> For the complete documentation index, see [llms.txt](https://gdplabs.gitbook.io/catapa/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://gdplabs.gitbook.io/catapa/digital-employee-de/list-of-de/claudia-de-hr/claudia-de-payroll-analyst/user-guide.md). # User Guide ### Introduction **DE – HR Compliance Analyst (Camila✨)** runs compliance checks against your HR data using CATAPA's Anomaly Detection engine, flags violations, and suggests corrective actions — without you needing to navigate the Anomaly Detection UI or scan raw reports. The DE operates **on-demand via CATAPA Chat**. You send one message describing what you want to check and for which period; Camila**✨** confirms the request, runs the check, and delivers results. **You only see what you're allowed to see.** The DE runs under your CATAPA login, so all results respect your existing role and data access. If you can't see certain employees' data in CATAPA directly, Camila**✨** won't include them either. ### What You Will Receive When the run completes, you receive two things: **In CATAPA Chat — Findings Overview** A structured summary delivered in chat as soon as it's ready, organized around four questions: * What is the critical issue this run? (Severity = High) * What issue occurs most frequently? * Are there any patterns or insights across findings? * What general recommendations apply to the company? > ⚠️ **HITL Advisory:** Review the Findings Overview before acting on it or sharing it externally. Camila's recommendations are grounded in rule metadata and CATAPA-curated guidance, but final action and accountability remain with you. **Via Email — Excel + PDF** * **Excel Report** — the engine's raw output, with per-rule sheets and per-employee finding rows. * **PDF Compliance Report** — a structured report containing: (1) Cover, (2) Overview with Findings Overview + executive summary + run data statistics (total, successful, and failed records processed), (3) Rules Checked Summary per rule, (4) Detailed Findings per rule per employee with per-finding recommendations. ### Benefits Camila**✨** helps HR and compliance teams **reduce manual compliance-checking effort** and deliver findings that are **more consistent, proactive, and easier to act on** through: * **Reduced manual effort** — Runs all in-scope compliance rules across the entire workforce in a single request, eliminating hours of UI clicks and Excel scanning. * **Earlier signal detection** — Surfaces violations before your payroll disbursement or review deadline, when corrections are cheap. * **Grounded recommendations** — Every finding row carries a corrective recommendation tied to the rule's metadata, with output checks on numbers and citations. * **Consistent, scalable execution** — The same rules run the same way every time, across the whole workforce, regardless of headcount. * **Parameter transparency** — Camila**✨** always confirms the rules, period, and delivery email before running. No silent assumptions. *** ### Who Can Use It Anyone who: * Has access to **Camila✨ DE HR Compliance Analyst** in CATAPA Chat, and * Has **authorized the CATAPA GL Connector** in their account can trigger a compliance run. There are no additional role restrictions. Camila**✨** runs under your CATAPA login, so all results are scoped to what your account is already authorized to see. > **First-time setup:** If you haven't authorized the CATAPA GL Connector yet, Camila**✨** will prompt you with setup instructions on your first request. Authorization persists until revoked. *** ### What You Need Before You Start To run a compliance check, include in your message: 1. **Which rules to run** (required) — name specific rules, or say "all rules" to run every available check. 2. **Period** (required) — the month and year range (e.g., "May 2026" or "January to March 2026"). Future periods are supported. **Input rules:** * **Rule names** can be exact, partial, or a common alias. If Camila**✨** can't match, she'll show you the available rule list. * **Period** must be month & year format. For a single month, start and end are the same. * If a required parameter is missing or ambiguous, Camila**✨** asks a clarification question before proceeding — no silent assumptions. *** ### **In-Scope Use Cases** Camila**✨** can run **any compliance rule configured on CATAPA's Anomaly Detection platform** — she dynamically discovers new rules as they're added, with no changes needed on your end. The **initial set of 8 rules** shipped with the MVP falls into two categories:
