Cerpro comparison (INTERNAL)¶
Internal only — not for Theegarten
From CERPRO_deepresearchreport_GPT.md (cerpro.io research). Customer leave-behind: Brief (EN).
Executive summary¶
- Cerpro (cerpro.io) is a workflow-first Berlin industrial SaaS company — not a PLM vendor — focused on drawing → structured QA documentation. [Source: CERPRO_deepresearchreport_GPT.md]
- QualiSpec (Dec 2024): extract dimensions/tolerances/features → EMPB, inspection plans, VDA-style exports. QualiChain: OEM–supplier quality orchestration. [Source: report]
- Pivot: contract-manufacturing quoting/digitization → narrow QA documentation wedge; validated via founder-led sales, references, Instagram-led top-of-funnel. [Source: report]
- Security: “Made in Germany” + AWS Frankfurt, but US LLM subprocessors (OpenAI, Anthropic, Google) in 2025 terms — weak for Marcel-style controlled-environment requirements. [Source: report]
- RapidDraft: compete upstream (engineering release); use Cerpro as proof that Mittelstand pays for document workflows; do not fight on EMPB automation as primary wedge. [Strategic]
Source files in this folder¶
| File | Role |
|---|---|
CERPRO_deepresearchreport_GPT.md |
Primary cerpro.io / public-source forensic report (7 sections) |
theegarten_pactec_quality_templates.md |
Maps Cerpro QualiSpec output types to Theegarten template pack |
quality_templates/* |
Example EMPB/inspection artifacts (Cerpro/RapidDraft export narrative) |
No separate cerpro.io/ scrape folder exists in the pilot directory — the GPT deep-research markdown is the cerpro.io research corpus. [Confirmed: directory listing May 2026]
Company & GTM forensics¶
| Field | Detail | Confidence |
|---|---|---|
| Legal entity | Cerpro GmbH, HRB 257881, Charlottenburg | Confirmed (register aggregations) |
| Founded | 14 June 2023, Berlin | Confirmed |
| Founders | Frederik Frei (CEO/sales), Sascha Müller (CTO/DS), Henrik Pitz (COO/finance) | Confirmed |
| Accelerator | Techstars Riyadh 2023; returned focus to Germany | Confirmed |
| Funding | ~€2M pre-seed, Nov 2025 (seed + speed, D11Z, EIT Manufacturing, Techstars) | Confirmed |
| Team scale | 1–10 employees; hiring sales (SDR, full-cycle AE) — GTM institutionalization | Inferred |
| Customers cited | 100+ companies, 5 European countries; “100+ paying” (promotional) | Medium-low |
| ARR estimate (report) | ~€0.6M–€1.2M band; sub-€1.5M likely | Medium-low inference |
| Burn estimate | ~€90k–€150k/month post-raise | Inference |
Strategic pivots (timeline)¶
- Geographic: Riyadh accelerator → Germany-focused
- Product: Broad CM digitization / RFQ quoting (early blog 2024) → QualiSpec inspection documentation (late 2024)
- GTM: Supplier / contract manufacturers first → QualiChain OEM–supplier platform (2025–2026)
GTM mechanics that work (copy for RapidDraft)¶
| Mechanic | Cerpro behavior |
|---|---|
| Wedge | Manual copy from drawing → Excel → inspection report (audit pressure) |
| Buyer | Head of Quality, QE, plant manager — not CIO first |
| Motion | Live before/after on customer drawings; 14-day trial; productive in one day |
| Channel | Founder-led demo + references + trade shows (Hannover, EMO) + Instagram for awareness |
| Sales cycle | Supplier SME: 2–8 weeks; OEM QualiChain: 3–9 months |
| Pricing | Annual base + usage per drawing analysis |
| Language | “Audit-proof,” “document-proof,” “plug-and-play” — not “foundation model” |
Product — cerpro.io (QualiSpec & QualiChain)¶
QualiSpec (core)¶
| Capability | Detail |
|---|---|
| Inputs | PDF, JPG, DXF, CAD |
| Extraction | Dimensions, tolerances, drill holes, features → characteristic model |
| Outputs | VDA, EMPB, inspection plans, quality documentation; PDF & Excel |
| Claims | 70–80% time savings (customer references); stamp/number inspection points |
| Integrations | Case refs: Majesty ERP/PDA; “validatable, document-proof” outputs |
QualiChain (strategic direction)¶
- Drawing as common data foundation for supplier quality
- Extraction + stamping + FAI reporting + supplier doc standardization
- Multi-party approvals, reconciliation OEM ↔ supplier
- Domains: development, QM, goods receipt, supplier quality
“More use cases” (execution risk)¶
Public page also claims: PMI extraction, feasibility analysis, similarity search, 3D→2D conversion, historical drawing digitization. Report flags very different technical tiers — sprawl risk if pursued too early.
Stack position¶
NOT: PLM / BOM master / digital thread backbone
IS: Workflow + document-intelligence on 2D drawing artifact
→ optional climb to multi-enterprise QM (QualiChain)
Coexists with Teamcenter, Windchill, CIM Database, Vault — augments operator-level QA paperwork.
