AI HR Assistant & Policy Chatbot

Key Features

Retrieval-augmented policy Q&A: handbook files are chunked, embedded and cited, so answers quote your actual `HR Knowledge Document` text rather than the model's general knowledge

Grounded leave-balance lookups via a whitelisted `@frappe.whitelist()` method that reads `Leave Ledger Entry` scoped to `frappe.session.user`

Latest-payslip and salary-component queries backed by read-only `Salary Slip` access, never exposing another employee's records

Custom DocTypes: `HR Knowledge Document`, `HR Assistant Conversation`, `HR Assistant Message` for auditable, threaded chat history

`hooks.py` `on_update` doc event re-indexes and re-embeds a handbook whenever a knowledge document is saved or superseded

Hourly `scheduler_events` job reconciles the vector store and prunes stale or orphaned chunks

Onboarding FAQ automation that answers new-hire questions (documents to submit, probation terms, IT setup) from your own onboarding pack

Draft generation for HR letters and employee responses — experience letters, warning letters, policy clarifications — always as a review-first draft, never auto-sent

Permission-aware answers: the assistant honors ERPNext role profiles and user permissions, so it can only surface what the employee is already allowed to read

Pluggable LLM provider (self-hosted, OpenAI, Anthropic, Azure OpenAI) configured in a single `HR Assistant Settings` DocType with no code change

Client script chat widget mounted in the Frappe desk plus an optional whitelisted REST endpoint for a portal or web-form surface

Full prompt/retrieval/answer audit log for every conversation, satisfying HR data-governance and traceability requirements

Fallback and escalation path: low-confidence or out-of-policy questions route to a human HR officer instead of hallucinating

Compatible with Frappe/ERPNext v15 and v16, delivered as a versioned, `bench`-installable app

Built to order, done for you

No DIY setup — a working app, built, installed and supported by ECOSIRE.

You order

Start with a one-time build price. We scope it with you at kickoff.

We build & install

ECOSIRE builds, configures and installs it on your ERPNext.

Go live + support

You go live in about 2–4 weeks, with a post-launch support window.

About this Product

HR teams spend a disproportionate share of the week answering the same repetitive questions: "How many annual leave days do I have left?", "What is the notice period in my contract?", "When is the payroll cut-off this month?", "How do I claim medical reimbursement?". ERPNext already holds the underlying data in Leave Ledger Entry, Salary Slip, Employee and HR Settings, and most companies keep their rules in a handbook PDF — but ERPNext core has no conversational surface that connects the two. Employees either dig through the Frappe desk (which most non-HR staff never learn), raise a ticket, or interrupt an HR officer. ERPNext gives you the records and the workflows; it does not answer questions in plain language and it does not read your policy documents.

ECOSIRE builds an ai_hr_assistant Frappe app that adds a chat surface inside ERPNext and grounds every answer in two sources: your uploaded handbook documents and the requesting employee's own live records. On the retrieval side we ingest handbook files attached to a custom HR Knowledge Document DocType, chunk and embed them, and store the vectors so a policy question is answered from your actual text with the source section cited — not from the model's general knowledge. On the data side, whitelisted server methods (@frappe.whitelist()) expose narrow, read-only lookups such as leave balance and latest payslip, always scoped to the logged-in user via frappe.session.user, so the model never receives raw SQL access and one employee can never read another's payroll.

Technically the app is a proper Frappe module: new DocTypes (HR Knowledge Document, HR Assistant Conversation, HR Assistant Message), a Frappe REST/whitelisted API layer for the chat client script, and hooks.py doc events that re-index a handbook whenever a HR Knowledge Document is saved. A scheduler_events hourly job reconciles embeddings and prunes stale chunks. Access is enforced with ERPNext permissions and role profiles — the assistant honors the same read rights the employee already has, and every drafted HR letter or response is a review-first draft, never an auto-sent action. The LLM provider (self-hosted, OpenAI, Anthropic or Azure) is configurable in a single HR Assistant Settings single DocType, and all prompts, retrieved chunks and answers are logged for audit.

Because scope, handbook content, integrations and the chosen LLM differ for every organisation, this is a build-to-order engagement. After a short scoping call we confirm the exact DocTypes, policies, guardrails and provider, then build against your Frappe/ERPNext v15 or v16 instance. Typical delivery is 2–4 weeks from confirmed scope. We build on staging, run UAT with your HR team, hand over the git repository and installable app, and provide a post-go-live support window. Nothing is generic: the assistant answers from your handbook, over your data, under your permission model.

