Dunes AI

work in progress

• Edtech. New batch at Edu Challenger, 10 teams join pre-accelerator: ScreenTreat, Wisit, Unive, Buildibots, Expy, Bound AI, Impact Solutions Partners, Gilia, Create Power House. Unive AI announces investment from FIRSTPICK, no time to waste (see below). Pod with Edit AI founders (operating as venture builder, prototype before company). Kris Vasiliauskas talks about their journey with edtech ventures (in LT).
• Advanced electronics. As we read that Apple challenger Nothing raises $200m promising ‘AI-native’ devices: 8Devices approaching EUR 20M revenue. Teltonika unveils four production plants.
• Health. Carelybot is a new monitoring device for healthcare. Estonian SafePAS (BSV Ventures) has received a €2.5M grant from EIS to further develop its DrugHunter rapid drug test.
• Consumer – use this deck as an inspiration, and go, build something. There is something being built already, but we are probably dismissive, bec not clear if that turns into a massive player. Burga will become a major brand – it's more than "ecomm" – an established brand and proprietary production. Also, considered small now, but – Brite has an impressive PMF, Traxlo has already signed 30k (!) gig workers, we might get surprised.
• Media. News aggregation is extremely hard, but also very appealing to founders? with intouch app, Liutauras and the team are making another effort. See how Repsense employs LLMs for media analysis in the interview below.
• Team Rapid. What's unfolding from the early Kilo team? Series of new ventures. kopa.ai is bringing agents into e-commerce growth, something these guys know incredibly well (they need Product Manager and UX Designer). Also in ecom, Callsy AI is hustling at 500 Global 500 Eurasia accelerator.

• Defence. Origin Robotics and DefSecIntel Solutions launch partnership to drive the Drone Wall forward. Expeditions Fund II already secured EUR 100M to invest. Poland becomes the biggest buyer of Taiwan's drones. UDS signed MOU with OQ Technology, to bring Direct-to-Device 5G NTN into UDS platforms. Ragnar reporting from Defence Tech Valley, always interesting takes.
• Get to know DECODO, why and how they rebranded.
• Space. ReOrbit’s EUR 45M Series A is timely inspiration for the likes of SysIDE from Sensmetry, as well as Astrolight, Blackswan Space, Delta Biosciences and others. Estonian AS Datel has signed a contract with the European Space Agency.

rounds and capital

• Ex-Sequoia partner Matt Miller launches $400m fund Evantic, with nexos.ai being one of the first five investments (alongside Lovable, n8n, others)
• Cyberbit, an Israeli cybersecurity company headquartered in the US, has acquired Estonian startup RangeForce to create AI-based cyber readiness for businesses and governments around the world.
• sun.store, a B2B digital marketplace for solar components, has raised €6 million in seed funding to accelerate its mission. Co-led by Contrarian Ventures, Market One Capital and Movens Capital. FJ Labs, Push Ventures and Aidiom also participated.
• Argyle got a new funding round – adding Mastercard as an investor, alongside continuing support from Bain Capital Ventures, Checkr, Inc., Rockefeller Capital Management, and SignalFire. This is C extension, that team knows how to fundraise.
• Unive AI, AI-driven college and career counseling platform, has secured €350,000 in funding from FIRSTPICK this July, and an additional €60,000 grant from Edu Challenger this September.

three questions

Mykolas Katkus, Co-founder and CEO, Repsense

What is Havel?

Havel is a Foreign Information Manipulation and Interference (FIMI) dashboard that tracks traditional media, social media, and short video content across platforms. It identifies stories, detects disinformation and propaganda, and helps users understand and plan strategic responses to narrative battlegrounds.

Currently deployed in three countries, Havel enables Strategic Communications (Stratcom) and FIMI teams to instantly assess the media landscape and develop effective response strategies. The platform is also used by law enforcement institutions, regulators, and financial institutions to identify fraud and prevent cyberattacks, as well as by political parties and other large stakeholders.

Where do you want to see Repsense in 5-8 years from now?

We have a clear goal: to become the reference product for multinational public and private stakeholders seeking to track, assess, and simulate the impact of public information on people's behavior and decisions.

To achieve this, we are developing models that will help us identify relationships between the supply side—media landscape evaluation—and the demand side—opinion polls and real-life decisions. We're also working on agent-based simulations to better understand these dynamics.

What tech stack and LLMs allow you to research at this scale?

We rely primarily on text embedding technology (the core of LLMs), which allows us to encapsulate the semantic meaning of texts into measurable vectors. This enables us to manipulate them, calculate distances, cluster communities, and even compute angles, projections, and mathematical products of the content we consume daily. This creates a "black box moment" for Repsense users, where suddenly a massive volume of media begins to make sense again.

This utilization of text embeddings requires us to extract maximum information from given content. Therefore, we continuously deploy technologies such as speech-to-text, optical character recognition, voice recognition, and image analysis, enhanced with an additional layer of LLMs (GPT, Claude, and especially the newest Google Gemini versions for qualitative object recognition and description) across our documents. This orchestration of these three latter components makes the Repsense toolkit a pioneer in deeper media understanding, especially where information is lacking (for example, half of TikTok videos have no title or description).

Finally, our data analysis philosophy is based on high-fidelity contributor data. Here are a couple of examples:

Context Weighting: There's a significant difference between your entity being mentioned in the title of a highly negative article versus being randomly mentioned on the last page of a lengthy newsletter. It's crucial for us to measure this difference numerically, so we actively use sentiment analysis, prominence scoring, text quality assessment, named-entity recognition, and other models.

Data Enrichment: All collected datapoints (author, domain, interactions) are initially just strings without information about source origin or features. By connecting secondary external databases (such as Ahrefs), we link Repsense data with other analysis networks to provide richer insights.

sponsors

Cloudvisor [AWS partner dedicated to startups]
Hostinger [making online presence accessible to everyone worldwide, hiring]
Google For Startups [cloud credits up to $350K, expert support & faster growth]
Oxylabs [Step into the world of web data gathering]
VIALET [Business accounts for growth, hiring]
Surfshark [Top 50 among fastest growing in EU, hiring]
15MIN Group [all the news you need to know]
Superangel [Pre-Seed, Seed and Series A for tech companies]

founder's guide

• How to network? (don't)
• Why university degrees are still relevant? 80% of founders either studied together or worked together. (We are concerned that in the Baltics students start working corporates jobs too early, and there is too few straight-out-of-school communities like Sintra mafia or basedspace community).
• Books not to read: the “anti-reading list”
• Spot-on tip for first-time founders
  • If your VC fundraising slide deck has a use of proceeds slide that is a pie chart, you will not get funding. You are welcome.
• With some possible acquirers (esp software), founders need to reach out directly.

roleplay

further insights

• The bigger the success story, the more likely founders are to use this phrase: the job is not finished (Ramp, Klarna)

ecosystem

• Which Baltic Airport is the best (vote)?
• Whatever Baltic countries think they are competing about, only talent matters.
• For an economy to be relevant in the next decades, it must master The Electric Tech Stack - writes Noah Smith. These are the underlying technologies for electric vehicles, drones, computers and phones, robots and more. These will define your industrial capacity, economic development, and sovereignty – listing Lithuanian startups working with atoms.