The PDF is the finance department's silent antagonist. It arrives as an invoice, a contract, or a trade deduction claim, its data locked inside a static format that requires manual entry, cross-referencing, and reconciliation. For a team of three founders in Berlin, that friction is a wedge. Cloudsquid, launched in 2023, sells an AI platform that promises to extract and transform that unstructured data with 99%+ accuracy, automating workflows for finance and operations teams [Cloudsquid].
Filip Rejmus, the CPO and co-founder, saw the problem up close on Uber's data team, pulling figures from PDFs and CSV exports to understand daily operations [LinkedIn, 2026]. He teamed with Sangwoo Bae, a former AWS and Kubermatic engineer, and Mike McCarthy, an early revenue leader at customer service automation startup Ultimate AI. Their collective pitch to investors was infrastructure, not just another OCR tool. In 2024, they secured a total of roughly $1.1 million in pre-seed and seed funding, led by High-Tech Gründerfonds and joined by BackBone Ventures and angels including Ultimate AI's Reetu Kainulainen and Udi Miron [The SaaS News, March 2025] [PitchBook, Oct 2024] [Finsmes, Nov 2024].
The Infrastructure Wedge
Cloudsquid's positioning is specific. It is not a general-purpose document parser but an AI agent platform built for production pipelines in finance and operations. The core offering is an API that ingests PDFs, images, emails, and CSVs, extracts the relevant data, and structures it for downstream systems like ERPs and CRMs [Cloudsquid]. The company's marketing leans heavily on a 99%+ accuracy claim for data extraction, a figure that, if proven at scale, would address a primary pain point for enterprises drowning in document-based processes [Cloudsquid].
The platform includes components for prompt refinement, observability, and production integrations, aiming to let engineering teams embed data extraction directly into products and internal tools rather than building custom solutions [HTGF]. This focus on being a developer-friendly infrastructure layer, rather than a point-and-click SaaS tool, is the intended moat.
Targeting the Back-Office Bottleneck
The initial use cases are textbook examples of operational grunt work. The company highlights solutions for consumer packaged goods (trade deduction reconciliation), retail (markdowns, vendor chargebacks, inventory reconciliation), and manufacturing (bill of materials, supplier invoices, master data cleaning) [Cloudsquid]. One published case study details how a mid-market CPG brand used the platform to automate trade-deduction revenue recovery, a process typically mired in manual spreadsheet work [Cloudsquid].
- Accuracy as a product. The 99%+ claim is the headline feature, directly attacking the error-prone nature of manual data entry and basic OCR.
- Production readiness. The platform is built as API-first infrastructure, designed for engineering teams to integrate into automated workflows, not just for business users to upload files one by one.
- Vertical specificity. By focusing on known, high-volume pain points in CPG and retail finance, Cloudsquid can tailor models and pipelines to specific document types and business logic.
This approach targets a market served by established players like Instabase, Hyperscience, and Rossum, which offer broader enterprise intelligent document processing platforms. Cloudsquid's bet is that a narrower, deeper focus on finance and ops workflows, coupled with a developer-centric API, can carve out a defensible niche.
The Founder Stack
The team's backgrounds map neatly to the company's challenges. Rejmus (product) and Bae (infrastructure) bring the technical depth to build a reliable data pipeline, while McCarthy's experience at Ultimate AI provides a template for scaling an AI-centric SaaS sales motion. Their early backing from HTGF, a German deep-tech and software investor, and from angels with direct experience in AI automation, suggests investor confidence in this founder-market fit [HTGF].
| Founder | Role | Prior Experience |
|---|---|---|
| Mike McCarthy | CEO & Co-Founder | Early team & revenue, Ultimate AI [HTGF] |
| Filip Rejmus | CPO & Co-Founder | Data teams, Taktile, Uber [HTGF] [LinkedIn, 2026] |
| Sangwoo Bae | CTO & Co-Founder | Software engineering, AWS, Kubermatic [HTGF] |
The Proof Threshold
The primary counterfactual for any early-stage AI data company is proof of scale. A 99% accuracy claim on a controlled dataset is one thing; maintaining that performance across thousands of unique document formats, degraded scans, and complex multi-page layouts in a live enterprise environment is another. Competitors like Instabase and Hyperscience have spent years and raised hundreds of millions to tackle similar problems, building extensive validation and human-in-the-loop systems [Private candid take].
Cloudsquid's answer appears to be a combination of vertical focus and ISO 27001:2022 certification, achieved in late 2024, which signals a commitment to the security and compliance requirements of financial data handlers [Cloudsquid]. The lack of publicly named enterprise customers beyond a single case study is typical for a seed-stage company but leaves the traction question open. The real test will be whether the platform's accuracy holds under the weight of a growing, diverse client base, moving from a promising API to a mission-critical piece of back-office infrastructure.
The Next Twelve Months
With just over $1 million in the bank, the immediate roadmap is clear: convert early pilots into referenceable customers and prove the unit economics of automating high-volume document workflows. The seed round from late 2024, led by High-Tech Gründerfonds, provides runway to build that evidence [PitchBook, Oct 2024]. The company is hiring in Berlin, indicating a focus on product and engineering execution before a potential commercial push [Welcome to the Jungle].
The bet rests on a simple premise. If finance teams can trust an automated system to read their most important documents, the efficiency gains are substantial. Cloudsquid's $1.1 million seed round, anchored by HTGF and BackBone, is a wager that its founders have the right technical wedge and market focus to earn that trust. The question for 2025 is which mid-market brand will be the second name on their case study page.
Sources
- [Cloudsquid] Company website and product documentation | https://www.cloudsquid.io/
- [The SaaS News, March 2025] Cloudsquid Raises Over €900K in Funding | https://www.thesaasnews.com/news/cloudsquid-raises-over-900k-in-funding
- [PitchBook, Oct 2024] Cloudsquid Seed Round Listing | https://www.crunchbase.com/organization/cloudsquid
- [Finsmes, Nov 2024] Cloudsquid Raises $1M in Pre-Seed Funding | https://www.finsmes.com
- [HTGF] High-Tech Gründerfonds Portfolio Page | https://www.htgf.de/en/portfolio/htgffamily/cloudsquid/
- [LinkedIn, 2026] Filip Rejmus Profile | https://www.linkedin.com/in/filiprejmus/
- [Welcome to the Jungle] Company Profile | https://www.welcometothejungle.com
- [Crunchbase] Company Profile | https://www.crunchbase.com/organization/cloudsquid