AI Is Changing Crowdfunding. But Are We Building Better Decisions?
· Methodology
After 15 years in crowdfunding, Michael Eis examines how AI can support project creators, regulatory due diligence for ECSPs and independent investment analysis - and why human judgement must remain central.
*After 15 years in crowdfunding, I believe artificial intelligence could transform how projects are created, evaluated and funded. But the real breakthrough will not be automation alone. It will be transparency, accountability and better-informed decisions.*
**By Michael Eis, Founder of CrowdDiligence and boomerang.dk**
For most of its history, crowdfunding has been about removing barriers.
We made it easier for entrepreneurs to access capital, for creative projects to find supporters, and eventually for ordinary investors to participate in opportunities that were previously largely reserved for professional investors.
Technology made this possible. But technology did not eliminate the fundamental challenges of crowdfunding.
Project creators still struggle to present their ideas convincingly. Investors still face significant information asymmetries. Crowdfunding platforms face regulatory obligations that require considerable documentation and verification. And conducting proper due diligence remains difficult, particularly for retail investors who lack the time, resources and experience of professional investment teams.
Artificial intelligence has the potential to change all three.
Having worked with crowdfunding for approximately 15 years, I am now exploring that potential through two platforms I founded: boomerang.dk, a Danish reward-based crowdfunding platform, and CrowdDiligence, an independent due diligence platform focused on equity crowdfunding.
Our work spans three areas: helping entrepreneurs develop better campaigns, supporting crowdfunding platforms with regulatory due diligence, and giving retail investors access to more structured and independent investment analysis.
Although these applications serve different purposes, they share an important principle.
AI should improve the quality of human decisions, not simply automate the process of making them.
From 15 Years of Experience to AI-Assisted Campaign Development
I founded boomerang.dk approximately 15 years ago, when reward-based crowdfunding was still a relatively unfamiliar concept in Denmark.
Over the years, I have seen countless projects, successful and unsuccessful, and observed how differences in presentation, communication, funding goals and reward structures can influence campaign outcomes.
One lesson has become particularly clear: having a good idea and communicating a good idea are two very different things.
Many entrepreneurs are excellent at developing products, producing creative work or running small businesses. But they are not necessarily experienced copywriters, marketers or crowdfunding specialists.
That creates an unnecessary barrier.
At boomerang, we use AI to help project creators develop their campaigns, drawing on patterns associated with successful crowdfunding projects and combining them with the practical knowledge accumulated through years of operating a crowdfunding platform.
The objective is to make that experience accessible to more people.
Instead of expecting every entrepreneur to understand crowdfunding psychology, campaign structures and reward design from the beginning, AI can help them develop clearer project descriptions, communicate their value proposition and present rewards more effectively.
However, there is an important distinction between learning from successful campaigns and simply copying them.
A campaign should reflect the actual project, its capabilities and its commitments. AI must not manufacture credibility, exaggerate expected results or promise something the creator cannot deliver.
There is also a methodological challenge. Successful historical campaigns can reveal useful patterns, but correlation is not causation. A particular writing style or reward structure may be associated with successful funding without being the reason for that success.
Our ambition is therefore not to create a machine that promises successful campaigns.
It is to give project creators better tools and better starting points, informed by experience.
That is one side of the crowdfunding equation.
The other side is considerably more complex.
Bringing AI Into Investor Due Diligence
Equity crowdfunding has expanded access to investments in privately held companies, but access to investment opportunities has not always been accompanied by equivalent access to independent analysis.
Professional investors can employ analysts, commission legal reviews, examine financial statements and conduct extensive investigations before committing capital.
Retail investors generally cannot.
They may have access to a campaign presentation, a Key Investment Information Sheet (KIIS) in Europe or Form C disclosures in the United States. But transforming those documents into a meaningful assessment of business risks, financial sustainability, management capabilities and investment terms requires considerable effort.
This is the problem CrowdDiligence was created to address. Every analysis starts from the official filings and is structured using the [platform's due diligence features](/how-it-works).
Our platform combines a [structured due diligence methodology](/how-it-works) with community participation and AI-assisted analysis.
We have developed a dedicated AI due diligence agent, guided by a detailed set of instructions that we continuously revise as we gain experience.
The agent is designed to examine available information, identify relevant findings, structure evidence and formulate assessments that can be incorporated into CrowdDiligence's existing due diligence framework.
