approve Queue for HR, and the date is enforced in code. HR can move that date on the Branding page. Watch deflection and SLA metrics, then layer in compensation, retention, and talent tools as you go.
Row-level security on every table, a TOTP second factor that HR and system admins must enroll in and re-verify each session, and an append-only audit trail of case activity enforced in the database. The AI only reads approved documents and role-scoped data. A filter strips Social Security and card numbers from every reply and email, and the assistant is instructed never to repeat bank details, diagnoses, or pay figures. Reasoning and drafting run on one model provider whose business terms bar training on your data. Knowledge-base search uses a second provider that receives only the text of a search and the policy text you index; employee records, case data and pay fields are never sent to it, and a deployment can run without it, falling back to keyword and FAQ matching. Email, HR legal research and notifications use a small set of named services. Every subprocessor, what it receives, and where, is listed in our subprocessor list, available on request.
Yes. It installs as a progressive web app on iOS and Android, with push notifications for approvals and alerts, and the AI help desk works the same on a phone as on a desktop.
It is built for small and mid-sized employers who want one system instead of six, including multi-company groups run by a single HR team. Plan tiers let a ten-person company start with the help desk and time off and grow into the full command center.
Case and SLA metrics, deflection rate, turnover pivots, headcount, retention risk per employee with the contributing factors, pay bands and pay-equity analysis, pulse trends, and compliance trackers, Turnover pivots, the EEO-1 summary, and the audit log export to CSV.
Employees can write in any of 49 languages and the AI answers in the language they used, with dates, numbers, and text direction correct for their locale, including right-to-left. The interface itself ships fully translated in English, Spanish, and Chinese. Other languages show an English interface until their translation pass lands, and the language picker says so rather than implying a translation that is not there.
A 30-minute walkthrough using your handbook, your roles, and the questions your employees actually ask.