Table of Contents
- Why Shared Services AI Matters for Portfolio Companies
- The Centralisation Decision: What to Move, What to Keep Local
- Finance Shared Services: The AI Opportunity
- Operations Automation Across the Portfolio
- Building Your Shared-Services Technology Stack
- Governance, Data, and Security at Scale
- The Savings Model: What Gets Cheaper, What Doesn’t
- Implementation Roadmap: From Pilot to Scale
- Common Pitfalls and How to Avoid Them
- Next Steps: Getting Started with Your Portfolio
Why Shared Services AI Matters for Portfolio Companies
If you run a portfolio of companies—whether through a PE firm, a venture studio, or as a multi-entity startup group—you’re sitting on a massive inefficiency. Each company has its own finance team, its own ops workflows, its own vendor contracts, its own data silos. Multiply that across 5, 10, or 20 portfolio companies, and you’re paying for 20 versions of the same problem.
Shared services is not new. IBM’s overview of the shared services operating model shows that centralisation has been a staple of large organisations for decades. But AI changes the economics. Where a traditional shared-services centre might save 15–25% on labour through process standardisation, AI-driven shared services can save 40–60% on transaction costs while freeing your portfolio CFOs and ops leaders to focus on unit economics, product-market fit, and growth.
The math is straightforward. If your portfolio spends $2M annually on finance and ops labour across ten companies, shared services with AI can cut that to $800K–$1.2M. More importantly, it cuts the time your portfolio companies spend on back-office work from weeks to days. That’s time your founders and operators get back to focus on revenue, product, and fundraising.
But shared services AI isn’t a plug-and-play box. It requires clear decisions about what to centralise (accounts payable, payroll, general ledger reconciliation, basic HR workflows), what to keep distributed (vendor selection, capital allocation, hiring decisions), and what technology to build or buy. Get those wrong, and you’ll centralise the wrong thing, upset your portfolio companies, and fail to capture the savings.
This guide walks you through the entire journey: how to decide what to centralise, how to use AI to automate the workflows that matter most, how to build a governance model that works, and how to measure and realise the savings.
The Centralisation Decision: What to Move, What to Keep Local
Not everything should be centralised. The art of shared services is knowing where to draw the line.
What Should Always Be Centralised
Certain finance and ops functions are pure commodity. They don’t vary meaningfully across your portfolio companies, and they’re expensive to replicate:
Accounts Payable (AP). Invoice receipt, coding, three-way matching, approval workflows, and payment processing. These steps are identical whether you’re paying a software vendor at a fintech or a logistics company. Centralising AP removes duplicate vendor master data, eliminates duplicate invoices, and lets you negotiate better payment terms across the portfolio.
Payroll and Benefits Administration. Salary processing, tax filing, superannuation contributions, leave tracking, and benefits enrolment. Payroll is highly regulated and standardised. A single payroll engine serving your entire portfolio, configured once per jurisdiction, is far cheaper and more accurate than each company running its own.
General Ledger (GL) and Month-End Close. Chart of accounts reconciliation, inter-company eliminations, consolidation, and financial reporting. If your portfolio companies report to a single holding company or fund, you need a single GL. This is where AI can shine: automated reconciliation, anomaly detection, and close-cycle compression from 10 days to 3.
Accounts Receivable (AR) for Standard Terms. If your portfolio companies have similar payment terms and customer profiles, centralising AR collection, dunning, and dispute resolution is a no-brainer. This is one of the highest-ROI automation targets: AI-driven dunning sequences can recover 15–20% more cash with zero manual effort.
Procurement and Vendor Management. Centralise the vendor master, contract terms, and approval workflows. This gives you portfolio-wide visibility into spend, eliminates duplicate contracts, and lets you consolidate volume to negotiate better rates.
Compliance and Audit Support. Tax filings, regulatory reporting, audit schedules, and documentation. Centralise the calendar and the workflows so your portfolio companies don’t each manage their own audit trail.
What Should Stay Local
Some functions are core to each company’s identity, competitive position, or risk profile. Keep them local:
Hiring and People Strategy. Your fintech founder may need a different hiring bar and compensation band than your logistics company. Keep hiring decisions, culture, and talent strategy at the company level. The shared-services centre can handle onboarding administration and benefits enrolment, but hiring is local.