That framework examines areas such as management, market, financials, legal matters, product, operations and risk.
Our analytical approach also includes weighted assessments, confidence intervals and Monte Carlo simulations to help communicate uncertainty rather than present a single numerical score as an objective truth.
The purpose is not simply to generate a polished investment report.
It is to create a structured analytical foundation that can be reviewed, challenged and improved.
And that distinction matters.
The Problem With AI-Generated Confidence
Large language models are remarkably effective at producing convincing explanations.
Unfortunately, a convincing explanation is not necessarily a correct one.
An AI-generated due diligence report can appear comprehensive while relying on incomplete information, questionable assumptions or sources that do not support its conclusions.
In financial analysis, this is more than a technical inconvenience.
A missing liability, an incorrectly interpreted financial figure or an unsupported market assumption can materially change an investment assessment.
We therefore believe that AI-generated findings should not automatically carry the same credibility as findings verified by human analysts.
CrowdDiligence addresses this through a methodology that distinguishes AI-assisted assessments from verified human contributions.
Under our methodology, unverified AI-assisted assessments carry a reduced provisional weight until they are reviewed and verified by a human DD Lead.
The objective is to recognise AI's analytical contribution without presenting its initial output as independently established fact.
We also address another problem that is particularly relevant in crowdfunding: conflicts of interest.
Entrepreneurs, employees, existing investors and other stakeholders may all have legitimate reasons to participate in a discussion about a company. Their knowledge can be valuable.
But their financial or personal interests can also influence their assessments.
Rather than automatically excluding those contributors, our methodology allows declared conflicts to be reflected through reduced weighting. Assessments from contributors flagged as biased on a particular project carry a substantially reduced weight.
These weightings are methodological choices, not scientifically established measures of reliability. They must remain open to validation and adjustment as we accumulate evidence.
Neither mechanism guarantees accuracy or eliminates manipulation.
But both reflect an important principle.
AI can contribute to the analysis. Interested parties can contribute their knowledge. Neither should automatically be mistaken for independent verification.
Extending AI Due Diligence to Crowdfunding Platforms
Our work at CrowdDiligence is not limited to independent investor analysis.
We are also designing a dedicated AI-assisted regulatory due diligence framework for European Crowdfunding Service Providers (ECSPs) operating under the European Crowdfunding Service Providers Regulation ([Regulation (EU) 2020/1503, ECSPR](https://eur-lex.europa.eu/eli/reg/2020/1503/oj)).
This second application addresses a different but closely related challenge.
Crowdfunding platforms have regulatory responsibilities when admitting projects and making investment opportunities available to the public.
Article 5 of ECSPR establishes minimum due diligence requirements concerning project owners. Other provisions of the regulation address the information provided to investors, including the Key Investment Information Sheet.
Traditionally, much of the associated work involves reviewing documentation, cross-referencing information, identifying inconsistencies and recording compliance assessments.
AI offers considerable opportunities to support these processes.
Our proposed architecture builds on CrowdDiligence's analytical framework but introduces a separate regulatory due diligence layer.
We are designing this framework around Article 5 checks, KIIS completeness and consistency reviews, evidence traceability, document versioning, human oversight and auditable assessments.
The objective is to help platform compliance teams identify potential gaps, organise supporting evidence and document their reviews more efficiently.
Importantly, we do not believe an AI system should independently determine whether a project satisfies regulatory admission requirements.
A finding generated by an AI agent should be linked to its underlying evidence, allowing a compliance officer to examine the source, challenge the interpretation and document the final decision.
This requires more than an AI-generated report.
It requires a structured system of evidence, clear responsibilities, version control and meaningful human review.
And it requires understanding the distinction between regulatory compliance and investment risk.
A crowdfunding project may satisfy regulatory admission requirements while still representing a highly speculative investment.
Conversely, a commercially promising business may have unresolved documentation or compliance issues.
These are fundamentally different assessments and should not be collapsed into a single score.
Our ambition is therefore to develop a shared evidence infrastructure supporting two distinct purposes: regulatory due diligence for crowdfunding platforms and independent investment analysis for retail investors.
The underlying documents and verifiable facts may overlap, but the conclusions, responsibilities and decision-making processes must remain separate.
This architecture is still being developed. We are not presenting it as a completed regulatory compliance solution.
However, it demonstrates how AI-assisted analysis could support multiple participants in the crowdfunding ecosystem without confusing their respective roles.
