# AI Tools for Enterprise HR Teams | Metatable

> Stop answering the same HR question twice.

For HR teams at 1,000+ employee organisations. Four AI tools for the people team: an HR assistant, mailbox triage, onboarding and testing, and CV pre-screening. Our team builds them on your own policies and connects them to Teams, Slack or WhatsApp. A person signs off every decision.

**How we work:** NDA signed before we see a single document. A working version in two weeks on your real policies or mail. Nothing is sent, rejected or used for HR decisions without a person. Exportable source code and database, no lock-in.

## 1. Who it's for

- **Executive sponsor: HR Director / CHRO.** Owns the budget and the outcome. Gets fewer repeat questions reaching the team, visible request volume, SLAs and topic trends, and more employees served with the same headcount.
- **HR business partners** (tool A, HR assistant): stop answering the same policy question every week.
- **HR operations and shared services** (tool B, mailbox triage): the shared inbox arrives sorted, with drafts for routine requests.
- **L&D and training leads** (tool C, onboarding and testing): see who actually understood mandatory training, by topic.
- **Talent acquisition** (tool D, CV pre-screening): an explained shortlist per vacancy; the recruiter still decides.
- **IT and information security** (reviewer): review deployment and data flow; nothing lands on their backlog.

## 2. Where it fits best

Large, distributed workforces. The strongest fit has three or more of: 1,000+ employees; a branch, store, plant or site network; heavy mandatory or regulatory training; high-volume hiring for repeat roles; a shared HR mailbox or helpdesk.

Where each tool pays off first:

| Sector | A. HR assistant | B. Mailbox triage | C. Onboarding & testing | D. CV pre-screening |
|---|---|---|---|---|
| Banking & financial services | first | first | first | useful |
| Insurance | first | useful | first | useful |
| Telecom | first | first | useful | first |
| Retail & consumer chains | first | useful | first | first |
| Energy & utilities | first | useful | first | useful |
| Manufacturing & logistics | first | useful | first | first |
| Healthcare networks | useful | useful | first | first |

## 3. The tools

Three of them run on one HR knowledge base (policies, procedures, templates, portal links, who owns what). The CV tool works from your vacancy criteria and CVs.

### A. HR assistant: routine questions answered in chat, with the source
Works in Microsoft Teams, Slack or WhatsApp, or a web app.

- Answers from approved documents only, linking the source and the right portal form
- Routes "who do I ask?" to the right HR partner, and drafts a request when there's no answer
- Shows HR the top questions and the gaps the knowledge base doesn't cover

Measure: questions resolved without contacting HR.

### B. Mailbox triage: the shared HR inbox, sorted before anyone opens it
- Category, urgency, summary and owner on every email, written back into Outlook
- Draft replies for routine requests; a person reviews and sends
- Sensitive cases skip the AI and go straight to a named person

Measure: time from email to first action.

### C. Onboarding and testing: onboarding that answers back, training that checks understanding
- Role-based day 1, week 1 and 30-60-90 paths, with cited answers to any question
- Scenario simulations scored against a rubric, showing the rule that was missed
- Question banks generated from your documents and approved by HR before use

Measure: first-attempt pass rate, time to productivity.

### D. CV pre-screening: a shortlist with the reasons written down
- HR writes the criteria in plain language; each CV is scored per criterion
- Every score shows the CV line behind it, or a phone-screen question when it's missing
- The recruiter makes every call; age, gender, photo and nationality are ignored

Measure: recruiter hours per vacancy.


## 4. Build or buy

Copilot Studio or an in-house bot is a solid start for Q&A. These tools need more than a chat window: a database, workflows, scoring rules and dashboards. We plug in next to what you've built.

| | Building in-house | With Metatable |
|---|---|---|
| Who builds it | Your IT or digital team, alongside their backlog | Our team, with HR as the owner |
| Scope | Q&A over documents | Q&A plus triage, scoring and simulations on one knowledge base |
| First version | When the backlog allows | Two weeks, on your real data |
| Analytics for HR | Built separately, if at all | Top questions, gaps, SLAs and trends included |
| What you own | Configuration inside a vendor platform | Exportable source code and database |

## 5. Control and security

Guardrails: answers come from your documents with citations; a person makes every hiring and reply decision; sensitive cases skip the AI; training scores are for learning only.

**Where does our data live?** Metatable runs on Microsoft Azure. The hosting region and deployment model are agreed with your security team before any data moves.

**Is our data used to train AI models?** No. Your documents, emails and CVs are processed for your deployment only and are not used to train models.

**Which chat tools does it work with?** Microsoft Teams, Slack or WhatsApp. We connect whichever one your staff open every day, and a web app is always there as a fallback. Mailbox triage works in Outlook through Microsoft Graph.

**Who can access it?** Access is restricted to an approved list of your employees. Integration with your corporate sign-in is scoped on the first call.

**Can data stay inside our perimeter?** Tell us on the first call if data can't leave your environment. Deployment options are agreed, under NDA, before any work starts.

**How is it priced?** A monthly subscription per tool, based on users, document or mail volume, integrations and deployment. Indicative pricing on the first call; a detailed proposal once you pick a tool.

**What about the EU AI Act?** Recruitment and employee evaluation are high-risk uses. CV screening never rejects anyone, training scores aren't used for HR decisions, and every score is traceable and logged.

## 6. Who builds it

Metatable was founded in 2023 by a team that spent 15+ years delivering custom software for clients through its own agency, then built a platform to do that work faster and more reliably. The same small senior team scopes your first tool, builds it on Metatable and runs it after launch. Metatable Inc., team in Kyiv, Lviv and Vancouver.

- **Yaroslav Dzoma**, CEO and co-founder: 15+ years running a custom software agency, after investment banking.
- **Andrii Melnyk**, Head of R&D and co-founder: 20+ years in software; leads the engine that generates and tests each tool.
- **Danila Verbinskiy**, CTO: platform architecture, infrastructure and deployment.
- **Dmytro Yushchenko**, Head of Design: 23+ years in product design.

More about the company: https://metatable.ai/about

## 7. How to start

1. Pick the tool with the most pain (usually the assistant or mailbox triage). A 20-minute call is enough to scope it.
2. Working version in two weeks, under NDA, on your real policies or mail.
3. Add the next tool on the same knowledge base; it reuses the owner matrix from the first one.

Book a call: https://cal.com/metatable/ai-dev-tool-demo or email yd@metatable.ai.

All examples on the page are illustrative; names and policy content are fictional.