Alinia AI

Yesterday, we celebrated Alinia's coming out of stealth. We presented the team behind the mission to friends and the local community. The energy was real, and it helped that we had one of the city's best views. Thank you to everyone who came out to help us celebrate and spread the word about how Alinia enables companies to roll out GenAi more safely. #gratitude #networking #startup #community #norrsken_bcn #barcelonatechhub #Innovation #Barcelona #AIEthics #ResponsibleAI #AISafety #AIalignment #GenAI #AIregulation #AI #LLM

At Alinia AI we are proud to announce our membership in the AI Alliance. An unparalleled pace of innovation in the AI space, coupled with a race for AI adoption and upcoming regulations, urges to collectively address safety and responsible deployment of this powerful technology. We will commit efforts to both trust & safety and policy advocacy working groups. Alinia is grateful for the opportunity, and would like to specially thank Sean Hughes, Andrea Greggo, Anthony J. Annunziata, and Melissa Turesky for their kind support and coordination over the process. You can read the announcement here: https://lnkd.in/dWm36XME #AIAlliance #AIAlignment #AIGovernance #ResponsibleAI #FoundationModels #LLM #EthicalAI #AIRegulation #TrustworthyAI #Enterprise #Regulation #GenerativeAI #genAI #AIEvaluation

Today our CEO, Ariadna (Ari) Font Llitjós, PhD was interviewed as the AI and RAI specialist by one of the most viewed morning news in Barcelona (Els Matins) Ariadna: “Almost everyone is interested in entering the generative AI space, but they lack the technical and RAI expertise. Alinia was born to meet this need and help companies evaluate and guide their gen AI applications” You can now view her segment in Catalan here https://lnkd.in/djUMcckk

🎆 Happy 4th of July!🎆 At Alinia AI, we're proud to celebrate this day with all of you. Thank you to our US partners and investors for your trust and collaboration. Let's continue to achieve great things together! Wishing everyone a joyous and safe Independence Day! #IndependenceDay #4thofJuly #AI #Innovation

The 3rd ML & AI Meetup Bern meetup today was a blast! Jan Badertscher from Nexplore AG introduced us how easy it is to run a custom AI model as coding assistent on your dev machine. Just download the Llm and a Visual Studio Plugin - and you have a private coding assistent with good quality. Carlos Muñoz Ferrandis from Alinia AI gave interest insights into the alignment and preparation necessary when going to production with your AI and LLM project. Thanks to Zühlke Group Bern for sponsoring the meetup location and the Apero!

🔊 The Gartner 2024 Hype Cycle for Artificial Intelligence was recently announced. 🔍Whereas both Foundation Models and Generative AI are already going down from their peak of inflated expectations, Responsible AI is about to reach it. These 3 trends are at the core of Alinia’s business: ▶ Our commitment is focused on landing the hype of generative AI by enabling the enterprise to deploy it confidently through customized control and guidance solutions. Companies need to make sure Foundation Models perform as expected when deployed in a specific business use case in their industrial scenarios and be able to get evidence of it. ▶ Responsible AI (RAI) practices and tooling are key for gen AI to be deployed at a massive scale. At Alinia, RAI is in our DNA due to our expertise and backgrounds; it's rooted in our product. Our approach to RAI aligns with both best practices and upcoming regulations. Do you want to know more about how we operationalize RAI at Alinia through our gen AI alignment process? Reach out: contact@alinia.ai Link to the post: https://lnkd.in/dsEiSFrZ

🔊 "A Roadmap to Pluralistic Alignment" - Practical insights extracted. This paper proposes ways to bring pluralism into AI systems through alignment, as well as related evaluation benchmarks to assess "pluralistic alignment". LLMs are used as testbed. 🔍Some of the identified questions and problems that resonate with what we see at Alinia talking to enterprise customers: ▶ Customization necessitates pluralism. Any guardrails placed on AI systems will require customization, within the bounds of those guardrails, to serve diverse use cases and values. ▶ AI/NLP has trended away from specialist systems and towards generalist systems (foundation models) for use in a diverse set of tasks by a diverse set of users. Yet, current alignment optimizes these generalist systems for a single objective – averaged human preferences.  ▶ Current LLMs often give advice confidently but inconsistently or in an opinionated manner, affecting users’ downstream judgments. ▶ Models which have been so heavily “aligned” towards a specific attribute such that they cannot be steered to other attributes fail to be useful (or usable) to populations who may not share that value or attribute. ▶ There is a fundamental tension between helpfulness and harmlessness for LLMs. These two attributes often get clumped together and are implicitly traded-off through data mixtures or vague human preferences. ▶ There has been a growing sense in the community of the importance of measuring which values and to whom we are trying to align LLMs. 🔍Some of the proposed approaches to pluralistic alignment that resonate with our efforts: ▶ Steerable Pluralistic Models. A pluralistic model might faithfully steer (or align) its responses to a given attribute or perspective, such as a value, framework, or population. ▶ An important application of steerable-pluralism is customization. Users often want to personalize models towards characteristic properties and perspectives. ▶ Through explicit multi-objective benchmarks, we can better understand how they trade-off and make informed decisions when selecting a model for a given application or domain.  ▶ Trade-off steerable benchmarks: A trade-off steerable benchmark measures whether a single model can represent solutions across a spectrum of objectives, allowing for tuning to trade-off functions of choice at deployment time.  ▶ Although some alignment techniques may require automatic methods (e.g., jury functions), we advocate for human-centered evaluations whenever possible. Interested in measuring the degree of alignment of your gen AI systems according to specific values and business objectives? Reach out at contact@alinia.ai ! Link to the paper: https://lnkd.in/diSDYpbc #AIAlignment #AIGovernance #ResponsibleAI #FoundationModels #LLM #EthicalAI #AIRegulation #TrustworthyAI #enterprise #regulation #generativeAI #genAI #AIevaluation