Why a Shared Evidence Infrastructure Matters
One of the most interesting opportunities in AI-assisted due diligence is not the ability to generate reports more quickly.
It is the possibility of creating a reusable and traceable body of evidence.
Consider a company raising capital through an equity crowdfunding platform.
The company provides financial statements, ownership information, management details, investment terms, market assumptions and other supporting documentation.
The platform needs to perform its regulatory checks.
The investor needs to understand the investment opportunity and associated risks.
Traditionally, these processes may involve separate reviews of overlapping information.
A well-designed AI system could help structure and reference that evidence more efficiently, while allowing different parties to apply their own analytical frameworks.
For example, a financial statement might be relevant both to a platform's verification procedures and to an investor's assessment of financial sustainability.
But the platform's regulatory responsibilities and the investor's investment decision remain distinct.
At CrowdDiligence, we believe this distinction should be reflected in the architecture itself.
Evidence can be shared or reused where appropriate. Conclusions should remain specific to the purpose of the assessment, and access to confidential information must be controlled.
AI-generated findings should identify the underlying sources and preserve sufficient context for reviewers to understand how the finding was produced.
If a document changes, the system should make it possible to identify which assessments relied on the previous version.
If a human reviewer disagrees with an AI-generated interpretation, that decision should be documented rather than silently replacing the original result.
This is particularly important in regulated environments, where a plausible answer is not sufficient.
An assessment must also be defensible.
We believe evidence traceability, human accountability and methodological transparency should become core design principles for AI-assisted financial due diligence. Our [white paper](/whitepaper) describes how these principles shape the CrowdDiligence framework.
Due Diligence Should Be a Living Process
Traditional due diligence reports are often produced at a particular moment.
The analysis is completed, conclusions are presented, and the document gradually becomes outdated.
But early-stage companies evolve continuously.
New financial information becomes available. Management teams change. Regulatory issues emerge. Funding rounds close. Previously reasonable assumptions may no longer hold.
This is particularly relevant in equity crowdfunding, where investors are dealing with young businesses operating under significant uncertainty.
We believe due diligence should be capable of evolving alongside the company.
AI provides an opportunity to make that process more practical.
Information can be reassessed when new documentation becomes available. Previously identified risks can be revisited, and inconsistencies between earlier and later disclosures can be highlighted for human review.
A platform's regulatory assessment may need to be updated when new documentation is submitted.
An investor's assessment may change when new operational or financial information emerges.
And a community-based analysis should be able to incorporate new evidence without losing the history of previous conclusions.
At CrowdDiligence, we are developing the methodology and workflows needed to support this approach.
Our broader ambition is to combine the scalability of AI with the independence and collective intelligence of retail investors, while also developing practical tools for regulated crowdfunding platforms.
The result should not be an automated machine telling people which companies to invest in.
It should be an analytical environment that helps investors and compliance professionals understand what they know, what remains uncertain and which risks deserve further investigation.
Why AI Governance Matters Now
In July 2026, the Global Equity Crowdfunding Alliance (GECA) announced its [AI Governance Task Force](https://thegeca.org/blogs/ai-governance-task-force-crowdfunding/), chaired by Neera Patel. You can also hear Neera Patel discuss the initiative in the [GECA podcast on AI governance in crowdfunding](https://thegeca.org/podcasts/neera-patel-ai-governance-crowdfunding-geca-taskforce/).
The initiative reflects a wider recognition that crowdfunding needs practical approaches to responsible AI adoption, including transparency, accountability, regulatory awareness and human oversight.
As a GECA member and a founder actively developing AI applications within crowdfunding, I welcome this development.
The industry needs more than general statements about responsible artificial intelligence.
We need to discuss actual implementations.
How should an AI-generated finding be verified?
How do we distinguish documented facts from analytical assumptions?
How should conflicts of interest affect crowd-generated assessments?
How do we maintain meaningful human responsibility when an AI agent performs much of the initial analytical work?
How can a platform demonstrate that an AI-assisted compliance finding is supported by appropriate evidence?
And how can smaller crowdfunding businesses adopt these technologies without building governance systems that are prohibitively expensive or operationally impractical?
These questions cannot be resolved by technology providers alone.
They require collaboration between crowdfunding platforms, investors, regulators, technology developers and experienced practitioners.
The EU AI Act adds an important regulatory dimension, but appropriate governance must also consider financial regulation, data protection and the actual risks created by a system's intended use.
