HY

Hypatos Financial Model

• Deep learning automation of semi-structured document processing across three service layers:

1. Document Understanding - RNN + CNN models for data extraction, classification, splitting.
2. Content Validation - heuristics + ML to flag non-compliant/inconsistent data.
3. Document Enrichment - ML to predict processing attributes (e.g., GL account coding, workflow routing).

• Hypatos Studio - ML pipeline software for clients to annotate data and fine-tune models on proprietary datasets; available cloud and on-premise.
• Delivery via API + system integration components (user/license management, document collectors, human-in-the-loop UI, monitoring, system connectors); multi-cloud, private cloud, and on-premise deployment options.
• No template/rule engineering required; confidence scores surfaced to operators.

• TAM: ~$2 trillion total annual spending on back-office business processes globally (2019).
• Methodology footnote on slide 4: based on outsourcing market ($195bn; Grandviewresearch 2017), × share of companies outsourcing HR/Finance/Procurement/Tax (50%; Deloitte 2016) × assumption on share of outsourced processes (25%), extrapolated to total spending.
• Donut chart shows ~roughly 55/45 split between industry-agnostic (A/P, O2C, T&E, Tax, Recruitment) vs industry-specific processes (mortgage/loan processing, insurance claims, medical prescriptions).

• Model: B2B software/API services. Revenue most likely structured as subscription + volume-based licensing (per-document or per-transaction), though explicit pricing is not stated in the deck.
• Channels: Direct enterprise sales (delivery team with pre-sales, solution architects, technical support called out in use of funds).
• Integration pattern: Embedded into existing ERP/ECM/RPA/CRM stacks - not a standalone workflow tool; implies multi-year enterprise contracts.
• Hypatos Studio likely sold as separate licensed software (cloud or on-premise), potentially with a training/implementation services component.
• No ARR, ACV, contract values, pricing tiers, or revenue figures disclosed.

• No revenue figures, ARR, customer count, or growth rates disclosed in the deck.
• Client logos referenced as blurred/anonymized "example clients" on the use-case matrix (slide 8) - identities redacted in the image.
• Focus use cases with named example clients implied across Finance (P2P & O2C) and HR (T&E, Payroll, HR Admin) as of 2020; Financial Services, Insurance, Public Admin, Logistics flagged as 2021+ expansion.
• Team stat: 40+ data scientists, engineers, and delivery professionals in Berlin and Warsaw.
• Team stat: 150 years cumulative tech work experience.

• Competitive framing: positions against RPA (rule-based, simple tasks) vs ML (complex human-understanding tasks). Hypatos occupies the ML/deep learning tier.
• Moat claims:
• No templates or manual rules required (vs legacy OCR/RPA competitors).
• High accuracy with transparent confidence indicators.
• High adaptability to long-tail and edge cases.
• Hypatos Studio enables client-specific model fine-tuning on proprietary data (creates data moat per client).

Team:

• Dr. Uli Erxleben - Founder & MD; ex-McKinsey (Berlin & Palo Alto), MD Rocket Internet North America, MD ProSiebenSat1 corporate venturing, serial entrepreneur.
• Cem Dilmegani - CCO; ex-McKinsey, telco integration, Solon, founder of AIMultiple.
• He Zhang, PhD - VP Machine Learning; ex-Head of Data Science Lesara, Lead Data Scientist HelloFresh, theoretical physicist at Max-Planck-Institut.
• 40+ person team across Berlin and Warsaw.

Funding ask:

• EUR ~10mn round.
• Runway: 24 months.
• Use of funds (four buckets):

1. Machine Learning - additional data scientists, ML pipeline automation, new document use cases and international models.
2. Engineering - team build-up, Human-in-the-Loop toolset, ERP/CMS/RPA integrations.
3. Delivery Team - pre-sales, solution architects, technical support.
4. Go-to-Market - marketing manager, sales reps, PR, content.

Archetype + why: B2B SaaS / AI-as-a-Service ARR model with a volume overlay. Hypatos sells recurring software licenses (API services + Studio) to enterprises on multi-year contracts. Revenue likely has two components: (a) base subscription/platform fee and (b) consumption-based volume fee (documents processed). A standard SaaS ARR build is the right spine, with a document-volume driver layered on top to model the consumption component. No M&A or SPAC indicators.

Forecast horizon & granularity: Monthly for Year 1 (to track burn vs. the 24-month EUR 10mn runway), then annual for Years 2–5. 5-year total horizon.

Key drivers & assumptions:

• New logos signed per quarter - 2–4 new enterprise clients/quarter in 2021, ramping to 6–8 by 2023; rationale: early-stage enterprise sales cycle is long (6–12 months), team is still being built.
• Average ACV (annual contract value) - EUR 100–200k/year per client for a mid-market/large enterprise A/P automation deployment; rationale: comparable enterprise document-AI deals; no deck data available.
• Document volume per client per month - 50,000–500,000 documents depending on client size; rationale: enterprise A/P departments process tens of thousands of invoices monthly.
• Volume-based fee per document (above base tier) - EUR 0.01–0.05/document; rationale: market benchmark for invoice-processing automation.
• Net Revenue Retention (NRR) - 110–120%; rationale: expansion likely as clients add use cases (e.g., starting with A/P, expanding to O2C, T&E); Studio upsell.
• Gross churn - 5–10% annually; rationale: sticky enterprise integrations but early-stage product risk.
• Gross margin - 65–75%; rationale: cloud hosting + ML compute costs for inference; Studio on-premise deployments may have lower margin.
• Sales cycle / ramp - 6-month average sales cycle; new client ARR recognized from month 7 post-pipeline entry.
• Headcount ramp: EUR 10mn / 24 months = ~EUR 417k/month total burn budget; allocate ~60% to personnel (ML + Engineering + Delivery + GTM), ~40% to infra, cloud, G&A.
• Average fully-loaded salary - EUR 80–100k/year blended (Berlin/Warsaw mix); rationale: Warsaw engineering salaries are ~40–50% of Berlin, blended team.
• Hypatos Studio licensing - separate line at EUR 30–50k per client annually; on-premise deployment adds professional services revenue (~EUR 20–40k one-time).
• Initial geography: DACH + UK; 2021+ expansion to broader EU.

Scenarios (Base / Bull / Bear - which variables flex):

• Base: 3 new logos/quarter ramp, EUR 130k ACV, 65% gross margin, NRR 110%.
• Bull: 5 new logos/quarter, EUR 180k ACV, 72% gross margin, NRR 125%; Studio adoption accelerates; 2021 vertical expansion hits on schedule.
• Bear: 1–2 new logos/quarter (long sales cycles, COVID-19 budget freezes plausible given June 2020 date), EUR 90k ACV, 60% gross margin; runway pressure by month 18.

Required sheets / outputs:

1. Assumptions - all drivers on one sheet, color-coded vs.
2. Revenue Build - monthly logo additions × ACV + volume fee overlay; ARR waterfall (new, expansion, churn).
3. Headcount Plan - by function (ML, Engineering, Delivery, GTM, G&A); monthly hire schedule vs. EUR 10mn budget.
4. P&L - Revenue, COGS (cloud/compute), Gross Profit, OpEx by function, EBITDA.
5. Cash Flow & Runway - monthly burn, cumulative cash, runway months from EUR 10mn raise.
6. Scenario Toggle - Base/Bull/Bear switcher feeding all sheets.
7. KPI Dashboard - ARR, MRR, Logo Count, NRR, Gross Margin %, Monthly Burn, Runway.