This document walks through how the 5 pilot companies are selected, what gets automated, the pilot timeline, the impact report you receive after 90 days, and how the model scales to the rest of the portfolio.
Back-Office Automation for Portfolio Companies
Per-Company Approach
Two of our consultants interview Finance, HR, IT, Procurement leads. Map top 20 repetitive processes. Score by volume, FTE cost, and automation feasibility.
Build the top 5-8 automations: AI agents, workflow automation, data pipelines. Mostly remote with check-ins every 3-4 days.
Track FTE hours saved, error rates, cycle times. Deliver an impact report with hard numbers, side-by-side with the baseline we measured in week 1.
What the Impact Report Looks Like
A 6-8 page document per company. Process-by-process: hours saved per week, error rate before and after, cycle time before and after, qualitative feedback from the people now using the automation. One consolidated portfolio summary page. Numbers, not narrative.
5 Companies in Parallel
Stagger starts by 1-2 weeks. All 5 complete within 12 weeks. Your time per company: one 30-minute intro call to connect us with the CFO/COO. We handle everything directly with portfolio company management.
Risks and Mitigations
Five things that can derail this kind of program. We have hit each of them on prior engagements and have a working response.
| Risk | Mitigation |
|---|---|
| Portfolio company management resists | Your warm intro to the CFO/COO. We frame the pilot as "free help, free numbers, no commitment." We have not had a portfolio company refuse a free pilot. |
| Data access blocked by IT or security review | Every automation runs on read-only data initially. No writes until reviewed. Most portfolio companies clear this in days, not weeks. We bring our own security documentation. |
| Discovery finds nothing repetitive enough to automate | Has not happened yet. Every back office above 50 FTE has 5-8 obvious targets. If discovery genuinely finds nothing, we drop that company and replace it. You only pay for completed pilots. |
| Pilot results look great in week 12 but adoption decays | Each automation includes a 30-day follow-up after the impact report. We measure actual usage, not theoretical capacity. Adoption issues surface and get fixed before the report is final. |
| Scope creep: portfolio company asks for more during the pilot | Common and a good signal. We capture the requests in a backlog for Phase 2 but do not expand the pilot scope. The pilot has to finish on time to produce credible numbers. |
What We Need From You
Beyond the 5 hours of intro calls, the pilot needs three things from your side: (1) a warm intro per portfolio company (one short email per CEO/CFO is enough), (2) one named sponsor inside your operating team to be our point of contact for status and escalations (roughly 1-2 hours per week), and (3) a willingness to share the resulting impact reports internally so the case for Phase 2 lands with the right people. Everything else we do directly with portfolio company management.
What Happens After 90 Days
You hold the impact report and a clear decision: continue with us into Phase 2, hand the playbook to an internal team, or stop. The pilot is a discrete commitment, not a contract trap. We design it that way intentionally.
Portfolio Archetypes
Infrastructure companies cluster into operational archetypes that predict their back-office patterns. The same invoice processing automation looks different at a data center versus a port terminal. Getting this right up front determines whether pilot results scale or stay local.
| Archetype | Examples | Back-Office Character | Highest-Value Automation |
|---|---|---|---|
| Asset-heavy, low headcount | Data centers, renewables, fiber networks | Small teams, often already digital. Back-office is a high % of total staff. | Financial reporting, vendor management, compliance |
| Asset-heavy, high headcount | Ports, terminals, airports | Large operational workforce. HR/payroll is massive (shifts, safety, training). | HR/payroll, procurement, maintenance scheduling |
| Service/operations-intensive | Waste management, facility services | Distributed operations, many small sites. | Scheduling, invoicing, customer service |
| Regulated utilities | Gas, electricity, water | Heavy regulatory overlay. Unionized workforce in many geographies. | Regulatory reporting, customer billing, asset management |
| Project-based | Infrastructure development, construction | Project finance, contractor management. Fewer recurring processes. | Project accounting, procurement, contractor compliance |
Our Recommendation: Cover 3-4 Archetypes
You will know your portfolio better than this framework, but our recommendation is to spread the 5 across 3-4 archetypes rather than concentrating on one sector. The reason is template reuse: the more variety in the pilot, the broader the rest of the portfolio that the templates work for. A starter shape that has worked elsewhere:
| Slot | Archetype | Purpose | What It Proves |
|---|---|---|---|
| 1 | Asset-heavy, low headcount | Quick win. Small team, modern systems, fast results. | Reporting and vendor management templates that are reusable across many similar companies. |
| 2 | Asset-heavy, high headcount | Largest absolute FTE savings. | HR automation at scale. Templates for ports, terminals, airports. |
| 3 | Service/operations-intensive | Tests distributed operations pattern. | Scheduling and billing automation for distributed sites. |
| 4 | Regulated utility | Tests regulatory constraints early. | What is possible within regulated environments. Avoids surprises at scale. |
| 5 | Largest back-office in your selection | Maximum absolute savings. | The headline ROI figure for the portfolio-wide investment case. |
This is a starting point, not a prescription. We expect to revise it together once we see the actual list of candidates.