Not every AI tool used in a financial context is automatically classified as high-risk under the AI Act. Equally, falling outside a specific high-risk classification does not make accuracy, transparency and accountability irrelevant.
The appropriate governance framework must reflect what the system actually does, who relies on its outputs and what consequences may follow from an incorrect assessment.
There is also a broader issue.
As AI-based due diligence becomes increasingly accessible, we risk creating a situation where many apparently independent analyses originate from similar models, datasets and assumptions.
A hundred AI-generated assessments are not necessarily a hundred independent opinions.
If they rely on the same incomplete source material or repeat the same analytical errors, apparent consensus may simply be duplicated uncertainty.
This is one reason why we believe independent verification, diverse human contributions and transparent methodology are essential.
The objective should not be to maximise the volume of analysis.
It should be to improve its reliability.
Three Applications, One Underlying Principle
Working with AI at boomerang and CrowdDiligence has reinforced an important observation.
Artificial intelligence does not have a single role in crowdfunding.
On the project creator's side, it can reduce barriers, improve communication and help entrepreneurs learn from practical experience.
On the crowdfunding platform's side, it can support regulatory verification, document reviews and compliance processes.
On the investor's side, it can help reduce information asymmetry, structure research and make sophisticated analytical methods more accessible.
Yet these applications also introduce different risks.
An AI assistant helping an entrepreneur write a campaign can unintentionally encourage exaggerated claims or create the appearance of credibility without supporting evidence.
An AI agent supporting a platform's regulatory due diligence can overlook relevant documentation, misinterpret legal requirements or generate an apparently complete assessment that fails to identify an important compliance issue.
And an AI agent assessing an investment opportunity can make equally damaging mistakes by overlooking financial risks, interpreting information incorrectly or expressing unwarranted certainty.
The appropriate controls must therefore depend on the purpose of the application.
This is why I believe our industry should move beyond asking how much of the crowdfunding process can be automated.
The more important question is how AI changes the quality, reliability and accountability of the process.
We should also be careful not to confuse technological sophistication with genuine investor protection.
An elaborate AI-based scoring system is not necessarily better than a simpler analysis if its assumptions cannot be explained or its conclusions independently verified.
Likewise, a faster regulatory review is not an improvement if efficiency comes at the expense of proper scrutiny.
Technology should be judged by the quality of the outcomes it supports, not merely by the amount of human work it replaces.
Crowdfunding Was Built on Collective Intelligence. AI Should Strengthen It.
One of crowdfunding's original promises was that the collective participation of many individuals could open up opportunities that traditional financial institutions overlooked.
That idea remains powerful.
But collective participation does not automatically create collective intelligence.
A thousand investors repeating the same unsupported assumption do not constitute a thousand independent assessments.
Likewise, an AI system generating an impressive report does not necessarily create genuine understanding.
The challenge is to build systems that combine information, experience and independent judgement in ways that improve decision-making.
At CrowdDiligence, our work with AI-assisted due diligence is still evolving.
We are continuously refining our agent's instructions, assessment methodology, verification practices and integration with community-based due diligence.
In parallel, we are developing a framework that could help regulated crowdfunding platforms apply AI to documentary and regulatory verification while preserving human accountability.
Through boomerang, we are also gaining experience with AI on the project creator's side, where the opportunities and risks are different.
We do not claim to have solved the challenges of AI governance or eliminated the risks of crowdfunding.
What we are doing is turning these questions into practical design decisions and exploring how they should work in real products.
After 15 years in crowdfunding, I see this as a natural continuation of the industry's original ambition.
First, technology made access to capital more democratic.
Now, AI gives us an opportunity to make access to knowledge, analysis and professional experience more democratic as well.
That may ultimately prove to be the more important development.
**The future of crowdfunding should not be about replacing human judgement with artificial intelligence. It should be about giving more people the tools to exercise better judgement.**
---
*Michael Eis is the founder of CrowdDiligence, an independent equity crowdfunding due diligence platform, and boomerang.dk, a Danish reward-based crowdfunding platform with approximately 15 years of operating experience. CrowdDiligence is a member of the Global Equity Crowdfunding Alliance (GECA).*
*CrowdDiligence provides research and analytical tools, not personalised investment recommendations. Equity crowdfunding involves substantial risk, including the possible loss of invested capital.*
Tags: AI, AI governance, due diligence, ECSPR, ECSP, investor protection, GECA