Our approach
AI only becomes truly helpful once it understands your work
The KIAD approach helps you move beyond experimenting with AI in a chat window and build it meaningfully into your grant work.
AI can do an astonishing amount. It drafts project ideas, describes target groups, answers application questions and suggests implementation plans. But in grant work it isn’t enough for a text to sound plausible. It has to fit your organization, your target group, your local context and what you can actually deliver. KIAD combines solid grant craft with a clear structure for using AI: Context, Input & Interaction, Assistants and Documentation.
AI can do a lot. But without context, it won’t be your proposal.
Many AI results are impressive at first glance. Suddenly a project sounds polished, the target group is neatly described, the activities seem logical. And yet something is missing: a good proposal doesn’t describe just any good project. It describes your project.
It has to show why your organization in particular can deliver it, what your target group really needs, what experience you bring, what the conditions on the ground look like, which partners are involved and why this proposal fits this exact funding program. AI can support all of this very well, but it needs your groundwork first. Without context it fills the gaps on its own, and the result sounds professional but stays interchangeable.
Key takeawayAI shouldn’t write just any proposal. It should help you develop yours: better, clearer and faster.
Born out of practice
KIAD wasn’t designed on a drawing board. It comes out of Mira’s AI Accelerator cohorts with nonprofit organizations. After one of the first cohorts she talked to many participants one on one: What did you really learn? What helped you in your day-to-day work? Where did “interesting” actually turn into “useful”?
It was never the one perfect tool or the one perfect prompt that made the difference. It was that organizations understood how much good context matters, learned to work with AI in dialogue, turned recurring tasks into their own assistants and documented what worked, so that not everything starts over with every new proposal.
Key takeawayKIAD is not a theoretical model. It’s the distilled experience of real proposals, real organizations and real learning.
The KIAD approach in four steps
The order matters. Many people start straight with prompts or tools, because that’s where results show fastest. But without context even good prompts stay limited, unclear interaction doesn’t improve results, and if nobody documents what works, a lot of knowledge disappears into individual chats. KIAD helps you use AI as a way of working, not as a collection of tricks. (The K stands for Kontext, German for context: the approach was developed in German.)
K
Context
You prepare your knowledge, your proposals and your materials so that AI can work with them meaningfully.
I
Input & Interaction
You learn to set good tasks, review results and keep working with AI in dialogue.
A
Assistants
You turn recurring tasks into assistants, agents or skills that hold a process, ask follow-up questions and guide you through it.
D
Documentation
You record what works, so that knowledge stays usable across your team.
K for Context
Understand first, then let it write
The results of an AI tool are only as good as the information you give it.
Grant proposals are almost never about generic text. They're about your organization, your target groups, your past projects, your language, your region, your funding logic and the requirements of one specific funding program. That's why good AI work doesn't start with the perfect prompt. It starts with context, for example:
- previous grant proposals
- project descriptions
- mission statements and strategy papers
- funding requirements
- information about your target groups
- studies, reports or regional data
- text modules that work well in your organization
- lessons learned from earlier projects
When this information is well prepared, AI has to guess less. The results fit better, are more grounded and much closer to your actual work.
Key takeawayGood context is the difference between a plausible AI text and a draft that truly fits your organization.
I for Input & Interaction
AI is not a search box, it's a dialogue
At first, many people treat AI like a better search engine: a short question in, an answer out. For grant proposals, that's rarely enough.
If you want specific, usable results, AI needs clear tasks, good follow-up questions and feedback. Think of it as a new colleague who should contribute professionally but first needs to understand what it's all about. A good prompt doesn't just describe what should come out. It also explains:
- the context you are working in
- the role the AI should take on
- the information it needs to take into account
- the language and tone that fit
- what to pay particular attention to
- how the result should be structured
And the first draft is usually not the end. You follow up, sharpen, correct, have it double-check and ask for alternatives. That's how an AI answer turns into a usable way of working.
Key takeawayGood AI work doesn't come from one perfect prompt. It comes from good interaction.
A for Assistants
AI that thinks along with the process
In grant work, many tasks come back again and again: structuring project ideas, checking funding requirements, revising drafts, giving feedback, sharpening the logic of impact, checking documents for gaps.
This is where your own AI assistants pay off. Depending on the tool they're called custom GPTs, assistants, agents, skills or workflows, and the names change fast. The idea stays the same: AI shouldn't just answer single questions, it should carry part of a process. An assistant can help with:
- developing a project idea
- giving feedback on a proposal draft
- checking funding criteria
- structuring application questions
- revising texts
- comparing drafts with previous proposals
- preparing internal coordination
- actively asking for missing information
One thing matters: an assistant doesn't replace your judgment. But it can prepare, sort, question and speed up tasks that would otherwise cost you a lot of time, again and again.
Key takeawayA good AI assistant doesn't simply write for you. It holds the process, asks better questions and makes your expertise easier to use.
D for Documentation
So AI doesn't start from scratch every time
Many organizations test AI here and there: a good prompt here, a helpful chat there, a half-working workflow somewhere in a browser tab. If nobody records what worked, everything starts over with the next proposal.
Documentation doesn't mean writing enormous manuals. It means making the most important lessons usable:
- Which prompts work well?
- Which materials does the AI need?
- Which tasks are actually a good fit?
- Where do errors creep in?
- Which information must never go into an AI tool?
- How do we review results?
- How can colleagues keep working with the templates?
That way AI doesn't just make life easier for individual people. It becomes a way of working the whole team can share and build on.
Key takeawayDocumentation turns individual AI experiments into a reliable way of working.
Examples
What KIAD can look like in practice
Context from past proposals
An organization collects successfully submitted proposals and prepares them so that AI can recognize typical structures, wording and project logic. New drafts come out much closer to the organization's own language and way of working.
A guideline assistant for funding programs
An assistant keeps the relevant guidelines, program logic and formal requirements in view while you write, so individual sections can be reviewed and sharpened far more precisely.
A feedback assistant for the final polish
A dedicated assistant reviews proposal drafts against typical criteria: clarity, funding logic, impact goals, gaps, formal requirements and fit with the target group.
For everyone who wants to build AI in, not just try it
KIAD is especially helpful if you see grant work as a recurring process, not a one-off task. It’s a good fit if you:
- write grant proposals regularly
- develop projects or prepare funding ideas
- want to make better use of past proposals and existing knowledge
- want to introduce AI to your organization or use it in a more structured way
- want to show colleagues how to use AI well and responsibly
- are looking for a clear way of working, not just new tools
It’s less suitable if all you want is a quick prompt that writes a finished proposal without any context. That’s exactly what this is not about.
KIAD is the common thread behind our offers
Not every offer covers the whole approach, and that’s deliberate. Our beginner courses help you build confidence with ChatGPT and other AI tools first. Digital products such as databases and templates support single building blocks: good context, clear prompts, assistants or documentation. Live seminars will pick up individual steps in depth.
You don’t have to implement the whole system right away. Start with the building block that fits your work now: a better start, a good context document, a clearer prompt, a first assistant or simple documentation for your team.
Ready to put AI to work with a system?
One good next step is enough: better context, a clearer prompt, a first assistant, or simple documentation of what works.