#CategoryUse Case
1Statutory (Tax)PTKP validation against gender and marital status
2Statutory (Tax)NPWP vs KTP consistency for applicable local employees
3Statutory (Tax)Employee Tax Object validation based on employment type & foreign labor status
4Statutory (Wage)Minimum wage (UMR/UMP) compliance based on employee location
5Statutory (Contract)PKWT end-of-contract compensation presence
6Statutory (Holiday)THR (Eid al-Fitr) payment timing (N days prior to Eid, configurable)
7Internal PolicyTake-Home Pay deduction limit vs. configured maximum
8Internal PolicyRegular salary payment date vs. configured latest allowed day
> To configure internal policies or add new rules, contact your CATAPA team. *** ### How to Run It via CATAPA Chat #### Recommended prompt template You can use this prompt and adjust it accordingly in CATAPA Chat: > ``` > Run all compliance checks for [period] > ``` > > ``` > Check [rule name] compliance for [period] > ``` > > ``` > Run PTKP and UMR checks for [period] > ``` > > ``` > Run all compliance checks for [period] and send the report to [email address] > ``` > > ``` > Did my compliance check finish? / What was the result of my last run? > ``` #### What you do 1. Open CATAPA Chat. 2. Send your request. 3. Confirm the summary Camila**✨** shows you (rules, period, delivery email) — or ask her to change it. 4. Wait for the **Findings Overview** in chat. 5. Check your email for the **Excel + PDF Compliance Report**. #### What Camila**✨** does behind the scene 1. **Parses your request** — extracts rules and period, asks for clarification if anything is missing. 2. **Confirms with you** — presents a summary and waits for your approval *(Human-in-the-Loop)*. 3. **Runs the check** — connects to the Anomaly Detection engine under your CATAPA identity and executes the selected rules. Camila**✨**'s progress shows as a thinking indicator in the chat sidebar. If your session times out, the run continues; you can ask Camila**✨** for the result later. 4. **Generates findings & recommendations** — parses the engine's Excel output, generates per-finding recommendations grounded in each rule's metadata, and builds the four-section Findings Overview. 5. **Delivers results** — Findings Overview to chat, then Excel + PDF to your email. She confirms both. #### 1) Statutory (Government) Regulations These checks help identify potential non-compliance with Indonesian payroll-related regulations, including: * **Tax compliance (PPh 21-related attributes)**: * **PTKP validation** against gender and marital status * **NPWP vs KTP consistency** for applicable local employees * **Employee Tax Object validation** based on employment type and foreign labor status * **Minimum wage compliance (UMR/UMP)** based on employee location and UMR/UMP references * **Selected government regulation checks**: * **PKWT compensation** presence of end-of-contract compensation when a contract ends in the checking period * **THR (Eid al-Fitr) payment timing** relative to Eid al-Fitr day (maximum of N days prior to Eid al-Fitr, where N is configurable). #### 2) Internal Company Policies These checks validate your payroll outputs against your company’s configured policies, including: * **Take Home Pay deduction limit** (Take Home Pay deduction percentage against the configured maximum). * **Regular salary payment date policy** (Regular salary payment date against the configured latest allowed day of month). To configure the policies, please contact our team. #### What Camila**✨** will not do Camila**✨**'s scope is intentionally narrow. Understanding these boundaries helps set the right expectations: * **Camila✨ does not correct data.** She identifies findings and recommends actions; corrections are made by your team directly in CATAPA. * **Camila✨ does not run automatically.** Every check must be triggered by a chat message. Scheduled runs are not available in the current version. * **Camila✨ runs one check at a time.** The Anomaly Detection engine allows only one active run per tenant. A second request while a run is in progress is blocked, not queued. * **Camila✨ does not show individual employee data in chat.** The Findings Overview contains aggregates only — names, NIK, salary, and other PII appear only in the Excel and PDF sent to your email. *** ### Understanding The Reports #### Findings Overview (in chat) Delivered in CATAPA Chat as soon as it's ready — before the email. Contains aggregate information only; no individual employee data appears in chat.
