Upload a pitch deck and get a Clarity Score out of 100 with the structural problems named and located. Five scored sections: Problem Definition, Solution Logic, Market Evidence, Business Model Physics and Deal Structure. Free, no account required.
Never. Your documents are processed to extract structural variables and are never added to any training corpus, model weights, or benchmark dataset. Raw files are held no longer than 30 days under our retention ceiling, and you can ask us to erase them at any time — we action deletion requests within 30 days. The structural data derived from your deck persists and can include identifying details such as your name, so that erasure right covers it too. You retain 100% ownership of your IP.
No. Every audit is private by default. We never sell, share, or broker your files to investors or any third party. If — and only if — you score 85 or above on a combined deck and financial model audit, you are offered the option to generate a private, read-only askOdin Verified link for your own data room. Anything that leaves here leaves because you sent it.
Conversational LLMs are optimised to be agreeable. Ask one to review your deck and it usually flatters the thesis and suggests phrasing tweaks. askOdin Crucible is deterministic judgment infrastructure: it extracts each document in state isolation, then runs deterministic rules across them to surface asset-liability mismatches, CAC/LTV inversions and duration risks — the contradictions a chat model reconciles away — and maps the counter-arguments an investment committee will use to pass.
No. Beautiful design never saves a round when the unit economics or market assumptions collapse. Crucible evaluates narrative rigour, logical coherence and mathematical defensibility — it compiles your math, it does not format your slides.
The Clarity Score™ (0-100) quantifies the structural defensibility of your thesis — how well your claims hold up against each other and against the evidence you supplied. A higher score means fewer brittle assumptions. The bands are 85 and above (PRIORITY), 70-84 (INVESTIGATE), 50-69 (WATCH) and below 50 (PASS): below 50 signals critical narrative contradictions or unbacked claims, while 85 and above represents institutional-grade clarity. A Kill Shot is separate from the score: a single terminal finding — a governance problem, or a claim that breaks physics or arithmetic — that floors the score to 0 regardless of how the rest reads.
Crucible audits five canonical pillars, each scored out of 20. Problem Definition: structural pain, market urgency and timing. Solution Logic: product mechanism, defensibility and moat. Market Evidence: TAM reality, customer validation and pricing power. Business Model Physics: unit economics, margin profile and capital efficiency. Deal Structure: round sizing, milestone credibility and realistic runway. Across the published benchmark, Business Model Physics is where decks break most often and Problem Definition least — founders describe what is broken in the world well, and how the company makes money badly.
Crucible is completely free for founders and no account is required to run your first audit. No credit card, no trial expiry. In exchange we ask only for an unvarnished review of the engine once your report is ready — founder feedback is what makes it sharper; the institutional tier is what funds pay for.
Pitch decks as PDF or DOCX, and financial models as XLSX or CSV — up to 10 MB per file, three files per audit. If your deck lives in Google Slides, Figma, Canva or Notion, export it to PDF first; that export is what an investor receives anyway. A complete structural audit takes about 3 minutes.
Yes. Your deck alone produces a complete narrative audit and your Clarity Score™. Adding a financial model lets the engine cross-check qualitative claims against spreadsheet reality — the most common place a contradiction hides — and it is the prerequisite for a verified data room record. No model yet? Run the deck now and re-run once you have one.
A 0 is not an error — it is a FATAL verdict, triggered when the engine finds irreconcilable data it cannot compile around: revenue that contradicts itself across slides, a model that violates its own narrative, a claim with no support. Generative tools smooth these over to keep you happy. Crucible names the fracture and points at the page, so you fix it before an investor finds it.
Sometimes the deck, often the business. A low score on the narrative sections usually means you have not explained something you understand. A low score on Business Model Physics usually means the model itself does not close, and no amount of rewriting changes that. Knowing which one you are facing is the point.