Capital Allocation and Investment Decisions. Each company’s board or leadership team should decide where to spend money. The shared-services centre should provide clean financial reporting and scenario modelling, but not make the decision.
Product and Commercial Pricing. Your portfolio companies may have very different unit economics. Centralising pricing decisions or cost allocation would be a mistake. The shared-services centre should provide transparent cost reporting so each company can make its own pricing decision.
Customer and Vendor Relationships. If a portfolio company has a strategic relationship with a customer or vendor, keep that relationship local. The shared-services centre can handle transaction processing, but not the relationship.
Data Privacy and Regulatory Decisions. If one portfolio company is in financial services and another in healthcare, they have different compliance obligations. Keep those decisions local, and let the shared-services centre provide the infrastructure and audit trails to support them.
The Grey Zone: What Depends on Your Portfolio
Some functions are candidates for centralisation depending on your specific portfolio:
IT and Infrastructure. If your portfolio companies have similar tech stacks and security requirements, centralising cloud infrastructure, identity management, and basic IT support is smart. If one company is a regulated bank and another is a consumer app, you may need separate infrastructure and security postures. The shared-services centre can provide a baseline (cloud accounts, identity, backups), and each company can layer on additional controls.
HR and People Operations. Centralise the transactional work (benefits, payroll, leave tracking). Keep talent strategy, compensation philosophy, and hiring standards local. This is where Oracle’s guide to shared services in finance and operations becomes useful: you can centralise the system and the process, but not the decision.
Data and Analytics. This is where AI and shared services intersect most powerfully. Centralise the data platform, the ETL, the data quality checks, and the analytics infrastructure. Let each company build its own dashboards and business logic on top. This is the foundation for portfolio-wide AI: a single source of truth for finance, ops, and customer data, with AI agents running anomaly detection, forecasting, and recommendations at scale.
Finance Shared Services: The AI Opportunity
Finance is where shared services delivers the fastest ROI. And AI is where finance shared services gets interesting.
The Traditional Finance Shared-Services Model
Traditional finance shared services centralises transaction processing: AP, AR, payroll, GL reconciliation. Workday’s explainer on finance shared services outlines the standard playbook: centralise the process, standardise the chart of accounts, implement an ERP, and hire a team to process transactions at 30–40% lower cost than distributed finance teams.
That model works. It saves money. But it’s labour-intensive. Your shared-services team spends their time on exception handling, manual matching, and chasing missing invoices.
AI Transforms Finance Shared Services
AI changes the game in three ways:
1. Automation of Exception Handling. In a traditional shared-services model, 70% of the work is routine (invoice matching, GL coding, reconciliation), and 30% is exceptions (duplicate invoices, missing POs, coding mismatches). AI flips that: it handles the 70%, leaving your team to focus on the 30%. But more importantly, AI can handle many of the exceptions too. Machine learning models trained on historical data can code invoices, flag duplicates, and reconcile accounts with 95%+ accuracy.
2. Real-Time Visibility and Anomaly Detection. Traditional shared services operates in batches: daily, weekly, or monthly closes. AI enables real-time monitoring. Imagine an AI agent that watches your AP queue, flags unusual invoices (wrong amount, wrong vendor, wrong account code) in real-time, and escalates only the true exceptions to your team. This compresses your close cycle from 10 days to 3, and catches fraud and errors immediately.
3. Predictive Cash Flow and Scenario Modelling. Once you have centralised finance data, AI can forecast cash flow, predict customer churn, and model scenarios. Your portfolio CFO can ask: “If we slow down customer collections by 10 days, how does that affect our cash runway?” and get an answer in seconds, not weeks.
The Finance Shared-Services AI Stack
To realise these benefits, you need:
A Central Data Platform. All your portfolio companies feed their financial data into a single data warehouse (Snowflake, BigQuery, Databricks). This is the foundation. Without it, you can’t automate, you can’t get real-time visibility, and you can’t train AI models. SAP’s introduction to shared services models emphasises the importance of a unified system; a modern data platform is the equivalent for AI-driven shared services.
An ERP or Finance System. QuickBooks, Xero, NetSuite, or SAP. This is where transactions live. Your shared-services team uses this system daily. The key is that all portfolio companies use the same system (or have clean APIs that feed a central data warehouse).