SectionWhat it shows
Critical IssueAny rule or finding with Severity = High requiring priority attention
Most Frequent IssueThe rule or finding type that appears most across the run
Patterns / InsightsCross-rule patterns, if a clear signal exists (optional section)
General RecommendationsCompany-level recommended actions, deduplicated across the run
#### PDF Compliance Report (via email) A structured report covering the full run: 1. **Cover** — run metadata (period, rules, date). 2. **Overview** — Findings Overview + executive summary. 3. **Rules Checked Summary** — per rule: total employees checked, total findings, rule-level severity (High / Medium / Low). 4. **Detailed Findings** — per rule, per employee: violation description and corrective recommendation. #### Excel Report (via email) The anomaly detection engine's raw output, with one sheet per rule that produced findings. Each row is a single finding for a specific employee, including finding-level severity where configured. #### Recommendations * Each finding with a violation receives a **contextual per-finding recommendation** generated by Camila**✨**, grounded in Rule Metadata and CATAPA-curated Recommendation Hints. * Recommendations cite only regulations explicitly authorized per rule — no unchecked external citations. * If grounding sources are absent or validation fails, Camila shows a *"can't generate"* message for that finding instead of a recommendation. See **Understanding recommendation caveats** below. #### Understanding recommendation caveats Camila**✨** generates per-finding recommendations using two grounding sources: **Rule Metadata** (what the rule checks and why) and the **CATAPA-curated Recommendation Hints Library** (guidance on what corrective action to suggest). The recommendation column in the PDF Detailed Findings section will show one of three states depending on how much grounding was available: | What you see in the report | What it means | What to do | | ---------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------- | | Recommendation text only | Fully grounded — both Rule Metadata and a well-matched (high-overlap) Hints entry were available, and validation passed | Act on the recommendation after your own review | | Recommendation text + *"Limited context for this finding — please verify before acting"* | Partially grounded — one grounding source was missing, or the topic-tag match between the finding's rule and the Hints entry was a low-overlap match | Use the recommendation as a starting point; apply additional judgment before acting | | "I couldn't generate a recommendation here — please contact admin or CATAPA support" | Could not generate — both grounding sources were absent, or validation failed after retry, or a system failure occurred | Escalate to your compliance lead or contact CATAPA support | > The AI caveat *"These are AI-generated suggestions — please review before acting"* appears at the **footer of the PDF Report** and applies to all recommendations in the report. *** ### After You Receive Findings Camila**✨** surfaces findings; **HR decides and acts**. * **Start with Critical Issues (Severity = High)** in the chat overview — these need attention first. * **Cross-check with the PDF and Excel** before applying corrections. The PDF gives context; the Excel gives row-level data. * **Apply corrections in CATAPA.** Camila has no write access to any system. * **Re-run the check after corrections** to confirm the findings are resolved. * **Escalate High-severity findings** to your compliance lead before your review deadline. *** ### Data Privacy * **Access boundary.** Camila**✨** runs under your CATAPA login. You only receive findings for employees you're already authorized to see — enforced at the engine level, not by Camila**✨**. * **What Camila✨ keeps.** Nothing. The Excel and PDF sent to your email are your responsibility afterward; parsed content and generated files are discarded after delivery. * **What appears in chat.** Aggregates only — finding counts, severity levels, rule names, and general recommendations. **No employee names, NIK, salary, or other PII ever appear in chat.** *** ### Troubleshooting & FAQ #### Camila**✨** returned an error or the run failed. What should I do? Re-send your request. Camila**✨** will check the status of any prior run and guide you from there. #### I didn't receive the email. 1. Confirm the email address in Camila**✨**'s summary is correct. 2. Check your spam folder. 3. Re-run if needed — safe to retry. #### I only received the Excel, not the PDF. PDF generation failed — Camila**✨** will have flagged this in chat. The Excel contains the full row-level data. Re-run to try the PDF again. #### A run is already in progress. Can I start another? No — the engine allows only one run per tenant at a time. Wait for it to finish, then ask Camila**✨** for the result in a new session; she'll resume from where the run ended. #### What if the data for the period is incomplete? Camila**✨** runs the check against whatever data is available at that moment. If data changes afterward, re-run for updated results. #### My CATAPA GL Connector authorization expired. What do I do? Camila**✨** will detect this and prompt you to re-authorize in chat. Follow her setup instructions. #### My chat session ended before the run finished. Did I lose the results? No. If your session ends mid-run, the engine keeps running. Open a new session and ask Camila**✨** for the previous run's status (e.g., *"Did my compliance check finish?"*). If it's done, she'll resume delivery — Findings Overview + email — without you re-triggering the check. If it's still running, she'll ask you to check back later.