Understanding where your data goes:

The Hub has a 4-tier access system:

Per-tool overrides: The Managing Partner can grant or deny access to individual tools regardless of level. For example, a staff member can be granted access to Controlling, or a manager can be denied access to CV Screener.

Audit trail: Every permission change is logged — who changed it, when, and what was changed. View in the Permissions Console.

Before using AI tools on sensitive engagements, verify:

1. ☐ Client consent — Does the engagement letter permit AI-assisted work? Most CLA engagements do. Check with the engagement lead if unsure.
2. ☐ Data classification — Is this data public, internal, confidential, or restricted? Confidential data can be processed; restricted requires partner approval.
3. ☐ Anonymisation — For GDPR-sensitive data (employee lists, salary data), remove or mask CNPs and personal identifiers before pasting.
4. ☐ Output destination — Where will the AI output end up? Internal memo (fine), regulatory filing (needs verification), client report (needs review).
5. ☐ Retention — AI outputs in localStorage are on your device only. For engagement records, save to the official document management system.

Situation: Maria, Tax Senior, received a 14-email thread from a German subsidiary asking about reverse charge VAT on construction services. The thread included questions from 3 people, attachments referenced, and a deadline for the next day.

Before AI Hub: ~45 minutes to read, summarise, research the regulation, and draft a reply.

With AI Hub: 8 minutes total.

Maria pasted the full 14-email thread into Email Intelligence and clicked Summarise.

Result: Structured summary with bottom line (“Client asks whether reverse charge applies to their subcontractor arrangement”), 4 action items with owners, and a deadline flag for March 15.

Maria opened Knowledge Base and searched “reverse charge construction”. Found the pre-loaded article on VAT reverse charge for construction services (Art. 331 Cod Fiscal).

Result: Confirmed the applicable threshold (€450K in construction revenue) and the conditions. She cross-checked against the AI summary.

Maria went back to Email Intelligence, selected Draft reply with “Professional” tone. The AI drafted a reply addressing each person in the thread, referencing the correct fiscal code articles.

Maria’s edits: Added the specific client contract reference number, adjusted one legal citation, and personalised the greeting. Sent in 3 minutes.

• Tools used: Email Intelligence (Summarise + Draft Reply) + Knowledge Base
• Total time: 8 minutes vs ~45 minutes
• Quality impact: All 3 recipients addressed, correct legal references, professional structure
• Human review: Maria added client-specific context and verified the legal citation — AI cannot replace this step

Situation: New audit engagement for a mid-size manufacturing company. The team needed to: draft the engagement letter, set up the project plan, calculate sample sizes, build ISQM-1 risk assessment, and brief the partner.

Before AI Hub: 2 working days of document preparation.

With AI Hub: One afternoon (4 hours), with higher quality output.

Selected “Engagement letter” template, entered client details, scope (statutory audit FY2025), fee, and team composition. Generated a 3-page letter in Romanian.

Review: Partner reviewed, adjusted two fee clauses, approved. Downloaded as Word.

Created new project from the “Audit engagement” template (9 pre-loaded tasks). Customised deadlines, assigned team members, added 3 client-specific risk areas.

Result: Complete project timeline with milestones from planning through sign-off.

Used Statistical Sampling to calculate sample sizes for receivables (population: 847 items, materiality: €50K). Used MUS for revenue testing. Generated reproducible random selections.

Documentation: Clicked “Sampling memo” to generate ISA 530-compliant documentation for the working papers.

Opened ISQM-1 Builder, reviewed the 29 quality risks. Updated status for 5 risks specific to manufacturing audits. Used AI Gap Analysis to identify additional risks from “new manufacturing client with complex inventory”.

Output: Full risk register + AI-generated evaluation narrative downloaded as Word for the quality file.

Generated a 6-slide partner briefing deck covering: client overview, risk profile, team & timeline, fee structure, key audit areas, and regulatory considerations.

Outcome: Partner had a complete briefing pack for the planning meeting. All documents consistent and professional.

Situation: Three new hires starting April 1 across Audit, Tax, and BPS. Each needs: offer letter, employment contract template, salary certificate, and onboarding checklist. Plus, the team had 12 CVs to screen for an open Senior Associate position.

Uploaded 12 CVs (PDF/DOCX), defined requirements: “ACCA/CECCAR qualified, 3+ years audit experience, Romanian and English fluent.”

Result: AI ranked all 12 candidates with match scores (42%–91%), highlighted strengths and gaps per candidate, and identified the top 4 for interview. Exported as CSV for the hiring tracker.

