Key takeaways
- Eva is the AI grant advisor inside Evalora, wired into the Evalora Engine, not a general chatbot bolted onto a website.
- It holds three layers of context at once: the call's criteria and weights, your scored analysis, and your organisation profile.
- Advice is tied to the criterion that is costing you points, with the fix written into your document rather than left in a chat window.
- Eva builds a lasting organisation profile, so the second proposal starts from everything it learned during the first.
- It advises and edits with your approval. It does not invent your project or generate a proposal from nothing.
Ask a general assistant to improve a grant application and it will do a competent job on the sentence in front of it. Ask it whether the application will pass the panel, and it has no way to know. It never read the call. It does not know that Impact carries a heavier weight than Excellence in your action type, that your section on sustainability answers a criterion worth five points, or that your organisation has run three similar projects that would make the capacity argument credible if anyone mentioned them.
Eva was built to close exactly that distance between good writing advice and advice that moves your score.
What Eva is, in one sentence
Eva is the AI grant advisor built into Evalora. It combines the evaluation criteria of your funding call, the scored analysis of your draft and your organisation profile to explain why you are losing points and to help you fix it in the document itself.
Behind Eva sits the Evalora Engine, the criteria driven evaluation system that reads the official call, builds the scorecard, scores your draft against it and runs the evaluator simulation. The Engine does the assessment. Eva is the part you talk to, and everything it tells you comes from what the Engine measured.
Plenty of tools now call themselves an AI grant advisor. What separates one from the next is not the model behind it, it is whether anything in the system has actually read the call you are applying to and measured your draft against it. Eva has the Engine for that.
Eva sits in a dock on every page of the app. On a results page it can be opened directly from any improvement suggestion, and it arrives already knowing which criterion that suggestion belongs to.
The three things Eva knows that a chatbot cannot
The difference is not the model. It is the context the model is given before it says a word.
1. The criteria, the weights and the rules of your call
Before Eva says anything about your text, the Evalora Engine has already read the official call document and extracted the evaluation criteria, their point weights, eligibility rules, budget limits, required sections and deadlines. Eva answers against that structure. When it says a paragraph is weak, it can name the criterion the paragraph is meant to satisfy and the points at stake. A general assistant can only do this if you paste the call in yourself, every session, and even then nothing holds the weights in place while you work.
2. Your scored analysis, criterion by criterion
Eva reads the analysis the Engine produced for your proposal: which criteria are strong, which requirements are unmet, where the compliance gaps are, and what the evaluator simulation predicted. That is why its first answer is usually specific. Not "consider strengthening your impact section", but a note about the missing quantified target in the criterion where your score dropped, and the sentence that would carry it.
3. Your organisation
Eva keeps a profile of who you are: team size, sectors, mission, budget range, past projects and the working principles you have told it about. It can also read documents you upload, such as a statute, an annual report or a previous application, and propose facts to add to that profile. When it later drafts a capacity paragraph, it can reach for a real project you ran rather than a placeholder.
How Eva works during a review
The flow is deliberately narrow. Eva is there to close the gap between a finding and a finished edit.
Step 1: You get a scored analysis
The Engine scores your draft against the criteria and lists improvements ranked by priority and point impact. Each one names the section it affects and what is missing.
Step 2: You open Eva on the suggestion you want to act on
Eva opens with the full context of that criterion and recommendation. You can ask why it matters, what an evaluator would look for, or how much of the point loss the fix recovers.
Step 3: Eva proposes the actual text
Rather than describing an improvement in the abstract, Eva drafts the change in your voice, grounded in your project and your organisation profile, and shows you where it goes.
Step 4: You accept, edit or reject, and the change is recorded
Applied changes go into the proposal document with a change log, so you can see what moved and why. Nothing is written to your document without your approval.
Step 5: You re-analyse and watch the score move
Every run is saved as a version. The point of the loop is not to feel productive. It is to see whether the score actually rose, and where the remaining points are hiding.
The organisation profile that keeps paying off
Most AI conversations are disposable. You explain your organisation, get an answer, close the tab, and explain it again next week. Eva treats that explanation as an asset.
As you talk, Eva saves standard facts to your organisation record and stores qualitative knowledge as organisational principles you can view and edit on the Organization page. That profile is scoped to your organisation alone. Nothing crosses between accounts. The practical effect shows up on your second application, when the capacity, governance and track record sections stop being written from scratch.
Eva compared with a general AI assistant
| Capability | General AI assistant | Eva in Evalora |
|---|---|---|
| Knows the call criteria and weights | Only what you paste in, each session | Extracted from the call and held throughout |
| Knows your current score | No score exists | Reads the per-criterion analysis and predicted score |
| Knows your organisation | Re-explained every conversation | Persistent profile and organisational principles |
| Ties advice to points | General writing advice | Advice ranked by criterion and point impact |
| Edits the actual document | Copy and paste by hand | Applies approved changes with a change log |
| Tracks what changed between drafts | No memory of prior edits | Every run saved as a version |
| Reads your statute and past proposals | Only if re-uploaded each time | Reads uploaded documents and proposes profile facts |
| Works in your language | Yes | Yes, replies in the language you write in |
For a fuller treatment of where general models genuinely help and where they structurally cannot, read what ChatGPT, Claude and Gemini miss on grant proposals.
What Eva will not do
- It will not invent your project. Eva works from your draft, your documents and your profile. If a fact is not there, it asks rather than fabricates.
- It will not overwrite your work silently. Changes are proposed, reviewed and logged. A field that already has content is never replaced without confirmation.
- It will not guarantee funding. No tool can. It closes the avoidable gaps so the panel judges your project rather than your paperwork.
- It will not interrupt you. Eva stays out of the way during an analysis run and caps its own proactive suggestions.
Frequently asked questions
What is Eva in Evalora?
Eva is the AI grant advisor built into the platform, wired into the Evalora Engine. It works with the criteria extracted from your funding call, the scored analysis of your proposal and your organisation profile, so its advice is tied to the points you are losing.
How is Eva different from ChatGPT, Claude or Gemini?
General assistants begin without your call, your score or your organisation, and you have to rebuild that context by hand every session. Eva already holds it, which is why it can name the criterion, the gap and the fix.
Does Eva write my proposal for me?
No. It proposes criterion-specific edits based on your own material, and you accept, edit or reject each one.
Does Eva remember my organisation between sessions?
Yes. Facts and working principles are saved to your organisation and reused on later proposals. Nothing is shared between organisations.
What languages does Eva speak?
The interface follows your browser language with English as the default, and Eva replies in whatever language you write in.
Is my document confidential?
Yes. Uploads are processed securely and used to produce your results. As stated in the privacy policy, Evalora does not use your uploaded documents to train AI models without your explicit, separate consent, and nothing is shared between organisations.
Work with an advisor that has read the call
Upload your funding call and draft, get a criteria based score, then let Eva help you fix what costs you the most points.
Start Free Trial