It’s summer. You’ve earned some time off with friends and family, not all-nighters on Horizon Europe.
WinGrants AI analyses the call text, extracts the evaluation logic, grounds your concept in evidence, maps related projects, detects consortium gaps, and drafts and evaluates a structured Part B proposal aligned with the EU evaluators’ criteria — so you only tweak and approve it, instead of writing it from scratch.
From solo PIs and early-career researchers to 30-partner consortia — WinGrants AI adapts to how your team already works.
Turn your R&D roadmap into non-dilutive EU funding. Match strategic priorities to live calls and ship submission-ready proposals — without pulling engineers off the product.
Back every researcher without overloading the grants office. One shared workspace, your institutional knowledge on tap, and consistent guardrails across every submission.
You’ve got the client relationships — we give you the firepower to deliver on them. Win more bids on a lean team, with no new hires and no months spent turning juniors into seniors. Partner with the engine itself, not a basic chatbot, keep your margin on top, and outbid agencies several times your size.
Researchers who use WinGrants work at —
Every missed deadline, every rejected bid — they share the same failure modes. We designed WinGrants AI around each one.
Mapping state-of-the-art, TRLs, regulations, policy context, and already-funded projects burns 3–5 days of PI time before a single word hits the page. With the 2026–27 programme cutting topics ~35% and writing them broader, you can't afford to look derivative.
A single call can lock your best people for 4–6 weeks — off research, teaching, supervision, and the next deadline. Two-stage calls double everything: two drafting cycles, two internal reviews, two chances to lose momentum.
Most teams build consortia from whoever replies first or owes them a favour. That fills the partner table — but quietly weakens credibility, country balance, and the Implementation score.
8 contributors, 3 weeks, 5 writing styles — and the proposal feels stitched together. Add generic AI on top and it sounds like everyone else's. Evaluators notice both.
If odds are 5–10%, one "perfect" proposal isn't a strategy. You need more shots, but better shots — higher throughput and evaluator-grade quality control, without flooding the EU R&I system with weak drafts.
An LLM chatbot can draft text. It won't own the strategy, check the rubric, verify citations, fix the budget logic, or align work packages. Assisted is not done.
Unpublished research ideas, consortium strategy, partner data, and pre-patented technical novelty don't belong in random personal LLM accounts. The real question isn't AI or no AI — it's governed AI versus shadow AI.
Large R&I agencies run €10–30k a bid — or take a slice of the award — and you wait in their queue. The expertise leaves when the invoice clears, and every new proposal restarts the meter. You're renting wins, not building the capability to repeat them.
Tour our workspace — from Opportunities Radar to AI Evaluation. Click any tab to see what your team gets on day one.
Drives technological progress through challenge-oriented, AI-powered solutions across three strategic sectors.
Multi-scale, multi-organ in-silico models for tailored prevention and diagnosis across high-burden diseases.
A telco-edge-cloud, secure multi-provider system hosting network functions and workloads beyond connectivity.
AI methodologies to guide treatment decisions and accelerate the development of better, personalised medicines.
Expected Outcome: Policymakers and social partners gain insight into the scope and characteristics of un(der)declared work and the actors involved.
Approach: A strong multidisciplinary design couples indirect survey methods, behavioural science, and an agent-based policy simulator across heterogeneous economies.
The "Incoming core assessment" element is the single most dangerous gap. Without a defined mass-sorting workflow, the proposal will fail the Excellence pillar.
| Acronym | ReXeR |
| Title | Remanufacturing for integrated value chain innovation |
| Call | HORIZON-CL4-INDUSTRY-2025-01 |
| Action type | IA |
| Duration | 42 months |
Book a 30-minute demo — then run your free benchmark on a proposal you already submitted: our draft against yours, our AI scorecard against the official EU ESR.
Specialised AI agents work in sequence across every stage of the proposal lifecycle — from call discovery to redrafted submission.
Scans open and upcoming calls across Horizon Europe, Digital Europe, EIC, UNITE, and non-Horizon EU programmes — then matches them to your org profile, expertise, and geographic scope.
Validates your concept note against call requirements before anything else gets built. Unclear objectives, scope mismatches, weak alignment — all flagged with specific fixes. Iterate until it passes.
CORDIS-powered partner matching from your validated concept note. Ranked partners with role justifications and contact details for direct outreach.