For each new hire: selected “Offer letter” template, entered name, role, start date, salary, practice area. Generated in Romanian, Labour Code compliant. Then generated “Employment certificate” for each.

3 offer letters + 3 employment certificates + 3 probation review templates = 9 documents, all downloaded as Word, all consistent in format and legal language.

Opened the “New Employee” workflow (9 steps). Checked off completed items for each hire: IT access requested, bank details collected, REVISAL registration prepared.

Situation: CLA Romania has 10 legal entities plus a holding. Every month: trial balances uploaded, mapped to management categories, budget vs actual compared, ICO overhead allocated, and partner/CEO reports generated.

Before AI Hub: 3–4 days of Excel work per month.

With AI Hub: 1.5 days, with better analysis.

Uploaded CSV trial balances for each entity via the Import & Mapping tab. The Romanian chart of accounts (plan de conturi) is automatically mapped to CLA management categories. Reviewed and corrected any unmapped accounts.

Key feature: Once an account is mapped, the mapping persists. After month 3, almost zero manual mapping needed.

Switched to Budget vs Actual tab — instant comparison across all entities. Drilled into variances: identified 3 entities over budget on personnel costs, 1 entity with unusual professional fees spike.

Then: Used Excel Intelligence (Variance Analysis mode) — pasted the budget vs actual figures. AI generated significance flags, likely causes, and recommended actions for the partner report.

Used Export tab to generate: Partner Report (per-entity P&L), CEO Report (consolidated with KPIs), and ICO Detail (inter-company overhead allocation). All formatted and ready for the monthly management meeting.

Situation: A client needed a review of a 28-page service agreement with a German counterparty by end of business. The contract was in English with Romanian-law governing clauses. Crina had 3 other deadlines that day.

Uploaded the PDF to Legal & Corporate AI → Contract Review. The AI analysed all 28 pages and produced:

• Clause-by-clause risk scoring (green/amber/red)
• 3 red flags: unlimited liability clause, auto-renewal without notice period, and missing GDPR data processing agreement
• 4 missing clauses: force majeure, dispute resolution mechanism, IP ownership, and subcontracting restrictions

Used the Contract Drafting module to generate replacement clauses for the 3 red flags. Generated a force majeure clause and GDPR DPA template, both bilingual (EN+RO).

Crina’s review: Adjusted the liability cap to match the client’s insurance coverage, modified the notice period to 90 days (client preference).

Generated an “Advisory memo” summarising the review findings, recommended changes, and risk assessment for the client’s board.

Situation: Peak season. The TP team manages 210 mandates and needs to prepare Local File documentation for 45 clients by the fiscal year deadline. Each Local File requires: functional analysis, method selection, economic analysis, and benchmarking commentary.

Opened the Client Portfolio tab, filtered by documentation status “Pending” and sorted by ANAF risk level. Identified 8 high-risk clients to prioritise first.

Checked the Transactions tab to review intercompany flows for the first client: management fees, licensing, and goods resale across 3 related entities.

Used the AI Analysis tab: described the transaction (“Management fees from Romanian subsidiary to German parent, €1.2M, subsidiary has limited risk”). AI generated:

• Recommended method: TNMM with operating margin as PLI
• Comparable range: 2.1%–7.8% (based on Orbis benchmarking data patterns)
• Risk flags: related-party concentration, thin capitalisation
• Draft functional analysis narrative

Downloaded as Word — became the starting draft for the Local File chapter.

Important caveat: AI output is a draft. The TP specialist reviews method selection, validates comparable ranges against actual Orbis searches, and adds client-specific facts. AI does not replace professional judgement on transfer pricing methods.

Situation: OUG 77/2011 requires pharmaceutical companies to calculate and pay quarterly clawback contributions based on CNAS consumption data. CLA manages 11 entities. Each quarter: receive CNAS notification, match against client data, calculate contribution, generate payment order, file. Deadline for T1: May 31.

For each entity: uploaded CNAS notification (PDF) and client-provided input documents via the Upload Data tab. The dashboard shows a 4-phase progress tracker (INPUT → OUTPUT → REPORTING → SETTLEMENT) for all 11 entities at once.

Entered the RON amount from CNAS, the tool auto-calculated: EUR equivalent at BNR exchange rate + 2% buffer, net amount after deductions. Generated the Payment Authorisation Form (PAF) ready for partner signature.

Drug Register: Cross-referenced CIM codes against the imported drug register to verify product coverage.

Compliance: The tool includes the full OUG 77/2011 procedure, email templates for CNAS communication, and the correct bank payment details — reducing the risk of filing errors.

