Using Academic Papers as Market Evidence in a Business Plan

#

Korean government startup-support reviews in 2026 assess problem recognition, market analysis, and team composition. Presenting authoritative statistics together with their sources and years is repeatedly cited as a way to make a plan more persuasive to reviewers.

Founders often write sentences such as 'the domestic market is estimated at approximately KRW X trillion.' Without a source, a reviewer cannot verify the figure directly. The same applies to AI review agents. Even if an agent generates a locally plausible sentence, logical verification has a gap when the document does not contain the evidence chain behind the figure.

The core problem is not the figure itself but the connection between the figure and its source. Citing a specific paper in a KCI-indexed journal for the same market-size estimate lets a reviewer also consider the paper's research methods.

#

In research published in Nature in September 2026, James Zou's team at Stanford introduced Paper2Agent, a framework that automatically turns paper text, code, and data into AI agents based on MCP servers. In an experiment involving 100 computational-biology papers, 74 were converted without human intervention, producing 593 tools.

For startup reviews, these results illustrate the difference between consulting a paper's abstract and using its methodology and data structure through an agent. An abstract conveys the conclusions, while a structured paper agent can trace the full context in which a particular figure was obtained under particular conditions.

A practical risk of using AI in business plans is treating sentence generation and structural verification as the same task. By contrast, loading verified academic papers into agents as structured knowledge creates a basis for asking a market claim: 'Which study and which conditions produced this figure?'

#

The Korea Citation Index (KCI) provides titles, authors, abstracts, publication years, research fields, citation counts, DOI, UCI, and open-access status for papers in indexed Korean journals through a REST API (XML). It can be used after applying for an API key through the Public Data Portal (data.go.kr).

Supported parameters include title, author, keywords, abstract, and date range in YYYYMM format. A request can retrieve up to 100 records (displayCount=100). The National Research Foundation of Korea also publishes separate file datasets with annual counts of papers, authors, and references in Korean academic research.

#

Guidance on TIPS document evaluation repeatedly recommends that team capability should go beyond members' names: the relationship between their fields of study, experience, core capabilities, and the business model should be explicit. AI review agents apply the same perspective when reviewing team capability.

AI review cannot be guaranteed to verify relationships across an entire document consistently. However, a plan that names academic sources gives agents a path for tracing the relationship between claims and evidence, which directly helps make feedback more specific.

#

When using academic papers as market evidence, the common practical omission is not citation formatting but checking whether the study's conditions match the business. Reviewers consider whether the paper covers the Korean market, whether the study period fits the business timing, and whether the sample is large enough to support generalization.

• Query industry or technology keywords in the KCI API → inspect highly cited papers → check open-access status
• Check the study subjects, period, and sample in the abstract, and compare them with your business conditions
• When stating a figure in the plan: figure + source (author, journal, year) + one sentence on the original study's conditions
• When describing research experience as team capability: state the connection between the paper's topic and a core product or service function in one sentence
• Before submitting to OpenSeed, check every sentence in the market-analysis section that contains figures without sources