Invoice Processing AI. Tools like Coupa, Ariba, or custom models trained on your own invoice data. These read invoices, extract data, code them to GL accounts, and flag duplicates. You should expect 85–95% accuracy on first-pass processing, with the remainder escalated to your team.
Reconciliation and Matching AI. Tools like Blackline or custom models that match invoices to POs and receipts, reconcile GL accounts, and flag unusual patterns. Again, 90%+ accuracy on routine matching, with exceptions escalated.
Forecasting and Scenario AI. Tools like Anaplan or custom models built on your data warehouse. These forecast cash flow, predict customer churn, and model scenarios. This is where you move from “what happened” to “what will happen” to “what should we do.”
The Finance Shared-Services Savings Model
Let’s work through the numbers. Assume you have a portfolio of ten companies, each spending $150K annually on finance labour (1.5 FTEs at $100K each). Total portfolio finance spend: $1.5M.
With a centralised shared-services model, you can consolidate that to 4–5 FTEs (1 director, 2–3 processors, 1 analyst). Cost: $400–500K. Savings: $1M–$1.1M (67–73%).
But that’s the labour savings. The AI savings come from:
- Faster close cycle: Your portfolio companies close in 3 days instead of 10. That’s 7 days of working capital freed up across the portfolio. At $1M in monthly expenses, that’s $230K in working capital released.
- Better cash collection: AI-driven dunning recovers 15–20% more cash. If your portfolio has $10M in annual AR, that’s $150–200K in incremental cash.
- Fraud and error prevention: AI catches duplicate invoices, coding errors, and unusual transactions before they become problems. Expect to prevent $50–100K in fraud and error annually.
- Better vendor terms: Centralised procurement gives you visibility into spend and lets you consolidate volume. Expect 5–10% better terms on repeat vendors. At $5M in annual spend, that’s $250–500K in savings.
Total first-year savings: $1.6M–$2.1M. Payback on the shared-services centre (setup, tooling, training) is typically 6–9 months.
Operations Automation Across the Portfolio
Finance is the obvious place to start, but operations automation across the portfolio is where the real leverage lives.
What Ops Workflows Should Be Centralised and Automated
HR Onboarding and Offboarding. When a new employee joins any portfolio company, they need a laptop, email, access to systems, benefits enrolment, and tax setup. This is identical across your portfolio. Centralise it: a single onboarding workflow that creates accounts, sends welcome materials, and enrols benefits. Use Workday, BambooHR, or a custom workflow. AI can automate the repetitive parts: sending welcome emails, scheduling training, and flagging missing documentation.
Expense Management and Reimbursement. Employees across your portfolio submit expenses, managers approve them, finance processes reimbursement. Centralise the process: a single expense system (Expensify, Concur, or custom) with a single approval workflow. AI can categorise expenses, flag policy violations, and process reimbursement. Expect 90%+ automated approval rates.
Customer Onboarding and Offboarding. If your portfolio companies have similar customer profiles, centralise customer setup: account creation, access provisioning, contract setup, and billing configuration. This is especially valuable if you’re consolidating customer data across the portfolio (for example, if a customer uses multiple portfolio companies, you want a single customer master).
IT Service Desk and Support. Centralise basic IT support: password resets, laptop provisioning, access requests, and basic troubleshooting. Use a ticketing system (Jira Service Desk, ServiceNow, or custom) with AI-powered triage and self-service. Expect to resolve 40–50% of tickets without human intervention.
Vendor Management and Contract Lifecycle. Centralise vendor onboarding, contract negotiation, renewal tracking, and offboarding. Use a contract management system (Ironclad, Airtable, or custom) with AI-powered clause extraction and renewal alerts.
Compliance and Risk Management. Centralise compliance calendars, audit schedules, policy management, and risk registers. Use a compliance platform (Domo, Alteryx, or custom) with AI-powered anomaly detection and risk scoring.
The Operations Automation Stack
To automate operations across your portfolio, you need:
A Workflow Automation Platform. Zapier, Make, or a custom solution built on your data platform. This is the glue that connects your HR system, finance system, customer platform, and IT systems. When a new employee is hired in your HR system, the workflow automatically creates an email account, provisions a laptop, and enrols benefits.
AI-Powered Triage and Routing. Tools like Workato or custom models that classify incoming requests (expense, access request, support ticket) and route them to the right team or process. This is where AI shines: it can understand natural language, extract intent, and route 80%+ of requests automatically.
