HROne

Partner integrations span five classes. An Integrations marketplace in the primary navigation lists ten named partners, each with its own page: background verification (OnGrid, CFirst, Spring Verify), digital credentialing (ItsCredible), assessments (Testlify), payments and expense (EnKash, SalarySe), tax and statutory filing (Cleartax), and rewards (Pineperks). Biometric device integration ships on every pricing tier, and the vendor states a plug and play model with open APIs for 15+ integrations.

No page documents the agent taking authenticated action inside those systems. The partners are presented as connections that move data into and out of HROne, verification results, assessment scores and tax filings, rather than tools One AI calls to complete a step, and the estate is modest at 15+. The WhatsApp Bot and Teams Bot are the agent's delivery channels rather than integrations.

Sourcehrone.cloud/integration and /pricingread 2026-09-08

One AI runs "autonomous AI workflows that execute multi-step HR actions", with the worked example "approve leave plus sync payroll plus notify manager", on natural language commands without manual orchestration. Its blog describes the mechanism: the system reads the natural language intent, breaks it into steps, runs each action alongside rule-based automation, and escalates to a person at edge cases.

The participants are distinct: the employee who asks, the payroll system that is synced and the manager who is notified are three parties in one command, and the platform publishes persona surfaces for CHRO, team manager, employee and CXO.

Underneath sits workflow machinery on the pricing matrix: F&F settlements with workflow, asset lifecycle workflow, expense workflow, a helpdesk escalation matrix with auto assignment rules, and confirmation, transfer and separation flows.

The product's own contrast is that a chatbot returns an answer and stops, while an employee AI agent completes the task itself.

Sourcehrone.cloud/one-ai, /blog/hrone-one-ai-suite and /pricingread 2026-09-08

The customer's own HR corpus grounds One AI. HROne is the system of record for the employee database with dynamic fields, an HR Handbook holding the company's policies, forms and videos, attendance and leave history, payroll and CTC structures, expense policies and ticket history. One AI reads it on every request: AI policy search and policy aware chat answer from the company's own handbook, AI insights compute attrition risk from that organization's leave patterns, performance dips and engagement drops, and InboxForHR resolves payslip and leave balance queries against the same store.

The corpus persists between runs, scales across an entire workforce far past any context window, and stays queryable in natural language across web, mobile, WhatsApp and voice.

Sourcehrone.cloud/one-ai and /pricingread 2026-09-08

Escalation and approval workflows exist, but nothing gates the agent's ordinary work.

One AI runs each action alongside rule-based automation and escalates to a person at edge cases, and because it processes sensitive employee data it keeps a person in the loop for approvals; the vendor frames orchestration plus human in the loop as why the agent can act autonomously without going rogue.

Approval workflows are real objects on the pricing matrix, covering expenses, exceptional expenses, F&F settlements, advance requests and manpower requests, and the published position is that AI augments HR teams as a co-pilot rather than replacing them.

One AI's pitch is that it executes 110+ actions in real time with no queuing, and no page documents an approval queue, confidence threshold, hold or reversal route before it applies leave, files a regularization or posts a payroll correction. The approvals that exist gate business transactions in the HR workflow, not the agent.

Sourcehrone.cloud/blog/hrone-one-ai-suite, /one-ai and /pricingread 2026-09-08

HROne states it is "ISO 27001 and SOC 2 certified", and both badges, an ISO 27001 mark and an AICPA SOC mark, are rendered in the site footer on every page; compliance with India's Digital Personal Data Protection Act 2023 is stated alongside.

Access is customer administered and evidenced twice: role based access with granular permissions and two factor authentication on the product page, and on the pricing feature matrix a Create Role As Per Your Company control and "Grant access rights across three layers", available on every tier including Basic. Supporting controls are AES-256 encryption at rest and TLS 1.2+ in transit.

The wording is certified rather than aligned with, but no auditor, report date or certificate number is published and no trust portal exists.

Sourcehrone.cloud/hr-software, /pricing and site footerread 2026-09-08

Audit logs are a named platform control, audit trail ready reporting is called out for regulated sectors such as BFSI, and reporting runs to 200+ reports and dashboards on every tier. HROne's blog states that because One AI executes leave, payroll and reimbursement actions, its audit trails must line up with Indian statutes: the Code on Wages, EPFO and ESIC contributions, the IT Act's data protection duties, and POSH and Maternity Benefit entitlements.

That link to the agent's actions is stated as a requirement the product must meet, not documented as a capability a buyer can inspect. No page describes an agent run trace or a per decision record showing which of the 110+ actions One AI took, on whose instruction, with what inputs, or why it escalated. On a platform whose agent executes payroll and leave transactions, that record is what a buyer would demand, and it is not published.

Sourcehrone.cloud/hr-software, /pricing and /blog/hrone-one-ai-suiteread 2026-09-08

One sentence on HROne's own blog documents retained memory: the system reads the natural language intent, breaks it into steps, runs each action alongside rule-based automation, and "escalates to a human at edge cases, then remembers that resolution for next time." That states what is retained, the disposition of an escalated edge case, and when it is reapplied.

The claim appears in a vendor blog post rather than on the One AI product page, the Employee AI Agent page or the pricing matrix, none of which mention memory. No lifetime is stated, no retention or deletion control over what the agent has learned is offered as distinct from employee data, and no per agent or per workflow memory boundary is published. The employee database, attendance history, leave ledger and ticket history are the application's own data model, not agent memory.