Auto-maps the scientific landscape and identifies research gaps — then evaluates each gap on time, cost, and technical feasibility so you pick the angle most likely to land.
Pre-mortem for every call. Frames the angle, sizes the budget, structures WPs, flags risks — and shows you how to position and differentiate from the few projects already funded in your area.
Drafts every section to the official template, then critiques and rewrites its own work against the ESR rubric — ingest → draft → evaluate → decide → redraft. Each loop runs on a different state-of-the-art model, so a blind spot in one is caught by the next.
The overarching ambition of ATLAS-WIND is to transform the inspection of offshore wind infrastructure from a human-limited, reactive process into a continuous, agentic AI-driven workflow delivering measurable reliability gains across the European offshore fleet.
We pursue 4 scientific-and-technological objectives, each aligned with destination DIGITAL-2026-02 expected outcomes. O1 advances the state of the art in domain-adaptive vision-language models
Generates the high-definition visuals every Horizon Europe proposal needs — drop-in ready for the official template. Saves a week of design work, gives evaluators the clarity they reward.
Two ways to pressure-test a draft before you submit — both calibrated against real Evaluation Summary Reports.
End-to-end coverage is what separates purpose-built tooling from everything else.
Swipe horizontally to compare all five options. The capability column stays pinned.
| Capability | WinGrants AIEnd-to-end AI proposal engine | R&I ConsultanciesDone-for-you grant writing agencies |
Solo effortYou + your team | General-purpose LLMsClaude · GPT · Gemini | AI toolsOther grant-writing AI tools |
|---|---|---|---|---|---|
| What it does the end-to-end workflow | |||||
Scope & automationCo-pilot vs. autopilot |
End-to-end Co-pilot and autopilot — the full proposal, like an R&I agency at software speed | Full-service — but human-paced & opaque | You do every step | Co-pilot only — you own the process | One step each |
Call discoveryFind the right grant, fast |
Personalised Matched to your profile, with competition estimates | Expert matching, on their schedule | Manual portal browsing | Often hallucinates calls | Basic keyword search |
Concept note validationCatch issues pre-drafting |
Pre-draft check Pass/fail + specific feedback + AI-recommended fixes | Senior review — if it's in scope | Subjective peer review | Generic feedback, no rubric | ✗ Usually not offered |
Consortium buildingRight partners, right roles |
CORDIS-ranked Ranked partners with role justifications + contacts | Strong networks — their contacts, not yours | Days lost on B2Match events | ✗ No CORDIS access | Basic directory lookup |
Competitive intelligenceGo/no-go & mapping |
Go / no-go Verdict, competitor mapping, risk pre-mortem | Thorough — billed by the hour | Days of desk research | Often outdated | ✗ Usually not offered |
Budget allocationHE financial rules |
Rules-enforced Auto-allocated to HE financial rules | Done for you — logic stays hidden | Complex spreadsheets | ✗ Doesn't apply HE rules | Static templates |
Proposal draftingExcellence / Impact / Impl. |
Multi-pass Single-sitting draft, then AI redrafts in-loop | High quality — 6–8 weeks, premium fee | 6–8 weeks of writing | Fast but generic, one-shot | Editor + partial generation |
Evaluation & scoringESR-aligned scorecards |
ESR-aligned 300+ criteria scorecard, calibrated to the panel | Experienced eyes — no formal rubric | Subjective self-review | ✗ No ESR rubric | Basic AI review |
| The outcomes why teams actually switch | |||||
Speed of delivery |
Hours, not weeks ~3–4 hours to a full draft | Weeks; capped by consultant capacity | 4–6 weeks per call | Hours of prompting, still partial | Per-step, no full draft |
Proposal quality |
13/15 drafts First drafts, 60–80% overlap with historic ESR scores | Strong — varies by who's assigned | Swings by team & deadline | Generic, off-rubric prose | ✗ No evaluator calibration |
Cost of service |
≤ €1k / proposal No retainer, no success fee | €10–30k upfront + 2–7% success fee | High opportunity cost | Subscription + your time | Several tool licences |
Volume & scalability |
Parallel Several live calls at once | Capacity-limited; they pick the wins | One proposal at a time | Bottlenecked by you | One step at a time |
Consistency & compliance |
Repeatable Rules-enforced and template-compliant, run to run | Hinges on the individual writer | Quality varies run to run | Hallucinations & drift | Inconsistent coverage |
Data sovereignty & IP |
EU-hosted On-prem & local-install options; your IP never trains public models | ✗ You hand over IP & strategy | Varies | ✗ Often US-hosted, trains on data | ✗ Vendor cloud only |
Guarantee |
100% guarantee Satisfaction Guarantee — Self-service, under Fair Use | Success-fee only — partial at best | ✗ No guarantee | ✗ No guarantee | ✗ No guarantee |
Three setups, one engine. Self-service for solo PIs, AI Enablement for research teams, Enterprise for institutions where IT won't sign off on managed AI.