Situation: Monthly partner meeting in 2 hours. The Managing Partner needs: firm-wide financial overview, practice performance summary, key risk items, and a board-ready slide deck. Previously, this meant 3 hours of assembling reports from different sources.

Opened Analytics Dashboard for the firm-wide KPIs: revenue by practice, utilisation rates, WIP aging, and top clients. Noted that Tax Advisory revenue is 12% above target while Audit is 5% below.

Checked consolidated Budget vs Actual for the quarter. Identified that Entity 3 (BPS) has a 15% overspend on personnel due to the new hires. Exported the Partner Report.

Reviewed the Control Room for open risk items: 2 engagements with overdue milestones, 1 client with outstanding fees >90 days. Flagged for discussion.

Generated a “Quarterly review” presentation with key findings from steps 1–3. 8 slides covering: financial summary, practice performance, risk register, headcount update, and recommended actions.

1. Open Email Intelligence — paste the email thread
2. Click Summarise to get key points and action items
3. Click Draft reply for a ready-to-send response

Pro tip: For partner-level replies, use the Prompt Library to find the right tone, then paste into Document Drafter for a polished final version.

Scope: Works here. Draft output — review before sending.

1. Before: Open Client Brief — enter client name, get context + talking points in 30 seconds
2. Before (detailed): Use Client Brief for client economics, risks, and talking points
3. After: Open Meeting Notes — paste rough notes or upload a Fireflies transcript — get structured summary + action items

Scope: Works here. Internal use — not for CRM record.

1. Open Document Drafter
2. Select template from sidebar: engagement letter, memo, proposal, NDA, LinkedIn post, press release, etc.
3. Fill in: client name, practice, language, tone, context
4. Click Generate document
5. Click Download as Word to save as .doc

Interview mode: Toggle on — AI asks 3-5 questions before drafting for better personalisation.

Drag & drop: Drop a reference file onto the Context field to base the draft on existing content.

Scope: Draft output — not for final storage or approval workflow.

• Excel Intelligence — describe what you need in plain language, paste your column headers, get the formula
• Tax & Financial Calculator — 15 calculators with current 2026 Romanian rates (salary, CIT, VAT, penalties, DCF)
• Journal Entry Assistant — describe any transaction, get the correct debit/credit with Romanian chart of accounts (OMFP 1802)

Scope: Works here. Verify formulas before applying to live data.

Open Legal & Corporate AI — all 10 modules in one tool:

• Contract Review — upload any contract, get risk scoring per clause, missing clauses, redlines
• Contract Comparison — side-by-side diff of two versions with change categorisation
• Contract Drafting — NDA, SPA, MoU, service agreements, addendums. Bilingual (RO+EN)
• Corporate Secretarial — GMS minutes, board resolutions, PoA, ONRC filings
• Due Diligence — upload DD documents, get red flag report with severity tiers
• Litigation Tracker — summarise court documents, calculate deadlines, draft appeals
• Legal Research — query Civil Code, Company Law, GDPR, AML. Citation-ready output
• Legal Translation — RO/EN/DE legal-grade translation
• M&A Deal Room — deal timeline, closing checklists, negotiation tracking
• Client Intake — auto-generate engagement letter, conflict check, KYC pack

Important: All outputs are marked as drafts. Lawyer review required before any document leaves the tool.

• HR Document Generator — salary certificates, employment letters, offer letters, termination notices, probation reviews, promotion notifications. 11 document types. Romanian Labour Code compliant. Download as Word.
• CV Screener — drag & drop CVs (PDF/DOCX), define job requirements, get AI-ranked candidates with match scores, strengths, gaps. Export as CSV.
• Timesheet — log weekly billable hours by client

Plus: Email Intelligence, Excel Intelligence, Meeting Notes, and Document Drafter are all heavily used by the HR team.

Open Transfer Pricing Management — your entire practice in one tool:

• Client Portfolio — 210 mandates, filterable by status, taxpayer type, service, industry
• Transactions — intercompany transaction register with TP methods and risk levels
• Benchmarking — method distribution, margin analysis, Moody's/S&P credit rating table
• Documentation — Local File, Master File, CbCR, APA tracking with deadlines
• Risk Assessment — per-transaction risk scoring + ANAF audit risk indicators
• AI Analysis — describe any TP scenario, get method selection + margin range + risks. Download as Word.
• Import/Export — upload Master Sheet XLSX to refresh all data. 6 export options.
• Q&A — 10 detailed FAQs + AI-powered "Ask a TP question"

Update data: Go to Import/Export tab → upload latest Master Sheet → all 210 clients refresh instantly.