Sourcehrone.cloud/blog/hrone-one-ai-suiteread 2026-09-08

Data is hosted "in India on AWS, Mumbai region", a single disclosed region with no choice attached. No second region, regional instance, customer tenant or residency selection surface is published.

The vendor's own pages disagree on the cloud: the product page names AWS Mumbai, while its Google Play listing states "100% data security, Microsoft Azure, GDPR and ISO27001 compliant". Both are vendor authored and neither is dated; the app listing is the likelier to be stale, so the product page reads as current. GDPR alignment and DPDP Act 2023 compliance are legal posture, not residency.

Sourcehrone.cloud/hr-software and the HROne Google Play listingread 2026-09-08

Under the heading "Meet the rest of the One AI crew", the One AI page shows a named set of AI tools: Expense Receipt Parser, Tailored Interview Question, Candidate Profile Creation, Auto JD Creation and Resume Parser, each with its own card and description, alongside the Employee AI Agent with its own page. The primary navigation also lists Employee AI Agent, HR AI Agent and Recruiter AI Agent. Each tool is a whole capability: remove the Resume Parser and Auto JD Creation still works. A prompt library and playbooks are published separately on the AI microsite.

The three navigation agents all resolve to the same /one-ai page, so the plural is a label set rather than three products. The tools are described as built in, not adopted or declined individually, and One AI appears nowhere in the Basic, Professional and Enterprise feature matrix, so there is no selection surface or commercial unit. The separately purchasable add-ons, WhatsApp Bot, Teams Bot, Work Plan, Business Intelligence, Workforce Planning and Payroll Outsourcing, are a pricing structure rather than a pack.

Sourcehrone.cloud/one-ai, /pricing and /ai-at-workread 2026-09-08

Work reaches One AI both on command and unprompted. The command path is broad: InboxForHR supports 110+ HR actions through natural language commands across web, mobile and WhatsApp, and the Employee AI Agent takes voice commands to apply for leave or retrieve information. Five channels are documented: web, the iOS and Android mobile app, WhatsApp Bot, Microsoft Teams Bot and voice, with the two bots sold as named add-ons.

On the unprompted side, One AI automatically detects payroll anomalies, mismatched attendance, missing inputs and double deductions, and flags compliance gaps before a run, and it runs pre-payroll audits ahead of the payroll cycle.

Attrition prediction surfaces at risk employees from leave patterns, performance dips and engagement drops and triggers interventions early.

The pricing feature matrix adds scheduled machinery underneath: auto attendance processing on a schedule, auto salary processing, auto leave credit, recurring scheduler based announcements, and a define auto assignment rule with an escalation matrix in the helpdesk module.

Sourcehrone.cloud/one-ai, /pricing and /add-onsread 2026-09-08

Customers get no model choice. No model picker, admin setting, provider list, routing description or bring your own key option is documented, and no provider is named for One AI, InboxForHR, the Employee AI Agent or the anomaly detection and attrition models.

HROne is specific about what its AI does and silent about what runs it. One AI is described as the only "deeply embedded AI" in an Indian HRMS, using generative AI for policy aware chat, JD generation and on the fly analytics, with the embedding presented as the differentiator against bolt on tools; that is vendor operated machinery by design. The published Prompt Library and Playbooks let a customer shape what they ask, not which model answers.

Sourcehrone.cloud/one-ai, /pricing and /ai-at-workread 2026-09-08

A plug and play model with open APIs for 15+ integrations is stated on the product page and in HROne's own app store listing, and a partner program with Alliance Partner and Business Partner tiers takes applications through a form on its site.

The API itself is not documented: no developer portal, API reference, endpoint list, authentication guide, SDK or documentation subdomain appears on the product, pricing or integrations pages or in the footer, and hrone.cloud/integration is a directory of ten named partner connectors rather than a build surface. HROne Studio at hrone.studio and a separate ProjectOne product appear in the navigation but are not presented as developer surfaces.

Sourcehrone.cloud/integration, /hr-software and /partnersread 2026-09-08

No test environment, sandbox, simulation, versioning, holdout or scored evaluation of the agent's behavior is documented, and nothing lets a customer exercise a change before One AI runs it on live payroll or leave.

Several features look like testing and are not. Pre-payroll audits and payroll anomaly detection test the customer's payroll data for mismatched attendance, missing inputs and double deductions before a run, a quality gate on the customer's operations rather than on the agent; Run Smart Diagnostic Check in the expense module is the same shape.

Attrition prediction and performance analytics measure the workforce. Bi-monthly product updates delivering 300+ enhancements a year are the vendor's own release practice. No bias, fairness or algorithmic audit of One AI is published at all, which on a platform whose AI predicts attrition and screens candidates is worth a buyer's attention.

Sourcehrone.cloud/one-ai, /pricing and /hr-softwareread 2026-09-08

One AI acts inside HROne's own modules, core HR, attendance, payroll, expense and helpdesk, through the platform's internal interfaces, which is programmatic execution in software the vendor built. No hosted or local browser, desktop session, or remote or local computer control is documented.

The WhatsApp Bot and Teams Bot place the agent inside platforms HROne does not own, but they reach those platforms through published messaging interfaces and are channels. Nothing navigates, clicks, fills or advances anything through an interface built for a person. Biometric device integration reads hardware through a device interface, and geo-tagged and selfie attendance are captured in HROne's own mobile app.

Sourcehrone.cloud/one-ai, /pricing and /add-onsread 2026-09-08