RPA (Robotic Process Automation). Tools like UiPath or Blue Prism for automating legacy system interactions. If one of your portfolio companies uses an old system that doesn’t have APIs, RPA can automate data entry and processing.
AI Agents for Customer and Vendor Interactions. Tools like Intercom, Drift, or custom models for handling customer inquiries, vendor requests, and employee questions. These agents can answer 60–70% of routine questions without human intervention.
The Operations Automation Savings Model
Let’s assume your portfolio has 500 employees across ten companies. Each company has 0.5 FTEs dedicated to HR operations (onboarding, offboarding, benefits administration). Total: 5 FTEs at $80K each = $400K annually.
With centralised, AI-driven operations automation, you can consolidate to 1–1.5 FTEs. Cost: $80–120K. Savings: $280–320K (70–80%).
But the real savings come from:
- Faster onboarding: New employees are productive 3 days faster. At 50 new hires per year, that’s 150 days of productivity freed up.
- Fewer support tickets: AI-powered self-service resolves 40–50% of IT support tickets without human intervention. At 10 tickets per employee per year (5,000 tickets total), that’s 2,000–2,500 tickets resolved automatically. At $50 per ticket to resolve, that’s $100–125K in labour savings.
- Better compliance: Centralised compliance management catches policy violations earlier. Expect to reduce compliance incidents by 30–40%.
- Better vendor terms: Centralised vendor management gives you visibility into contracts and lets you consolidate volume. Expect 5–10% better terms.
Total first-year savings: $400–500K. Payback is typically 3–6 months.
Building Your Shared-Services Technology Stack
Now that you understand what to centralise and automate, let’s talk about the technology. ACCA’s technical article on shared service centres covers the operating model design; here’s the technology design.
The Core Data Platform
Everything starts with data. You need a central data warehouse that ingests data from all your portfolio companies’ systems:
- Finance systems (QuickBooks, Xero, NetSuite, SAP)
- HR systems (Workday, BambooHR, Guidepoint)
- Customer platforms (Salesforce, HubSpot, custom)
- Operational systems (Slack, email, ticketing systems)
Your options:
Snowflake or BigQuery. Cloud data warehouses with strong AI/ML integrations. Snowflake is particularly good if you need to handle multi-tenant data (each portfolio company’s data isolated but queryable together). BigQuery is strong if you’re already in the Google Cloud ecosystem.
Databricks. If you need advanced AI/ML capabilities (LLMs, fine-tuned models), Databricks is excellent. It integrates well with open-source tools and gives you flexibility.
Custom Data Platform. If you have specific requirements (real-time streaming, complex data lineage, specific compliance needs), you might build a custom platform using PostgreSQL, Kafka, and dbt. This is more work but gives you full control.
For most portfolios, Snowflake or BigQuery is the right choice. Cost: $1K–5K per month depending on data volume.
The Finance and Operations Systems
You need systems for finance and operations transactions:
Finance: NetSuite is the gold standard for multi-entity finance (it’s designed for holding companies and portfolios). Xero is good for smaller portfolios. QuickBooks is fine for individual companies but doesn’t consolidate well. Cost: $1K–5K per month depending on company size.
HR: Workday is excellent for larger portfolios (100+ employees). BambooHR is good for smaller portfolios. ADP is solid but less flexible. Cost: $500–2K per month.
Customer: Salesforce is the standard. HubSpot is good for smaller portfolios. Cost: $500–2K per month.
The Automation and AI Layer
On top of your data platform and transaction systems, you need automation and AI:
Workflow Automation: Zapier, Make, or a custom solution. This orchestrates workflows across your systems. Cost: $500–2K per month.
Invoice Processing AI: Coupa, Ariba, or a custom model. Cost: $2K–10K per month depending on invoice volume.
Reconciliation AI: Blackline or a custom model. Cost: $2K–5K per month.
Forecasting and Scenario AI: Anaplan or a custom model. Cost: $2K–5K per month.
Customer/Vendor AI: Intercom, Drift, or a custom model. Cost: $500–2K per month.
Building vs. Buying
For most organisations, the answer is: buy the core systems (finance, HR, CRM), build the glue (workflows, data pipelines), and selectively build AI where you have unique requirements.