Selection Criteria
Within each archetype, score candidates on five dimensions:
| Criterion | What It Means | Why It Matters |
|---|---|---|
| Back-office scale | Total FTE in Finance, HR, IT, Procurement | Drives absolute savings. 1,000 FTE vs. 100 FTE = 10x different impact. |
| Process standardization | Modern ERP/finance systems vs. fragmented legacy | Companies on SAP, Oracle, Workday are dramatically faster to automate. |
| Management alignment | Local CEO/CFO willingness to engage | Cooperative management = 2x faster execution. Resistance kills pilot timelines. |
| Data accessibility | Digital systems with API access vs. paper-based | Determines what can be automated in weeks vs. what needs months of prep. |
| Sector coverage | How many other portfolio companies share this sector | Templates built for one water utility work at another. Sector coverage drives reuse. |
The Selection Process
Plot 190 companies on back-office scale vs. automation readiness. Identify top 20 candidates.
Score top 20 on all 5 criteria. Select 5 covering 3-4 archetypes.
Confirm willingness, data access, no active blockers (restructuring, system migration).
Template Reuse: How 5 Becomes 190
The pilot produces 25-40 automation templates (5-8 per company). Each template is tagged by function, process type, and system requirements. When scaling to additional companies:
- Most automations at a new company are reconfigurations of existing templates rather than ground-up builds
- The remainder require new development for company-specific edge cases
- Each subsequent company is meaningfully cheaper and faster than a pilot company
- The pilot produces the actual reuse rate. The 60-70% range that gets cited in the industry is directional, not yet our verified number for your portfolio.
This is the real answer to scaling. Not linear deployment. Template reuse across archetypes.
Back-Office Process Landscape
A typical portfolio company (5,000 employees) has roughly 400-800 people in back-office functions (Finance, HR, IT, Procurement). That is 8-16% of total headcount. Your target of 20% reduction in these functions means eliminating 80-160 FTE per company.
| Function | Typical Back-Office FTE (illustrative) | High-Automation Processes | Pilot Coverage |
|---|---|---|---|
| Finance | 120-250 | Invoice processing, expense reconciliation, month-end close, variance reporting, intercompany transfers | 2 processes |
| HR | 80-150 | Onboarding workflows, leave management, compliance tracking, headcount reporting | 1-2 processes |
| IT | 100-200 | Ticket triage, access provisioning, asset management, status reporting | 1 process |
| Procurement | 60-120 | PO processing, vendor onboarding, contract extraction, spend analysis | 1 process |
| Reporting | 40-80 | CFO dashboards, board deck prep, KPI aggregation, LP reports | 1 process |
FTE ranges are illustrative based on infrastructure benchmarks for companies in this size range. Actual numbers vary significantly. The pilot replaces these with measured baselines per company. The pilot total per company is 5-8 processes, sequenced to hit the highest-value ones first.
What the Pilot Actually Delivers
The Path from 1-2% to 20%
Reaching 20% requires systematically working through all back-office process clusters across each function. Each cluster follows the same discover-build-measure cycle. Based on pilot results, you can model the full program: how many clusters per company, what team is needed, and what the realistic timeline looks like. The pilot gives you the unit economics to make that decision.
| Phase | Scope | Aspirational Reduction |
|---|---|---|
| Pilot (Months 1-3) | 5-8 processes per company | 1-2% of back-office FTE |
| Phase 2 (Months 4-9) | 10-15 additional processes per company | 5-8% of back-office FTE |
| Phase 3 (Months 10-18) | 25-40 cumulative processes per company | 12-20% of back-office FTE |
Phase 2 and Phase 3 numbers are aspirational ceilings, not commitments. They depend on pilot results, change management capacity at each company, and your appetite for the investment. The pilot replaces these with real unit economics.
Real-Time Visibility as a Byproduct
Once back-office processes run on automated pipelines, the data they generate becomes available in real time. The same instrumentation that powers automation also produces live dashboards: month-end variance, headcount, cash, capex, and EBITDA trend at the portfolio-company level, refreshed nightly instead of quarterly with a 4-6 week lag.
| Today | With Pilot Automation in Place |
|---|---|
| Quarterly PDFs from each company | Live dashboards updated nightly from source systems |
| 4-6 weeks lag from period close | Same-day visibility on the metrics that matter |
| Inconsistent format across companies | Standard templates with side-by-side comparability |
| Anomalies surface after the next board review | Anomalies surface in the daily report |
This is not a separate workstream. It is what falls out of the pilot automatically once the data plumbing is in place.
Runtime
Automations are built as a combination of AI agents (Claude / Anthropic API), workflow orchestration (Python services), and data pipelines that read directly from source systems (ERP, HRIS, ITSM, finance platforms). The runtime is hosted inside the portfolio company's own cloud environment by default. No data leaves their tenancy unless they explicitly configure it to.
Data Flow
Source systems → read-only connectors → processing layer (LLM + workflow engine) → outputs (automated actions, dashboards, reports). Writes back to source systems are gated behind a security review per process and only enabled after the portfolio company's IT and security teams have signed off.
