Deal Term Insights is a research-informed first read of venture financing terms. It explains clauses, highlights unusual provisions, and gives a directional assessment of how founder- or investor-friendly a deal appears. It is not legal advice.
Built on real financing documents, not just term sheets
This distinction matters. The tool is not based only on preliminary term sheets or generic templates. It is grounded in filed venture financing documents, especially Certificates of Incorporation in the United States and Articles of Association in Germany.
These documents show the terms that actually made it into the legal structure of the company. In negotiations, investors and founders may ask for many things. What matters is what is ultimately agreed and filed. My PhD work looked at those final documents at scale and connected the resulting contracting data to how ventures developed over time.
That is what allows the tool to benchmark a clause against observed practice rather than against anecdote or wishful thinking. When the tool flags a term as unusual, restrictive, or investor-protective, the judgment is anchored in real negotiated deal terms and in the entrepreneurial finance literature.
Three evidence bases
The tool draws on three bodies of evidence: my own contract research, a proprietary German governance database, and the entrepreneurial finance literature.
1. Hand-coded US venture financing documents
During my PhD, I analyzed roughly 500 Certificates of Incorporation from US-based ventures. The sample covers companies founded between 1990 and 2021 and spans several venture-backed industries, including software, medical equipment, biotechnology, pharmaceuticals, and healthcare services.
The documents were hand-coded with a focus on how investors and founders allocate control rights, approval rights, and protective provisions. This makes it possible to compare how contractual governance differs across sectors, stages, and funding environments.
For European readers, the US Certificate of Incorporation is important because many core preferred-stock rights must be set out in the company's charter. Terms that would often sit in a private shareholders' agreement in Europe are therefore visible in the filed constitutional document in the United States. The Certificate of Incorporation is not a pitch document or a wish list. It is part of the financing's legal architecture and contains many of the core rights associated with preferred stock.
The work asks which rights investors receive, why particular approval rights are used, and when these rights may support or constrain venture development.
2. Proprietary German Articles of Association database
The tool also draws on a proprietary corpus of roughly 18,000 German Articles of Association across about 4,700 firms. The database covers German startups and venture-backed companies founded between 1990 and 2025, giving a population-scale view of filed governance structures over several funding environments.
The documents were coded using computational methods, including large language models, and linked to transaction and company data. This makes it possible to study governance provisions across a much broader set of companies than manual coding alone would allow.
German Articles of Association do not usually contain the full economics of financing, which are often set out in private shareholders' agreements. Their value lies in their ability to reveal the filed governance structure of real companies and to identify patterns, deviations, and unusual provisions at scale.
3. Entrepreneurial finance research and model documents
The interpretation is also anchored in established entrepreneurial finance research, including work by Kaplan and Strömberg, Lerner, Gompers, and related research on venture contracts, control rights, and investor protection.
Where relevant, the tool also compares terms to widely used model documents, including NVCA, Y Combinator, and GESSI forms. These are not treated as the market itself, but as useful reference points for understanding how different contractual architectures are usually framed.
Why contracts matter, and how the tool reads them
Venture contracts matter because they are opaque. They are the mechanism through which investors and founders allocate cash flow and control, divide risk and reward, and set the terms on which they will work together after the money goes in. Yet they are hard to read, harder to compare, and rarely visible across deals. That opacity is the problem this tool is built to reduce. It makes a single contract legible and benchmarks it against observed practice, so both sides can see what a term does, how usual it is, and how it helps them manage risk rather than be surprised by it later.
The tool does not treat any term as automatically good or bad. Investors need protection for the capital they put at risk. Founders need enough operating freedom to build the company. The useful question is not simply whether a term exists, but what it does, which decision or economic outcome it touches, and how it may affect the company over time.
Three ideas from my research shape this reading, and they apply across the whole contract, not to any single clause type.