Answer four quick questions — we’ll highlight your best fit and exactly what you trade for it.
Buy one credit, get one complete proposal. No subscription, no commitment.
A private, single-tenant workspace your whole team works in year-round.
Runs where IT requires it — your infrastructure, sovereign cloud, or campus compute.
Every hard question a buyer actually asks — with the honest answer, not the brochure version. Filter by topic.
What you’re really worried about
Ten partners’ confidential strategy leaking — or quietly training someone’s model.
The honest answer
Your inputs and outputs are yours. We never reuse your concept, consortium or drafts for another client, and we don’t train models on your data. WinGrants is ISO/IEC 27001:2022 certified and follows information-security and data-privacy best practices; all processing is in the EU and nothing is retained beyond what’s needed to deliver your work. To remove any conflict of interest, the founder has permanently withdrawn from Horizon Europe as an applicant — we run no competing proposals. For maximum isolation, AI Enablement uses a bring-your-own-keys model: your data is processed entirely under your own accounts — database and AI providers alike — so nothing is shared with us at all, and Enterprise adds a fully private on-premise deployment.
Data-flow + Trust & Security hubRaised by an industrial R&D consortium partner
What you’re really worried about
Procurement and legal need contract-grade clarity, not a badge.
The honest answer
Compliance is a data flow and a contract, not a logo. All processing happens in the EU (AWS Frankfurt) under GDPR, and we don’t train on your data — our model providers are contractually bound not to either. Partner research uses only public, business-level information (organisation websites, work emails), not personal data. Confidential consortium material stays within your workspace, and a private on-premise option exists for security-sensitive groups. We’re ISO/IEC 27001:2022 certified and operated by Healthdev OÜ, a registered Estonian (EU) entity; a Data Processing Agreement is available. Full detail is on our Trust & Security page.
GDPR FAQ · DPA · sub-processor listRaised by universities & RTOs
What you’re really worried about
Who can see our workspace — and for how long?
The honest answer
Everything runs in the EU on AWS (Frankfurt). Each account is fully isolated — only you can access your data — and we don’t train on it. Data isn’t retained beyond what’s needed to deliver your proposal. We maintain a live controls dashboard (policies, access, change management, vulnerabilities) and we’re ISO/IEC 27001:2022 certified. For security-sensitive organisations, a private on-premise deployment is available, running entirely on your own servers (no GPUs required). Full detail is on our Trust & Security page.
Live controls dashboard + ISO certificateRaised by security-conscious research offices
What you’re really worried about
Why pay you for something my €20 tool already does.
The honest answer
A single model gives you one model’s view. WinGrants orchestrates several at once: up to three models collaborate on each draft, and up to ten form the evaluation committee — so your proposal gets their combined strength rather than any one model’s ceiling. That multi-agent system, plus the Horizon-specific workflow around it (call analysis, structured drafting, a ~350-criteria scorecard, redraft loops and consortium logic), is what you can’t reproduce by prompting a single chatbot — and we maintain it as templates and rules change.
Side-by-side: generic model vs WinGrantsRaised by a university research office
What you’re really worried about
Only funded proposals hold the winning pattern.
The honest answer
We didn’t, and we won’t claim we did. We engineer context from public call texts, official templates and evaluation criteria. Most proposals don’t fail because someone lacked secret examples — they fail obvious criteria, weakly, and that’s exactly what the system catches.
Data-sources map · rules traced to criteriaRaised by research managers
What you’re really worried about
Is this a forced subscription, and how does the cost scale?