Open Translation Workbench:

1. Text Translation — side-by-side, 6 languages, domain-specific (Legal, Tax, Accounting)
2. Document Translation — drag & drop PDF/DOCX, full extraction + translation
3. Review & Compare — paste source + translation, AI checks accuracy and terminology (1-10 score)
4. CLA Glossary — 22 pre-loaded terms. Add your own. Applied automatically during translation.
5. Quality Check — 7 automated checks: completeness, terminology, numbers, formatting, legal precision

Open Clawback Compliance:

• Dashboard — all 11 entities with 4-phase status (INPUT/OUTPUT/REPORTING/SETTLEMENT)
• Upload Data — upload CNAS notifications, client input, output documents (drag & drop)
• Payment Generator — enter amount, auto-calculate EUR with exchange rate + 2% buffer, generate PAF
• Drug Register — CIM codes, DCI, ATC. Import via XLSX.
• Regulations — full step-by-step procedures + email templates + payment details

Deadlines: T1: May 31 | T2: Aug 31 | T3: Nov 30 | T4: Feb 28

Open Audit Sampling Tool:

1. Statistical Sampling — enter population size, confidence level, tolerable misstatement → get required sample size
2. Monetary Unit (MUS) — probability-proportional-to-size sampling for balance testing
3. Random Selection — generate random numbers with reproducible seeds for audit documentation
4. Systematic Selection — fixed interval with random start

Bonus: Click "Sampling memo" to generate ISA 530 compliant documentation with AI.

Open ISQM-1 Builder:

• Overview — readiness across all 8 SoQM components with risk heat map
• Components — click any component to see objectives, responses, and related risks. Status is editable inline.
• Risk Register — 29 quality risks with likelihood/impact scoring
• Monitoring — 8 activities with deficiency register
• AI Evaluation — Claude generates a full annual SoQM evaluation narrative. Download as Word.
• AI Gap Analysis — describe any situation (new service line, staff turnover) → get ISQM-1 gaps identified

Every signal in the hub follows this chain:

1. Signal detected — correction pending, adoption drop, filing due
2. Decision created — stored in database with title, impact, priority, required role
3. Authority checked — only users with sufficient role can act (staff → senior → manager → director → partner)
4. Action taken — approve, fix, assign, escalate, snooze
5. Outcome recorded — resolved, partially resolved, unresolved, reopened
6. ROI linked — economic impact tracked (cost avoided, risk mitigated, revenue protected)
7. Memory stored — who acted, when, what happened — builds institutional learning

On the homepage: "Today's Decisions" shows your open items. Each card has action buttons based on your role. If you can't act directly, you can escalate.

Compliance rule: Critical compliance decisions cannot be snoozed without partner approval.

If you see a lock icon on a decision card, your role is below the required level. Click "Request approval" to escalate to the right person.

Audit trail: Every action is logged — who, what, when, status change. Even unauthorized attempts are recorded.

Navigate to Daily Pulse from the Hub sidebar or the top nav strip (◊ Daily Pulse).

At the top of the page, review the Firm Decision Health score — an A–F grade based on five factors:

• Decision backlog
• Decision latency
• Outcome quality
• Escalation rate
• Cross-practice risk

Review the key performance indicators: open cases, escalated items, resolved this week, and active signals.

Look for red cards in the escalation alerts section — these are cases that exceeded their time thresholds and need immediate attention.

Work through two priority lanes:

• Decide Now — priority score ≥ 80. These must be handled first.
• Fix Today — priority score 60–79. Handle after the Decide Now lane is clear.

Check the recent activity feed to see who has been active in the last 24 hours and what actions were taken.

On the homepage, find the "Finish this first" card — this is your highest-priority item, selected by the decision engine.

The reasoning line explains why this item outranks others. It considers: overdue status, severity level, client tier, and the computed priority score.

The consequence line tells you what happens if you don't act — potential penalties, revenue loss, or client impact.

Each item shows trust metadata: source system (where the signal came from), confidence level, owner, and detection date.

Look for change markers on each item:

• NEW — appeared since your last visit
• ESCALATED — escalated due to inaction or severity
• NOW OVERDUE — past its deadline
• UPDATED — new information added
• STILL PENDING — waiting longer than expected

Use micro-actions directly from the homepage without opening the full case:

• "On it" — acknowledge that you're working on it
• "Later" — snooze for 24 hours
• "Done" — mark as resolved

Open any case from the Decision Queue.

Scroll to the "Network Context" section within the case view.