First, terms are not all the same. A provision that touches day-to-day execution has a different effect from one that protects the basic economics of the investment, or that governs an existential decision such as a sale, a liquidation, or a major financing.
Second, terms must be read in context. Risk, uncertainty, stage, and the nature of the company matter. Early-stage ventures, capital-intensive businesses, regulated sectors, and companies with greater technical or commercial uncertainty may justify stronger investor protections than simpler or later-stage companies. A clause is therefore assessed against the type of decision it covers and the problem it appears to address, not only against a generic template.
Third, a term's effect changes as incentives change. A provision that looks like a sensible safeguard at the time of investment can later become a hold-up lever if an investor disengages, faces different incentives, or no longer has the information to use the right well. For that reason the tool reads terms by their likely consequences, not only by their legal label.
Approval rights are one clear example of this from my sample. The same kind of consent right can mean very different things. A consent right over hiring, budgets, spending, product development, business lines, leases, technology acquisitions, or commercial partnerships affects how the company operates day to day, and can become restrictive because it gives investors influence over ordinary execution. A consent right over a sale of the company, liquidation, a major financing, charter changes, redemption, dividends, or transfers of core assets is different: it is more directly tied to protecting the investor's expected gain and the basic economics of the deal.
Approval rights are only one illustration. The same logic runs across liquidation preferences, anti-dilution protection, redemption rights, pay-to-play provisions, drag-along and tag-along rights, information rights, board composition, vesting, and transfer restrictions. The tool therefore does not just identify or count clauses. For each term it asks what the clause does, which decision or economic outcome it affects, how unusual it is against real filings, and whether it looks proportionate to the financing context.
Company database and limits
The company database is an exploratory research asset built from venture-financed company records and coded filing data. It is not a complete register of all companies or all startup financings. The data includes timing and stage signals, which matters because contract terms differ across seed, Series A, growth, and later-stage financings. The tool should therefore be read as a structured benchmark and diligence aid, not as a definitive market survey for every jurisdiction or stage.
How to read the output
The headline assessment is a band: investor-heavy, investor-protective, balanced, or founder-friendly. It is not a precise grade. Standard terms are expected and are noted as such. The main value is in identifying what deviates from observed practice, why it matters, and how it may affect negotiation or diligence. The output should be treated as a structured starting point for discussion, not as a conclusion.
Scope and limits
The tool reads one document at a time, a term sheet, charter, or shareholders' agreement. A single document rarely contains every term of a deal, so where the economics sit in one document and the governance in another, read them together. Coverage is strongest for the United States and Germany; other jurisdictions are a directional cross-border read. The calibration is anchored in real filed contracts and model documents and continues to improve as more examples are added, so the band is a considered judgment, not a guarantee.
Privacy and AI processing
Uploaded or pasted documents are used only to generate the current review. When a hosted AI model is used, the tool first creates a de-identified working copy by replacing direct identifiers, such as company names, emails, addresses, and register numbers, with placeholders. The placeholder map remains in memory for that request only, allowing the displayed report to restore names locally while the hosted model sees only the reduced version. By default, the raw contract and the analysis output are not stored or shared. If a user opts in at the end of the questionnaire, the tool saves only a de-identified research record locally, without the raw contract or placeholder map.
Jurisdiction coverage
The strongest calibration today is the United States and Germany. US financing documents and model form evidence anchor many of the economic and preferred-stock terms. The German corpus anchors the governance layer in filed Articles of Association. Other jurisdictions can still be reviewed, especially where the smart AI layer can translate the legal effect into the tool's clause categories. In those cases, the output should be read as a directional cross-border assessment rather than a local legal opinion.
In short
Built on my PhD research, Deal Term Insights turns messy venture financing documents into structured deal intelligence, making negotiated terms more transparent and easier to use in investment decisions. It draws on roughly 500 hand-coded US venture financing documents, a proprietary database of roughly 18,000 German Articles of Association across about 4,700 firms, and the entrepreneurial finance literature. It is not legal advice.