However, if you have a large portfolio with unique requirements, building custom AI models for invoice processing, reconciliation, and forecasting can be worth it. The ROI is typically 6–12 months.
If you’re considering building custom AI, PADISO’s AI Quickstart Audit is a useful diagnostic. It’s a fixed-fee, two-week assessment that tells you what to ship first, what to retire, and what 90 days could unlock. This is especially valuable if you’re building a shared-services AI platform across a portfolio.
Alternatively, if you need help designing and building your shared-services technology stack, PADISO’s Platform Design & Engineering service can help. We’ve built shared-services platforms for PE firms and multi-entity startups; we understand the data architecture, the AI requirements, and the governance model needed to make it work at scale.
Governance, Data, and Security at Scale
Once you’ve centralised finance, operations, and data, you need governance, data quality, and security. This is where most shared-services projects fail.
Data Governance
Centralised data is only useful if it’s clean, consistent, and trustworthy. You need:
Data Ownership. Each portfolio company should own their own data (GL accounts, customer records, employee records). The shared-services centre should own the consolidated data and the data quality rules. Make this explicit: document who owns what, who can access what, and who can change what.
Data Quality Rules. Define rules for what constitutes “good” data: required fields, valid values, format standards. Enforce these rules at the point of entry (in your finance and HR systems) and in your data warehouse (reject bad data, flag anomalies).
Data Lineage and Auditing. Track where data comes from, how it’s transformed, and where it goes. This is critical for compliance and debugging. Use tools like dbt, Lineage, or custom data lineage tracking.
Data Retention and Deletion. Define how long you keep data and when you delete it. This is critical for privacy (GDPR, CCPA) and compliance.
Security and Compliance
Centralised data is a security and compliance risk. You need:
Access Control. Use role-based access control (RBAC) to ensure people only see data they need. Your finance team shouldn’t see employee salaries. Your HR team shouldn’t see customer data. Use your data platform’s native RBAC (Snowflake’s dynamic data masking, BigQuery’s column-level access control).
Encryption. Encrypt data in transit (TLS) and at rest. Your data platform should support this natively.
Audit Logging. Log all access to sensitive data. Who accessed what, when, and why. Use your data platform’s native audit logging.
Compliance Frameworks. Depending on your portfolio, you may need SOC 2, ISO 27001, GDPR, HIPAA, or industry-specific compliance (APRA for financial services, ASIC for funds). Design your shared-services platform with compliance in mind from day one.
If you’re serious about compliance, PADISO’s Security Audit service helps you get SOC 2 and ISO 27001 audit-ready in weeks, not months. We work with Vanta to automate compliance evidence collection; this is especially valuable if you’re centralising data and systems across a portfolio.
Alternatively, if you’re in financial services, PADISO’s AI for Financial Services service covers APRA, ASIC, and AUSTRAC compliance by design. We help Australian banks, wealth managers, and fintechs build AI-driven shared services that are compliant from day one.
Change Management and Adoption
Technology is only half the battle. The other half is getting your portfolio companies to use it. This is where most shared-services projects fail.
Executive Sponsorship. Your holding company CEO or PE sponsor needs to be visibly committed to shared services. This sends a signal that it’s not optional.
Clear Value Proposition. Each portfolio company should understand what they gain: faster close, better visibility, fewer manual tasks. Don’t lead with “we’re centralising to save money.” Lead with “you get better financial reporting and faster decision-making.”
Training and Support. Your shared-services centre should provide training and support to portfolio company teams. This is not a one-time event; it’s ongoing.
Feedback Loops. Regularly solicit feedback from portfolio companies. What’s working? What’s not? What would make your life easier? Use this feedback to improve the shared-services platform.
The Savings Model: What Gets Cheaper, What Doesn’t
Let’s be concrete about what shared services AI actually saves.
What Gets Cheaper
Labour in Finance and Operations. This is the obvious one. Centralising and automating finance and ops labour saves 60–75% of the labour cost. If your portfolio spends $2M on finance and ops labour, you can cut that to $500K–800K.
Transaction Processing. AP processing, AR collection, payroll, GL reconciliation. These are the high-volume, low-complexity tasks that AI is best at automating. Expect 70–85% reduction in transaction processing labour.
Vendor Costs. Centralised procurement gives you visibility into spend and lets you consolidate volume. Expect 5–10% better terms on repeat vendors.
