A practical guide to using general-purpose AI for Horizon Europe proposal writing
Large Language Models (LLMs) such as ChatGPT and Claude have transformed the way many researchers, consultants and innovation managers prepare grant proposals. They can generate text, improve writing, summarise scientific literature and help structure complex ideas in seconds.
Can they be used to write a Horizon Europe proposal?
Absolutely.
Can they replace an experienced proposal writer?
Not entirely.
Like any tool, they have strengths and limitations. Used correctly, they can significantly accelerate proposal preparation. Used carelessly, they can produce generic, inconsistent or even inaccurate content.
This guide explains how to use ChatGPT or Claude effectively throughout the proposal writing process.
Step 1. Start with a Clear Project Description
AI performs best when it understands your project.
Before asking ChatGPT or Claude to generate proposal sections, prepare a concise project description including:
- The problem you want to solve
- Your proposed innovation
- Scientific or technological objectives
- Expected impact
- Target users or beneficiaries
- Main consortium partners
- Planned work packages
The more context you provide, the more useful the responses will be.
Step 2. Build the Proposal Section by Section
Avoid asking the AI to write an entire Horizon Europe proposal in one prompt.
Instead, work incrementally.
Typical prompts include:
- Draft the Excellence section.
- Improve the State of the Art.
- Rewrite this paragraph to improve clarity.
- Suggest measurable project objectives.
- Strengthen the expected impact.
- Propose suitable project risks and mitigation measures.
Treat the AI as a collaborative writing assistant rather than an automatic proposal generator.
Step 3. Provide the Call Text
One of the biggest advantages of LLMs is their ability to analyse long documents.
Provide the relevant Horizon Europe call text and ask the AI to:
- Summarise the objectives.
- Identify expected outcomes.
- Highlight evaluation criteria.
- Compare your project against the call requirements.
- Suggest missing elements.
This helps ensure that the proposal remains aligned with the funding opportunity.
Step 4. Iterate Frequently
Strong proposals rarely emerge from a single draft.
Use ChatGPT or Claude to iterate repeatedly:
- Improve readability.
- Remove repetition.
- Simplify technical explanations.
- Adapt the writing style for evaluators.
- Strengthen transitions between sections.
Each iteration should improve clarity without changing the scientific content.
Step 5. Keep Humans in Control
AI should never replace scientific expertise.
Researchers and proposal coordinators remain responsible for:
- Scientific accuracy.
- Technical feasibility.
- Budget decisions.
- Consortium strategy.
- Ethical considerations.
- Compliance with the call.
AI can accelerate writing, but it cannot make strategic decisions for your project.
Advantages of Using ChatGPT or Claude
General-purpose AI assistants offer several important benefits.
Faster drafting
Generating a first draft takes minutes rather than hours.
Better writing quality
AI can improve grammar, clarity and readability while maintaining the original meaning.
Brainstorming support
When ideas become difficult to express, AI can suggest alternative structures, wording or examples.
Continuous refinement
Proposal sections can be improved repeatedly throughout the writing process.
Accessibility
Because these tools are general-purpose, many researchers already use them and require very little training.
Limitations to Consider
General-purpose AI also has important limitations.
No knowledge of your organisation
ChatGPT and Claude know nothing about your consortium unless you provide the information yourself.
Limited project memory
Maintaining consistency across a 70-page Horizon Europe proposal can become difficult when working through independent conversations.
No structured proposal workspace
The proposal remains a collection of prompts rather than a structured collaborative project.
No Horizon Europe workflow
General-purpose AI does not understand work packages, deliverables, milestones or proposal structure unless explicitly instructed.
Manual document management
Users must manually organise versions, prompts, documents and supporting information.
Digital Sovereignty Considerations
Many European universities, research organisations and public institutions are paying increasing attention to digital sovereignty.
Claude is operated by Anthropic, while ChatGPT is operated by OpenAI. Both are non-European providers.
For many organisations, this is not a concern. Others may have internal policies governing the use of external AI services, particularly when proposals contain confidential research ideas or commercially sensitive information.
Researchers should therefore ensure that the use of external AI tools complies with their institution's data governance and confidentiality policies before uploading proposal content.
A More Structured Alternative
General-purpose AI is an excellent starting point, but Horizon Europe proposals involve much more than generating text.
Dedicated solutions such as Grant Genius are designed specifically for the grant writing process.
Rather than relying on isolated prompts, Grant Genius provides a structured proposal workspace built around Horizon Europe. It keeps proposal sections organised, maintains consistency across the application, supports reusable project knowledge and allows proposals to be exported as standard Microsoft Word documents at any time—avoiding vendor lock-in.
Grant Genius also embraces open standards wherever possible. For example, scientific references can be imported through the BibTeX format, allowing researchers to continue using their preferred reference manager, whether it is JabRef, Zotero, Mendeley, Citavi or another compatible solution.
For teams that appreciate the flexibility of ChatGPT or Claude but need a workflow designed specifically for European grant writing, a dedicated platform can help overcome many of the limitations of a purely DIY approach.
Final Thoughts
Writing a Horizon Europe proposal with ChatGPT or Claude is entirely possible—and for many teams, it is already part of their daily workflow.
The key is to recognise what these tools do well: accelerating writing, improving clarity and supporting brainstorming.
At the same time, successful proposals still depend on careful planning, scientific expertise and structured collaboration.
General-purpose AI is an excellent assistant. Combining that AI capability with tools specifically designed for Horizon Europe can make the entire proposal development process more efficient, consistent and easier to manage.