The honest answer
WinGrants is usage-based, not a forced monthly subscription. The standard full package — AI drafting, redraft loops and three evaluations — is €1,000 per proposal, and the cost can be split across your consortium. Pre-proposal tools (Opportunities mapping and the Concept Note Health Check) are free to start. For high-volume teams there’s an AI Enablement licence from €15,000 plus a €5,000 one-time setup; universities wanting a private, IP-secure on-premise deployment can license it up to €30,000 plus a €5,000 setup. For context, R&I consultancies, our direct competition, typically charge €10,000+ per proposal plus 2–7% commission.
Full pricing breakdownRaised by research-team leads
What you’re really worried about
Anchored to consumer-AI pricing.
The honest answer
Different unit. One €1,000 credit covers a submission-ready proposal workflow — compare it to a €10,000–100,000 consultant, or to the cost of one preventable rejection on a multi-million-euro call. Write one proposal a year and a chat subscription is cheaper; if a rejection costs you the grant, it isn’t.
ROI: hours saved + consultant cost avoidedRaised by a deep-tech SME
What you’re really worried about
Seasonal cashflow and internal budget approval.
The honest answer
Reasonable — so Self-Service stays credit-based with no subscription commitment. A team can prove value on one real call before anyone signs an annual line item, and walk into the budget meeting with a benchmark rather than a brochure.
Single-call pilot + ROI one-pagerRaised by a research-team lead at an industrial partner
What you’re really worried about
New-vendor risk — is this even a real business?
The honest answer
We launched recently and we won’t fake a victory lap — no invented logos, no named wins we haven’t earned. What we can show is the work. Start with a free trial on a call you’ve already submitted to: we generate the proposal and you compare it against the one you actually submitted — so you see both the jump in quality and how fast you get there. If you used an agency or consultancy on that call, it’s also a direct read on how our quality and speed compare to a paid service provider. Beyond that, we have paying customers at universities across Southern and Northern Europe. And on reliability: we ran an experiment evaluating proposals the EU had already scored — on almost all of them our evaluation flagged more issues than the official ESR did, with 60–80% overlap with the real EU ESRs. That depth comes from up to ten AI models reviewing your proposal separately and aggregating their findings into a single scorecard — which is why it surfaces weaknesses even the EU evaluators missed.
Book a free benchmarked trialRaised across multiple early calls
What you’re really worried about
Where are the wins and references?
The honest answer
We split this honestly. Funded outcomes take time — Horizon cycles run many months and we launched recently, so we won’t claim funded wins we haven’t earned yet. Paying customers: yes — universities across Southern and Northern Europe already run on WinGrants. Success evidence you can verify yourself: a free trial on a call you’ve already submitted to lets you compare our AI draft against your submitted proposal, and our evaluation against your official EU ESR — quality and speed side by side, plus a read on how we compare to an agency if you used one. And our evaluation is measurably reliable: on proposals the EU had already scored, our scorecard flagged more issues than the official ESR on almost all of them, with 60–80% overlap with the real ESRs — because up to ten AI models review your proposal separately and aggregate their findings into one scorecard.
Run it free on your own past callRaised by prospective pilots
What you’re really worried about
The evaluation is circular and not credible.
The honest answer
Scoring isn’t the drafting model marking its own homework. Evaluation runs as a separate multi-model committee against a granular rubric of ~350 criteria mapped to the official Excellence, Impact and Implementation logic. We’ve tested and tuned it across 20–30+ iterations, and it now flags more issues than the official EU ESR does — we see roughly 60–80% overlap with real EU ESRs, with our scorecard catching additional weaknesses the official review missed. Treat the red flags as more important than the number.
ESR-overlap benchmarkRaised by a research-SME founder
What you’re really worried about
Polish hides weak consortium logic, poor fit, shallow ambition.
The honest answer
We separate writing quality from funding strategy, and check the strategy first: call fit, consortium fit, state-of-the-art gap, work-plan credibility, impact pathway and the objections an evaluator will raise — before any redrafting. The most dangerous draft is the one that reads beautifully and says nothing.
Polished-but-weak vs strategically-strong exampleRaised by coordinators & senior writers
What you’re really worried about
Liability quietly transferred to me.
The honest answer
You are — the tool is decision-support, and you review and approve every section before submission. Research notes are source-grounded, citations are flagged for verification, and the ~350-criteria scorecard surfaces weak claims and unsupported references early. AI should cut the review burden, not remove accountability.
Citation-verification workflow + QA checklistRaised by consultancies
What you’re really worried about
Using us might be non-compliant or look AI-generated.
