# AGIgent

> AGIgent is a reviewed AI agency in Budapest, Hungary, listed in the AIAgencies.eu register and last assessed 2026-09-03.

Canonical page: https://www.aiagencies.eu/agency/agigent
Generated: 2026-09-28 · Curated by Gabor Kiss · No paid placements — inclusion, score, and rank are never sold.

## Facts
- Location: Budapest, Hungary
- Website: https://agigent.ai
- Hourly rate: €85-130/hr (directory-listed)
- Team size: 2-9
- Founded: 2024
- Services: AI Development, AI Consulting, AI Agents
- Industries: Manufacturing, Public Sector
- LinkedIn: https://www.linkedin.com/company/agigent

## About
AGIgent is a Budapest AI engineering firm working on physics-informed models, satellite and sensor signal processing, energy forecasting and long-horizon agent frameworks for research and industrial clients.

## Documented clients
Documented engagement: Ideas Science. It carries a source URL on file.

## Curator assessment
Reviewed 2026-09-03 by Gabor Kiss against published evidence.

The one published study with numbers is anonymized: a CLIP pre-filter cascaded in front of a real-estate data provider's legacy image-matching algorithm, routing 90% of comparisons away from it for a 70-80% cut in processing cost at a 1% rise in error rate.

The rest of the visible portfolio is staged research work—the GRIDFOR grid-forecasting project completed an ESA Kick-Start in August 2025 and lists its feasibility study as in execution, the PACAI antenna-calibration work is described as submitted, and the retail queue predictor was proven in test stores—so ask which of these has an operator running it today before you budget for a production system.

Best for research institutes, space programs and grid operators funding a feasibility-to-pilot track where physics-informed modeling is the point, rather than buyers who need a deployed system on a fixed date.

### Key strengths
- Partner roster names ESA and RAL Space (STFC, UKRI), with AGIgent as prime contractor on the ML4QS quantum-magnetometer denoising work
- Three named founders with LinkedIn profiles, a stated address at Faulhorn Labs in Budapest and a named investor (Stockhorn Capital)
- The engineering writing explains method choices—the cascade design in the image study, the tokenization and transparent-model posts—rather than name-dropping models

### Good to know
- The founder cards on the homepage still carry lorem-ipsum placeholder quotes and the blog stops in mid-2024—ask what the team looks like beyond the three founders
- MAVIR, E.ON Hungária, ELMŰ, HungaroMet, ALTEO and Veolia appear as feasibility-study participants, not as paying clients—confirm the commercial relationship before counting them as references
- The image-comparison study names no client and the RF signal work is under confidentiality—ask for a speakable reference

---
Source: AIAgencies.eu — the curated register of European AI agencies. Rate provenance: "agency-confirmed" = disclosed via submission or claimed profile; "directory-listed" = Clutch or two agreeing B2B directories (lone unverified sources are never written). Full method: https://www.aiagencies.eu/methodology · Rate index: https://www.aiagencies.eu/rates