Skip to content
Peopl

TechnologyJun 11, 2026

Bring Your Own AI to HR

Sunrise over terraced rice fields and layered mountains, catching the morning light

Most HR data lives behind one vendor's screens. Getting anything out usually means clicking through someone else's UI, or filing a request and waiting for a report. The data is yours. The way you reach it is not.

That is changing, and not because of another chatbot bolted onto another product.

The AI you already use can now reach your HR system

There is a new standard called the Model Context Protocol, or MCP. In plain terms, it lets the assistant you already use, Claude, ChatGPT, Copilot, connect to other systems and act on them, with permission. Peopl is open over it. So you point the assistant you already use at your Peopl data and ask it to work. No new tool, no new login.

But the interesting part is not Peopl talking to your AI. It is what your AI does next.

Your AI is the wiring

An assistant can hold many connections at once. Yours probably already reaches your email, your calendar, your documents, maybe your e-signature tool or your finance system. Add Peopl, and your assistant can move between all of them.

That is the shift. Peopl is the system of record, and the safe place changes happen. Your AI is the wiring between Peopl and everything else you use. The integrations stop being something a vendor has to build. Your assistant already speaks all of these tools, so it assembles the workflow on the fly, shaped to your exact stack.

Here is what that unlocks.

An offer letter that signs itself. You tell your assistant to promote Alex. Peopl prepares the change, routes it for approval, and once it is approved, generates the promotion letter from your data, the way it already does. From there your assistant carries it the last mile: out for e-signature, then filed back to your drive. Peopl never built an e-signature integration. Your AI already had one.

The board deck, as a real file. "Build me the quarterly people deck." Your assistant pulls headcount, turnover, comp spend, and the hiring funnel from Peopl, and hands you back an actual slide deck with the charts drawn, ready for the CFO. Not an answer in a chat window. A file on your desktop.

Interviews that schedule themselves. Your assistant reads the interview stage of your pipeline from Peopl, checks the panel's calendars for open slots, emails the candidate their options, and books the call. The coordination grind, gone.

Your bench and the whole market, in one list. "Find me a senior engineer with payments experience." Your assistant searches your Peopl talent pool by meaning, searches the open market on the web, dedupes, and returns one ranked shortlist. The people you already have and the people you could hire, side by side.

The number that finally matches. Your assistant reconciles Peopl's payroll queue and approved comp changes against the ledger in your finance system, and flags every mismatch. The spreadsheet that lives in the gap between HR and Finance, retired.

None of these are Peopl features. We did not integrate with your e-signature tool, your calendar, your finance system, or your slide software. We never could, because the right combination is different for every company. Your assistant builds it. The more tools you connect, the more your HR can do, and Peopl never ships a thing.

Your team is already doing this, just less safely

Be honest about what happens today. Someone exports a roster or a comp sheet to Excel and pastes it into a consumer AI tool to get a quick answer. It works. It also sends your people data to a model nobody vetted, with no guardrail and no record that it happened. That is not a hypothetical. It is a Tuesday afternoon.

Bring-your-own-AI does not start that habit. It makes it formal. The same instinct, to ask AI about your people, now runs through a governed door: the data comes through stripped of SIN, banking, and medical details, only the right people can connect it, every call is logged, and any change still needs a person to approve it. You are not fighting the behaviour. You are giving it a front door.

Why this is safe

An AI with a door into HR data raises fair questions. The design answers them.

  • It reads people data, never SIN, banking, or medical information.
  • It can prepare a change, but it cannot make one. Every change routes through your existing approvals, with a person in the loop and a full audit trail.
  • It is your AI, not ours. When your assistant reads Peopl data, it runs under your own agreement with the AI provider you already use, not a model we chose for you. Your AI, your terms. And your data leaves with you whenever you want. No lock-in.

The AI proposes. A person decides. That line does not move.

The shape of HR that is coming

For years, the deal with HR software was simple: put your data in, take the screens and reports we hand back. Data in fast, value out slow.

That deal is ending. Your HR data is becoming something any capable AI can reason over and act on, wired into every other tool you already pay for. The work of HR turns into a conversation, in the place you already work, with the assistant you already trust. And because it rides an open standard, it gets more capable every month, on its own.

You do not have to wait for that. With Peopl, you can connect your AI today.

Peopl Pulse

Your monthly guide to what’s changing in Canadian HR, in plain language.

We break down new legislation, share practical tips, and help you navigate real-world people challenges.

By subscribing you consent to receive the monthly Peopl Pulse email. Unsubscribe anytime.