Compliance and Audit Costs. Centralised compliance management and audit-ready documentation reduce the cost of audits and regulatory filings. Expect 20–30% reduction.
Working Capital. Faster close cycles and better cash collection free up working capital. This is real money: at $1M in monthly expenses, a 7-day improvement in close cycle releases $230K in working capital.
What Doesn’t Get Cheaper
Tooling and Infrastructure. You’re adding systems (data warehouse, workflow automation, AI tools). This costs money. Budget $50K–200K for initial setup, and $20K–50K per month for ongoing costs.
Shared-Services Team. You’re still paying people. You’ve just consolidated them. The shared-services centre needs skilled people who understand finance, operations, and technology. Expect to pay market rates: $80K–150K for processors, $120K–200K for analysts, $150K–300K for directors.
Change Management and Training. Getting your portfolio companies to use the shared-services centre takes time and money. Budget 10–20% of your savings for change management and training.
Data Integration and Cleaning. Integrating data from ten different systems and cleaning it up takes time. Budget 2–4 months and $20K–50K for data integration.
The Net Savings Model
Let’s work through a realistic example. You have a portfolio of ten companies, each spending $150K on finance labour and $50K on operations labour. Total: $2M annually.
Centralised labour cost: 5 FTEs (finance) + 1.5 FTEs (operations) = 6.5 FTEs at $100K average = $650K.
Tooling and infrastructure: $30K setup + $25K per month = $330K annually.
Change management and training: $100K.
Total shared-services cost: $1.08M.
Savings from labour consolidation: $2M − $650K = $1.35M.
Savings from faster close and better cash collection: $250K (conservative estimate).
Savings from better vendor terms: $100K (conservative estimate).
Total gross savings: $1.7M.
Net savings (gross savings − shared-services cost): $1.7M − $1.08M = $620K.
Payback period: 6–9 months.
This is conservative. Many organisations see higher savings (especially if they have poor vendor terms or slow close cycles). But it shows the model: shared services AI typically pays for itself in 6–12 months and delivers $500K–$2M in annual net savings for a portfolio of ten mid-sized companies.
Implementation Roadmap: From Pilot to Scale
Now that you understand the opportunity, how do you actually build it? Here’s a realistic roadmap.
Phase 1: Assess and Plan (Weeks 1–4)
Week 1: Inventory Current State. For each portfolio company, document:
- Finance labour (headcount, cost, processes)
- Operations labour (headcount, cost, processes)
- Systems used (finance, HR, CRM, others)
- Data architecture (where does data live, how is it integrated)
- Compliance requirements (SOC 2, ISO 27001, GDPR, industry-specific)
Week 2: Identify Centralisation Opportunities. For each function (AP, AR, payroll, GL, HR onboarding, etc.), score:
- Standardisation: How similar is this process across portfolio companies? (1–5 scale)
- Volume: How many transactions per month? (high volume = higher ROI)
- Complexity: How many exceptions and manual steps? (high exceptions = lower ROI for AI)
- Savings Potential: How much labour could be saved? (estimate in hours/month)
Focus on high-standardisation, high-volume, low-complexity processes first.
Week 3: Design the Target Operating Model. Document:
- What will be centralised and automated
- What will stay distributed
- What systems you’ll use (data warehouse, finance, HR, workflow automation, AI)
- Governance model (who owns what, who can access what)
- Reporting and KPIs
Week 4: Build the Business Case. Document:
- Current state costs (labour, systems, vendor costs)
- Target state costs (labour, systems, tooling)
- Savings (labour, transaction processing, vendor terms, working capital)
- Payback period and ROI
- Implementation timeline and costs
If you want help with this assessment, PADISO’s AI Quickstart Audit is designed for exactly this: a fixed-fee, two-week diagnostic that tells you where you are, what to ship first, and what 90 days could unlock.
Phase 2: Pilot (Weeks 5–12)
Start with one function and one portfolio company. Accounts Payable is usually the best starting point: high volume, high standardisation, high automation potential.
Week 5–6: Set Up Infrastructure. Spin up your data warehouse (Snowflake or BigQuery), set up your finance system (NetSuite or Xero if you don’t have one), and set up invoice processing AI (Coupa or custom).