Inferred technical architecture¶
Upload → object storage (AWS Frankfurt)
→ preprocess (raster/vector, OCR, title block, tables)
→ deterministic parsers (dims, tolerances, GD&T normalize)
→ multimodal LLM for ambiguous notes (Anthropic/OpenAI/Google per terms)
→ characteristic graph + template mapping
→ workflow (review, stamp, export EMPB/inspection plan)
→ usage metering per drawing analysis
Hard: GD&T semantics, note-to-feature association, revision drift, customer-specific symbols.
Commoditizing: generic PDF → JSON extraction (Werk24, etc.).
Security, GDPR & European trust (cerpro.io)¶
| Topic | Cerpro public posture | Diligence read |
|---|---|---|
| Marketing | “Data Security made in Germany,” AWS DE, 7-day backup, GDPR | Opens SME doors |
| Hosting | AWS Frankfurt (subprocessor) | Control plane DE |
| AI stack | OpenAI, Anthropic, Google, Microsoft, Hotjar (Sep 2025 terms) | Not sovereign |
| Enterprise IAM | No public SSO/SAML/RBAC | Weak for OEM IT |
| Deployment | Browser SaaS, multi-tenant implied | No public on-prem/VPC |
| DPA | Processor clauses, breach notification, deletion, audit rights | Credible for SME |
| Training clause | Anonymized data may improve software (Jun 2025 terms) | Marcel will ask — process unspecified |
| Marketing site | Google Analytics, Ads, Meta pixels | Normal startup; not air-gapped story |
Assessment (report): Moderately credible for SME / mid-market QA docs; early for enterprise platform or Marcel-style controlled local environment (R01).
Why this matters for Theegarten¶
Marcel’s requirements (local/controlled AI, know-how, no opaque training) are stricter than Cerpro’s default public SaaS + US model stack. If Theegarten evaluates Cerpro-like tools, the same security questions apply — RapidDraft can differentiate with on-prem/VPC path and explicit model routing. See Data security.
Customer acquisition & positioning¶
First wedge (confirmed pattern)¶
Manual reading of drawings → copy to Excel → inspection reports for audits
References span medtech, aerospace, plastics, CNC, mechanical engineering — same ICP cluster as packaging-machine suppliers.
Category narrative¶
“AI-powered technical-drawing intelligence for quality workflows” → stretching to supplier quality orchestration.
Vulnerabilities (where RapidDraft wins)¶
| Cerpro weakness | RapidDraft opportunity |
|---|---|
| Downstream of design decisions | Upstream release authority |
| No PLM-native release control | CIM Database–centric integration story |
| Model commoditization | Engineering-intent graph + rules + PLM lineage |
| Sovereignty gap | Controlled deployment + EU-only inference option |
| Use-case sprawl | Wedge discipline on one release workflow |
Strategic lessons for RapidDraft (from report § Strategic Lessons)¶
Copy¶
- Wedge discipline — one workflow, visible ROI
- Founder-led pilots until motion repeats
- Operational buyer language (release defects, traceability — not “AI transform”)
- Paid pilot → annual fee + volume (drawings / release packages analyzed)
- Template + standards depth as moat layer
Avoid¶
- Leading with supplier EMPB / QualiSpec parity
- Competing as generic drawing extraction
- Cloud-only without controlled-environment story for OEM engineering
- Promising 3D→2D or full PLM replacement early
Recommended RapidDraft sequence¶
- Drawing + release quality first
- Review automation second
- Traceability / auditability third
- Knowledge capture / manufacturing intelligence fourth
Moat ambition¶
Cerpro moat ≈ workflow embedding + correction history + templates.
RapidDraft moat target ≈ engineering-intent graph (3D ↔ drawing ↔ tolerance rationale ↔ change history).
Risks, blind spots & founder assessment¶
Unresolved in public data: exact ARR, pricing tiers, model mix in production, private deployment offers, deep PLM integrations, revision-control quality inside product.
Founder assessment (report): Competent industrial operators; pivot discipline; serious but not impregnable — moat is commercial/workflow more than technical.
Outcome range: Strong supplier-QM specialist or boxed in by enterprise security / commoditized extraction.
Cerpro ↔ Theegarten template pack¶
Theegarten quality templates (Quality forms) mirror Cerpro QualiSpec output targets:
| Cerpro output | Theegarten pack |
|---|---|
| EMPB / FAI | 01_EMPB, TP_EMPB docx |
| Inspection plan | 02_Pruefplan, master sheet |
| Series inspection | 05_Series_Inspection |
| Document register | Quality documentation summary |
Example part ZG-2451-A03 Rev C demonstrates extraction → characteristic rows — useful for Phase 2 demo (“downstream inspection doc lineage”), not Phase 1 release check.
Positioning at Theegarten (if Cerpro comes up)¶
| If they ask… | Say… |
|---|---|
| “Do you compete with Cerpro?” | Different step: we focus on engineering release in CIM/SOLIDWORKS; inspection automation is adjacent. |
| “Can you do EMPB?” | Phase 2 / partner path; primary pilot is release package quality. |
| “Why not Cerpro for drawings?” | They excel at QA documentation after release; we reduce errors before QA and manufacturing. |
Implications for RapidDraft demo¶
- Never trash-talk Cerpro to Marcel.
- Use this intel to sharpen deployment and PLM differentiation.
- Quality templates = optional credibility, not primary story.
- Full raw report:
../CERPRO_deepresearchreport_GPT.mdin repo parent folder.