The honest answer
The rules people cite restrict EU evaluators using AI during peer review — not applicants using AI to prepare a submission. The EU’s Responsible Use of AI guidance doesn’t prohibit AI-assisted writing — it requires that you don’t plagiarise and that you represent references and contributions honestly. WinGrants is built for exactly that: text is grounded in your project and sources, with citations, and the multi-model redraft loop avoids the repetitive style that ‘looks AI-generated’. You remain the author and are responsible for the final submission.
AI-use policy + the actual guidelineRaised by a university research office
What you’re really worried about
Expected a done-for-you agency — or afraid of being replaced.
The honest answer
You do — WinGrants accelerates the work, it doesn’t replace the researcher. A Horizon proposal has traditionally taken around four weeks of writing; with WinGrants you can spend one to two weeks perfecting it instead, focused on the final ~10% that actually differentiates you. The tool drafts ~90% from your concept note, consortium and notes; you own the science, the strategy and every final decision. Nothing is submitted without your review and approval.
What we do vs what you doRaised by principal investigators
What you’re really worried about
Losing authenticity and precision — or a templated draft.
The honest answer
No — two organisations on the same call get very different proposals, because the output is built around your inputs: your concept note (followed line-by-line), your consortium and partner profiles, and your Research and Strategy Notes, personalised to your organisation’s expertise, track record and scientific angle. Drafting runs through multiple models in a redraft loop, avoiding the generic ‘one-model AI’ fingerprint. You should recognise your own science in the final draft — generic AI text is exactly what evaluators punish, so the tool pushes for specificity, not polish.
Voice-preservation before/afterRaised by principal investigators
What you’re really worried about
Will it actually cover my instrument?
The honest answer
Today the platform covers Horizon Europe RIA and IA calls — the consortium-based collaborative schemes it’s purpose-built for, in English. Coming soon: EDF, Digital Europe and EU4Health. Our priority is and will remain consortium-based grant schemes, where consortium-building, multi-partner drafting and evaluation give the most leverage. Infrastructure-type calls with confirmed partners are handled well, since the tool keeps your fixed partners and personalises only the rest.
Supported calls & roadmapRaised by research offices across instruments
What you’re really worried about
Can I try it before committing?
The honest answer
Our trial is a structured, demo-gated benchmark on a call you’ve already submitted to: we generate a WinGrants proposal and evaluation for that past call so you can compare directly — our draft against your submitted proposal, and our AI scorecard against the EU ESR you received. Two honest reasons for the gating. First, each trial draft costs us real money to generate, so we grant trials selectively to teams seriously evaluating us — the demo is how we gauge that. Second, we don’t offer free trials on upcoming calls because that cannibalises paid work: historically, free drafts on live calls got used once and never came back. And to be clear, none of this is about your data — everything you share is deleted at the end of the trial and exists only to help you answer the question that actually matters: how much time and cost would you save if day one started from a draft that’s 80–90% there, so your team spends its time perfecting the proposal instead of writing it from scratch? Most teams pick a proposal from two or three years ago, so no current research direction is exposed — and everything runs under our zero-data-retention trial terms. The whole flow is productized in-app: book the demo, tick the trial consent, upload your materials, and each deliverable unlocks automatically. For a live call, a single self-service credit (€1,000) runs the full pipeline end to end.
Book a benchmarked trialRaised by first-time users of the tool
What you’re really worried about
Does it actually help me find partners, or just draft?
The honest answer
The Consortium Builder analyses your call and concept note, predicts the likely work packages, then mines the CORDIS database of 25,000+ organisations with proven EU-project experience to suggest partners matched to each work package. It builds partner profiles, shows geographic distribution, and surfaces verified contacts so you know who to approach. It’s a software tool, not an agency — it identifies and shortlists partners; you make the outreach. If you already have a strong network, skip this module and use only drafting and evaluation.
Consortium Builder overviewRaised by a large-consortium coordinator
What you’re really worried about
Drafting speed isn’t my actual pain.
The honest answer
Agreed — and we don’t pretend faster drafting is the whole story. We target the consolidation: consortium mapping, role and budget architecture, contradiction detection across partner inputs, and change-aware redrafting. The real bottleneck is messy multi-partner integration — that’s the part we de-risk.
Messy-input → structured-output demoRaised by a large-consortium coordinator