Week 7–8: Integrate Data. Connect your pilot portfolio company’s finance system to your data warehouse. Get clean GL data, AP data, and AR data flowing in.
Week 9–10: Configure Invoice Processing. Set up invoice processing AI. Train it on historical invoices from your pilot company. Aim for 85%+ accuracy on first-pass processing.
Week 11–12: Pilot with Real Volume. Process real invoices through the shared-services centre. Measure:
- Accuracy (% of invoices processed correctly on first pass)
- Throughput (invoices processed per FTE per day)
- Cycle time (days from invoice receipt to payment)
- Cost per invoice
Target metrics:
- 85%+ accuracy
- 3x throughput improvement (from 20 invoices/day to 60 invoices/day)
- 2x cycle time improvement (from 5 days to 2.5 days)
- 60%+ cost reduction per invoice
Phase 3: Expand (Weeks 13–24)
Once you’ve proven the model in AP, expand to other functions and portfolio companies.
Weeks 13–16: Add AR and Payroll. Set up AR and payroll processing in the shared-services centre. Integrate data from the pilot company. Aim for similar metrics: 85%+ accuracy, 3x throughput improvement, 60%+ cost reduction.
Weeks 17–20: Onboard Second Portfolio Company. Bring a second portfolio company into the shared-services model. Start with AP, AR, and payroll. Reuse the infrastructure and processes from the pilot.
Weeks 21–24: Add GL Reconciliation and Close. Set up automated GL reconciliation and close-cycle compression. Aim to compress close from 10 days to 3 days.
Phase 4: Scale (Months 7–12)
Once you’ve proven the model across multiple companies and functions, scale to your entire portfolio.
Months 7–9: Onboard Remaining Portfolio Companies. Bring remaining portfolio companies into the shared-services model. Reuse infrastructure and processes. This should be relatively fast (4–8 weeks per company) since you’ve already done the hard work of integration and configuration.
Months 10–12: Add Operations Automation. Expand beyond finance to operations: HR onboarding, IT support, vendor management, compliance. Reuse the workflow automation platform and AI tools you’ve built.
Phase 5: Optimise (Ongoing)
Once you’ve scaled, continuously optimise:
Monthly: Review key metrics (accuracy, throughput, cost, cycle time). Identify bottlenecks and opportunities for improvement.
Quarterly: Review portfolio company feedback. What’s working? What’s not? What would make their life easier?
Annually: Review AI model performance. Retrain models on new data. Add new AI capabilities (forecasting, anomaly detection, scenario modelling).
Common Pitfalls and How to Avoid Them
Shared services AI is powerful, but it’s easy to get wrong. Here are the common pitfalls and how to avoid them.
Pitfall 1: Centralising the Wrong Things
The Problem: You centralise hiring decisions, capital allocation, or product pricing. Now your portfolio companies are frustrated because they can’t make decisions without the shared-services centre’s approval.
The Fix: Be clear about what you’re centralising (transaction processing, data, infrastructure) and what you’re keeping local (decisions, strategy, relationships). Centralise processes, not decisions.
Pitfall 2: Poor Data Quality
The Problem: You centralise finance data, but it’s dirty: missing GL codes, duplicate vendors, inconsistent formats. Your AI models are garbage in, garbage out. Your portfolio companies don’t trust the data.
The Fix: Invest in data quality from day one. Define data quality rules. Enforce them at the point of entry. Use data validation and anomaly detection to catch bad data before it enters your data warehouse. Expect to spend 20–30% of your shared-services effort on data quality.
Pitfall 3: Insufficient Change Management
The Problem: You build a beautiful shared-services platform, but your portfolio companies don’t use it. They continue using their old processes because they don’t understand the new system or they don’t see the value.
The Fix: Invest heavily in change management. Get executive sponsorship. Communicate the value proposition clearly. Provide training and support. Collect feedback and iterate. Expect change management to take 3–6 months for each portfolio company.
Pitfall 4: Underestimating Implementation Complexity
The Problem: You thought you’d centralise AP in 8 weeks. Six months later, you’re still integrating data and configuring workflows.
The Fix: Pad your timeline. Data integration and configuration always take longer than you think. Budget 2–4 months for initial setup, and plan for 4–8 weeks per portfolio company for onboarding. Start with a pilot and learn before you scale.