What you get

Installable source code for your version of the `ai_hr_assistant` Frappe app (v15 or v16), delivered as a bench-installable package

Installation and configuration on your instance, including LLM provider setup and initial handbook ingestion

Technical documentation: DocType reference, whitelisted API methods, hooks and scheduler events, and provider configuration

User guide plus a live training session for your HR team and administrators

UAT on a staging environment with a documented test plan and a rollback plan before production cutover

Git repository handover with full commit history and branch/tag for your release

Post-go-live support window for defect fixes and configuration adjustments

Prompt, guardrail and escalation configuration tuned to your policies during the engagement

Who this is for

HR Operations Manager

Owns the HR shared inbox and fields the same policy and leave questions every day. Needs a self-service assistant that deflects repetitive queries accurately while keeping a human escalation path and a full audit trail.

People / HRIS Administrator

Runs the ERPNext HR module and cares about permissions, data governance and provider choice. Needs the assistant to respect role profiles, keep payroll data private per user, and let them swap the LLM provider without a rebuild.

CHRO / Head of People at a scaling company

Watching headcount grow faster than the HR team. Wants to cut response time on employee questions and free officers for higher-value work, with the assistant grounded strictly in the approved handbook to avoid inconsistent guidance.

IT / ERPNext Systems Owner

Responsible for what gets installed on the Frappe bench. Needs a properly structured app with clean DocTypes, hooks and scheduler events, staging UAT, a rollback plan, and the git repository handed over for maintainability.

How AI HR Assistant & Policy Chatbot Compares

Criterion	ECOSIRE	Custom Build	Competitor	ERPNext Native
Policy question answering	RAG grounded on your handbook with cited sections	Depends on in-house build; often ungrounded	Generic chatbot, rarely reads your documents	No conversational answers at all
Live ERPNext data lookups	Whitelisted read-only methods scoped per user	Possible but must design permissions yourself	Usually no ERPNext data integration	Data exists in DocTypes but no chat access
Data privacy / permissions	Honors role profiles; per-user session scoping	Only as strong as your own guardrails	Often bypasses ERPNext permission model	Enforced, but no assistant layer
Fit to your policies	Built to your confirmed scope and handbook	Fully bespoke but you carry all the work	One-size-fits-all, limited configuration	Manual; HR answers each query by hand
Provider choice / self-host	Configurable: self-hosted, OpenAI, Anthropic, Azure	Whatever you wire yourself	Locked to the vendor's LLM	Not applicable
Delivery model	Build-to-order, 2–4 weeks from scope, staging UAT	Weeks to months depending on your team	Instant install but generic	Already present, no new capability
Maintainability	Clean DocTypes/hooks + full git repo handover	Varies with your engineering discipline	Closed source, vendor-dependent	Standard ERPNext, no extra surface
Ongoing support	Post-go-live support window + optional contract	You own all maintenance	Subscription support, limited customization	Community/partner support for core only

Frequently Asked Questions about AI HR Assistant & Policy Chatbot

Is this a ready-to-download app from the Frappe Cloud Marketplace?

No. This is a build-to-order engagement. ECOSIRE builds, installs and supports the `ai_hr_assistant` app for your specific ERPNext instance, handbook, guardrails and chosen LLM provider. There is no instant download — we build it to your confirmed scope.

How long does delivery take?

Typical delivery is 2–4 weeks from confirmed scope. The scoping call fixes the DocTypes, policies, integrations and provider; the timeline depends on the size of your handbook, the number of data lookups, and any portal or web-form surface you want beyond the desk chat widget.

How is employee payroll and leave data kept private?

Every data lookup runs through narrow, read-only whitelisted server methods scoped to `frappe.session.user`, and the assistant honors your existing ERPNext permissions and role profiles. The LLM never gets raw SQL access, and one employee can never read another employee's payslip or leave records.

Will the assistant make up answers to policy questions?

Answers are retrieval-augmented and grounded in your uploaded handbook documents, with the source section cited. Low-confidence or out-of-policy questions escalate to a human HR officer rather than guessing. Drafted letters and responses are always review-first, never auto-sent.

Which LLM provider does it use, and can we self-host?

The provider is configurable in a single `HR Assistant Settings` DocType — self-hosted/open-weight models, OpenAI, Anthropic or Azure OpenAI. If data residency requires it, we can wire a self-hosted model so no HR data leaves your environment.

Which ERPNext versions are supported?

Frappe/ERPNext v15 and v16. We build and test against your target version and hand over a versioned, bench-installable app tagged for your release.

What kind of support and updates do we get after go-live?

The engagement includes a post-go-live support window for defect fixes and configuration adjustments, plus the full git repository so your team can maintain or extend the app. Longer-term maintenance, version upgrades and new features can be arranged as a separate agreement.