Funding research, on autopilot
An internal tool that watches the EU funding portals Agriventures used to scan by hand, decides which calls are actually worth their attention, and hands the team a clean shortlist with the details already pulled out.
Client
Agriventures
Services
Data pipeline, AI integration, Internal tools
Industry
Agritech / EU funding
Date
2025 — Ongoing
Challenge
Agriventures helps farms and agribusinesses apply for EU grants. The hard part isn't the application — it's finding the calls in the first place.
The team was spending hours every week opening the same handful of funding portals, scrolling through new calls, and copy-pasting the bits that mattered: amount, eligibility, deadlines, sector tags. Most calls weren't relevant. The few that were got buried under the rest.
Manual triage doesn't scale, and each missed call is a client opportunity lost. The team needed an always-on assistant that watched the same portals they did — and only surfaced what was worth their attention.
Solution
We built the team an assistant that handles the part of their job that wasn't a good use of their time — finding the calls — so they can focus on the part that is: advising clients.
The same portals Agriventures used to open one by one are now monitored continuously in the background. New calls are detected as they appear, the noise is filtered out, and the team only sees the ones that look relevant to the kinds of clients they serve — so the hours spent each week scanning the same sources are gone, and the calls that used to slip through during busy weeks don't slip through anymore.
Each call shows up already broken down into the details that matter — amount, eligibility, sectors, deadlines — so the team isn't reading long official PDFs to extract a handful of facts. They open the tool, review what's new, and decide what to send to which clients. Research turns into a continuous, reliable feed instead of a manual sprint, and Agriventures can take on more clients without growing the research team to match.
Tech & outcome
Stack
What was hard
It had to cover many portals without falling behind
Agriventures watches a lot of sources, and a slow or sequential pipeline would mean falling behind real-world publishing. We designed it so adding another portal doesn't slow anything else down — the team can always say yes to a new source.
AI summaries had to stay affordable at scale
Letting AI read every page sounds great until the bill arrives. We carefully filter what actually goes to the model, so the team gets clean, structured summaries without the cost growing every time a portal posts something new.
Outcome
Hours of weekly manual research are now a continuous feed running in the background, more relevant calls reach clients while fewer opportunities slip through during busy weeks, and the cost of using AI stays predictable as the team scales rather than running away with the workload.
Got an idea in the dark? Let's bring it to light.