Integration Touchpoints
| Layer | Typical Systems | Connection Method |
|---|---|---|
| ERP | SAP, Oracle, Infor LN, NetSuite, Microsoft Dynamics | Native APIs, ODBC, Snowflake/data warehouse where present |
| HRIS | Workday, SAP SuccessFactors, BambooHR | Native APIs |
| ITSM | ServiceNow, Jira Service Management, Freshservice | REST APIs |
| Finance | Concur, Coupa, Bill.com, native ERP modules | Native APIs, file-based fallback |
This section is a high-level orientation. A full architecture review (data residency, network topology, model provider choices, audit logging, observability) is part of the discovery phase per portfolio company.
Security Posture
- Data residency: Automations run inside the portfolio company's own cloud environment. Data does not transit our infrastructure by default.
- Read-only first: Every automation starts read-only. Write access is enabled per process only after security review.
- Model providers: Anthropic Claude is the default. The architecture is provider-neutral. Azure OpenAI, AWS Bedrock, or self-hosted open-weight models are supported alternatives.
- Contracts and compliance: NDA and DPA in place before discovery. SOC 2 posture and insurance documentation available on request.
Transferability (Avoiding Vendor Lock-In)
Every automation we build is designed to be handed over. The deliverables are:
- Source code in a Git repository the portfolio company owns
- Infrastructure-as-code definitions for the runtime environment
- Runbook covering deployment, monitoring, common failure modes, and recovery
- 30-day handover period with paired support if the portfolio company elects to take it in-house
After the pilot, the choice is open: continue with us, take it fully in-house using the playbook, or run a hybrid model. The decision belongs to the portfolio company, not us.
This section is a draft summary. Detailed security documentation (data flow diagrams, compliance posture, sub-processor list, incident response process) is available on request and reviewed per portfolio company during discovery.
What We Bring
Built on Our Own Architecture
The pilot automations we build for portfolio companies use the same patterns and infrastructure we run internally, hardened across 72 skills and 7 subagents. This document was generated by that system.
Operating Inside Industrial Companies
15+ years inside the operations of manufacturing and industrial businesses. The tooling, the people, the change-management resistance, the politics around back-office costs: infrastructure and manufacturing share more than they differ at this level. We have already worked through the patterns your portfolio companies will throw at us.
AWS Partner
AI applications built on AWS Bedrock. Cloud infrastructure, migration, and managed services capability for production-grade deployments.
Delivery Speed
Small firm, fast decisions. Discovery starts within 1 week of go. First portfolio automations ship within 6 weeks of kickoff.
Why Us vs the Alternatives
You have other choices. Here is the honest version of how we compare:
BLCG / H2W Labs
Operating partner mindset. We use AI inside our own business every day. Small enough to move in days, deep enough to build production systems. Senior people on every engagement.
Big 4 Consulting
Strong frameworks, slow start. 6-12 weeks before code ships. High day rates and pyramid staffing. Often hands off to a junior team after the kickoff slides.
RPA Vendors
UiPath, Automation Anywhere and similar. Strong tools but tool-first thinking. Great for narrow repetitive tasks, weaker for the language and judgment work that AI now handles. License-heavy commercial model.
In-House Build
Eventually the right answer at scale. Hard to start cold without a working playbook. The pilot exists to give you that playbook before you commit to hiring.
The meta-point: this entire document was generated by the system we are proposing. That is the most honest demo we can offer.
Your Core Team for This Engagement
Lennart Hector
Richard Weiher
Heiko Steinbach
Carlos Claure
This is the core team you would work with directly. Lennart owns the relationship. Richard owns the architecture. Heiko runs the multi-company coordination, which is the hardest part of a 5-company parallel pilot. Carlos and a handful of additional senior consultants from our DACH and U.S. teams handle the on-site discovery work. We do not staff junior people on engagements like this.
Selected References
Trojan Technologies
Multi-year engagement spanning training, ERP optimization, and system integration. Multiple expansion deals from a single entry point. Closest match to your utility and infrastructure portfolio profile.
Tsurumi Europe
Multi-country ERP support across DE, UK and FR operations. AI Product Assistant in production. Digitalization roadmap that reduced manual data lookups across the European service organization.
W. Neudorff
16 parallel ERP workstreams, multi-year engagement, ~10 consultants at peak. Full transformation across production, logistics, quality management, and integrations. Our largest engagement to date.
IMA Schelling
Multi-year ERP extensions, system integrations, and ongoing operational optimization. Account upgraded from project work to ongoing transformation partnership.
More Clients We Work With
30+ manufacturing and industrial clients across Germany, Austria, United States, Canada, and Japan.
Pilot Eligibility Review
A 60-minute call to walk through the candidate companies against the selection framework above, agree on the 5 pilots, and map out the discovery sequence. We come prepared with the framework; you bring the portfolio context. By the end of the call we have a shortlist and a starting date.
Optional: Technical Deep-Dive
A second 60-minute session with Richard Weiher (delivery lead) on automation patterns, data access models, and security review. Recommended for tech leads who want to validate the approach before warm-introducing the pilot to portfolio company management.
The Calendly link reserves a 30-minute holder. If you pick a session length above we will extend the booking on our side once you confirm. Or just reply to the WhatsApp thread with a date that works.
l.hector@blconsultingservice.com