The center node is the current case. Connected nodes are grouped by type:

• Signals — what detected this case
• Clients — which clients are affected
• Practices — which practice areas are involved
• People — who owns or is assigned
• Related Cases — click any related case to navigate directly

The decision graph rebuilds daily at 06:15 UTC via an automated cron job. New connections appear after the next rebuild.

Open a case from the Decision Queue or click any case link anywhere in the hub.

The header shows: priority circle (numeric score), severity, confidence, and status tags.

The economic impact grid shows three figures: penalty risk, revenue exposure, and delay cost.

This section provides the context and rationale for why this case was created and why it matters.

The evidence section lists facts and inferences, each with a source and weight, supporting the case assessment.

Historical cases with similar characteristics and their outcomes. Learn how this was resolved before.

Lists blockers that prevent resolution. These must be cleared before the case can move forward.

Graph connections showing related signals, clients, practices, and other cases. See the Decision Graph tutorial for details.

Detailed financial exposure breakdown beyond the summary grid — includes projected costs, risk ranges, and impact timeline.

Three options are presented:

• Recommended — the optimal path with success probability and cost
• Delay — what happens if you wait, with projected impact
• Escalate — hand off to a higher authority with rationale

Sequenced steps to resolve the case, each with a target module link so you can jump directly to the right tool.

Actions Taken — a timeline of who did what and when.

Outcome — resolution details and feedback once the case is closed.

On the homepage, check the "Pending" bar showing: actionable cases, overdue items, corrections, coordination cases, and upcoming deadlines.

Each pill in the Pending bar links directly to the relevant tool — click to jump straight to the work.

The "Waiting on others" lane (orange) shows BLOCKED cases — these need external input, not your action. Don't waste time on these until the blocker is resolved.

The outcomes strip shows: "This week: X resolved · Y fixed · Z actions" — a quick summary of what's been accomplished.

The key principle: this view separates what YOU can act on from what is waiting on someone else. Focus your energy on the actionable items first.

The system automatically escalates cases that aren't acted on within these windows:

• CRITICAL: 4 hours → auto-escalate
• HIGH: 24 hours → auto-escalate
• MEDIUM: 72 hours → auto-escalate
• LOW: 7 days → auto-escalate

When a case is escalated:

1. Status changes to ESCALATED
2. It appears in the Escalation Alert section on Daily Pulse
3. An AUTO_ESCALATE action is recorded in the case timeline with the reason and time threshold

The escalation engine runs daily at 06:20 UTC. Escalations are processed in batch, not in real-time.

To prevent a case from being escalated, take one of these actions before the threshold:

• Acknowledge — click "On it" to signal you're working on it
• Resolve — complete the required action
• Snooze — defer with a reason (resets the escalation clock)

Every item flows through these states:

NEW → ACKNOWLEDGED → IN PROGRESS → READY FOR REVIEW → DONE → VERIFIED

At any point, an item can also become BLOCKED (waiting on someone else) and later resume.

Every item shows a coloured tag telling you what changed since you last opened the hub:

• NEW — This item appeared since your last visit.
• ESCALATED — This item was auto-escalated because nobody acted in time.
• NOW OVERDUE — The deadline passed since your last visit.
• NEEDS REVIEW — Submitted for your review since last visit.
• BLOCKED — Became blocked since last visit.
• UPDATED — Something changed (status, owner, priority).
• STILL PENDING — Was here before and has not been resolved yet.

When you mark an item as blocked, the system asks for a reason. This helps managers see what is actually stuck and why:

• Awaiting client input — You need information from the client.
• Awaiting manager approval — A manager needs to sign off.
• Awaiting colleague response — Another team member needs to provide something.
• Awaiting system update — Waiting for a system sync or data update.
• Awaiting missing document — A required document has not been received.
• Awaiting external authority — Waiting on ANAF, a court, or another authority.
• Dependency conflict — Another task must complete first.
• Capacity constraint — The team is at capacity and cannot take this on now.

The actions you see depend on your role and the item’s current state:

Staff / Consultants: On it, Start work, Done, Submit for review, Blocked, Later

Managers: All of the above, plus Approve, Return for changes, Reassign

Partners: All of the above, plus Verify, strategic oversight actions

Hover over any action button to see a tooltip explaining exactly what it does.

Snoozing defers an item until tomorrow without changing its state. But:

• After 3 snoozes, you receive a warning.
• Repeated snoozing signals that the item may need to be reassigned, resolved, or escalated.
• Managers can see snooze counts in oversight views.

If you find yourself snoozing the same item repeatedly, consider marking it as blocked (with a reason) or asking for help.