Pitfall 5: Over-Automating
The Problem: You try to automate everything. You build AI models for every function. But your AI models are only 70% accurate, and you end up with more exceptions and manual work, not less.
The Fix: Start with high-volume, low-complexity processes where AI can achieve 90%+ accuracy (AP, AR, payroll). Leave exceptions for humans. Don’t try to automate your way to zero humans; automate your way to smaller, more skilled teams that focus on exceptions and strategy.
Pitfall 6: Ignoring Compliance and Security
The Problem: You centralise financial data and customer data, but you don’t think about compliance. Then you get audited and realise you don’t have the access controls, audit trails, or encryption you need.
The Fix: Design compliance and security into your shared-services platform from day one. Use CFO’s article on how shared services can move finance toward higher-value activities as a starting point, but layer on security and compliance. Use your data platform’s native access controls, encryption, and audit logging. Get SOC 2 or ISO 27001 certified if you’re handling sensitive data.
Next Steps: Getting Started with Your Portfolio
If you’re running a portfolio and you’re serious about shared services AI, here’s what to do next.
Step 1: Assess Your Current State
Inventory your portfolio companies’ finance and operations costs, systems, and processes. Identify the top 3–5 centralisation opportunities (the ones with the highest standardisation and volume).
If you want a structured assessment, PADISO’s AI Quickstart Audit is a fixed-fee, two-week diagnostic. We’ll tell you where you are, what to ship first, what to retire, and what 90 days could unlock. It’s AU$10K and comes with a clear roadmap.
Step 2: Build the Business Case
Estimate the costs and savings of shared services AI for your portfolio. Use the models in this guide as a starting point. What’s your payback period? What’s your three-year NPV?
If you need help building the business case or designing the technology stack, PADISO’s Services include fractional CTO leadership, platform design and engineering, and AI strategy. We can help you design a shared-services platform that works for your specific portfolio.
Step 3: Start with a Pilot
Don’t try to centralise everything at once. Pick one function (AP is usually best), pick one portfolio company, and run a 12-week pilot. Measure accuracy, throughput, cycle time, and cost. If it works, expand to other functions and companies.
Step 4: Build Your Shared-Services Team
You need a small team to run shared services: 1 director (strategy and vendor management), 2–3 processors (transaction processing), 1 analyst (data quality and reporting). If you’re in a regulated industry, add a compliance person.
If you need help hiring or want fractional CTO leadership for your shared-services platform, PADISO’s Fractional CTO services are available in Sydney, Melbourne, and other cities. We can help you design the team, build the technology, and execute the roadmap.
Step 5: Invest in Technology
Budget for a data warehouse (Snowflake or BigQuery), finance and HR systems (NetSuite, Workday), workflow automation (Zapier or Make), and AI tools (invoice processing, reconciliation, forecasting). Total first-year cost: $50K–200K. Ongoing cost: $20K–50K per month.
If you need help selecting, implementing, or building custom AI tools, PADISO’s Platform Development services cover data platforms, multi-tenant SaaS, and AI automation. We’ve built shared-services platforms for PE firms and multi-entity startups.
Step 6: Plan for Compliance
Depending on your portfolio, you may need SOC 2, ISO 27001, GDPR, or industry-specific compliance. Design your shared-services platform with compliance in mind from day one.
If you’re in financial services, PADISO’s AI for Financial Services covers APRA, ASIC, and AUSTRAC compliance by design. If you need SOC 2 or ISO 27001, PADISO’s Security Audit service helps you get audit-ready in weeks via Vanta.
Conclusion
Shared services AI is one of the highest-ROI investments a portfolio can make. By centralising finance and operations, standardising processes, and automating with AI, you can save 40–60% on labour costs, compress your close cycle from 10 days to 3, improve cash flow, and free up your portfolio companies to focus on growth.
The key is to start small (one function, one company), prove the model, and scale. Invest in data quality and governance from day one. Don’t try to automate everything; automate the high-volume, low-complexity processes where AI can achieve 90%+ accuracy. And invest in change management: technology is only half the battle.
If you’re ready to build shared services AI for your portfolio, PADISO can help. We’re a Sydney-based venture studio and AI digital agency that partners with ambitious teams to ship AI products, automate operations, and build platforms that scale. We’ve helped PE firms, venture studios, and multi-entity startups centralise finance and operations with AI. Get in touch to discuss your portfolio.