Want to move to Spain? Here's a list of well-funded startups from June 2024: - TravelPerk (business travel) raised up to $135M. - Tinybird (real-time APIs) raised $30M. - Bespoke Pixel Studio (gaming) raised €25M. - Cabify (transport) raised €20M. - Bcas (student finance) raised €17M. - Job&Talent (marketplace) raised €18M. - NeuronUP (cognitive rehabilitation) raised €10M. - Samara (solar energy) raised €9M. - Mitiga Solutions (data analytics) raised €8M. - Badi (room rental app) raised €7M. - FOSSA Systems (IoT satellite) raised €6.3M. - Ray studios (tattoo removal) raised €4.62M. - Danelfin (AI stock analytics) raised €2M. - Alinia AI (alignment platform) raised $2.4M. - Sincrolab (digital therapies) raised €1.7M. - Leemons LXP (edtech) raised €1.62M. - Bit2Me (crypto) received investment from Unicaja. - Azitek (IOT) raised €750K. - inBiot Monitoring SL (biotech) raised €500K. Get dealflow in your inbox monthly, subscribe with link below 👇 #startups #emprendimiento #spain #tech 📸: credit to contxto

🔊 "A Roadmap to Pluralistic Alignment" - Practical insights extracted. This paper proposes ways to bring pluralism into AI systems through alignment, as well as related evaluation benchmarks to assess "pluralistic alignment". LLMs are used as testbed. 🔍Some of the identified questions and problems that resonate with what we see at Alinia talking to enterprise customers: ▶ Customization necessitates pluralism. Any guardrails placed on AI systems will require customization, within the bounds of those guardrails, to serve diverse use cases and values. ▶ AI/NLP has trended away from specialist systems and towards generalist systems (foundation models) for use in a diverse set of tasks by a diverse set of users. Yet, current alignment optimizes these generalist systems for a single objective – averaged human preferences.  ▶ Current LLMs often give advice confidently but inconsistently or in an opinionated manner, affecting users’ downstream judgments. ▶ Models which have been so heavily “aligned” towards a specific attribute such that they cannot be steered to other attributes fail to be useful (or usable) to populations who may not share that value or attribute. ▶ There is a fundamental tension between helpfulness and harmlessness for LLMs. These two attributes often get clumped together and are implicitly traded-off through data mixtures or vague human preferences. ▶ There has been a growing sense in the community of the importance of measuring which values and to whom we are trying to align LLMs. 🔍Some of the proposed approaches to pluralistic alignment that resonate with our efforts: ▶ Steerable Pluralistic Models. A pluralistic model might faithfully steer (or align) its responses to a given attribute or perspective, such as a value, framework, or population. ▶ An important application of steerable-pluralism is customization. Users often want to personalize models towards characteristic properties and perspectives. ▶ Through explicit multi-objective benchmarks, we can better understand how they trade-off and make informed decisions when selecting a model for a given application or domain.  ▶ Trade-off steerable benchmarks: A trade-off steerable benchmark measures whether a single model can represent solutions across a spectrum of objectives, allowing for tuning to trade-off functions of choice at deployment time.  ▶ Although some alignment techniques may require automatic methods (e.g., jury functions), we advocate for human-centered evaluations whenever possible. Interested in measuring the degree of alignment of your gen AI systems according to specific values and business objectives? Reach out at contact@alinia.ai ! Link to the paper: https://lnkd.in/diSDYpbc #AIAlignment #AIGovernance #ResponsibleAI #FoundationModels #LLM #EthicalAI #AIRegulation #TrustworthyAI #enterprise #regulation #generativeAI #genAI #AIevaluation

🙌 𝐑𝐞𝐜𝐞𝐧𝐭 𝐅𝐮𝐧𝐝𝐢𝐧𝐠 𝐑𝐨𝐮𝐧𝐝𝐬 𝐟𝐨𝐫 𝐅𝐞𝐦𝐚𝐥𝐞-(𝐂𝐨)-𝐋𝐞𝐝 𝐒𝐭𝐚𝐫𝐭𝐮𝐩𝐬 👩💻 Anita Schjøll Abildgaard, Co-founder of Iris.ai, has raised €7.64M in Series A funding. The Oslo-based startup specializes in AI for scientific text understanding and helps clients unlock cutting-edge scientific insights. ✈️ Tara Spielhagen Hale, Co-founder of Swiipr , has raised £6M in Series A funding. The company offers a digital platform for airline disruption payments. 🔋 Mariela Atanasova, Co-founder of Decade Energy, has raised €3.6M in Seed funding. The Paris-based startup provides a software platform to help businesses electrify their truck fleets. 🎨 Tonia Samsonova, Founder of Exactly.ai, has raised $4.3M in Seed funding. The platform helps artists use AI to scale and monetize their artwork. 🛍️ Tash Grossman, Co-founder of Slip, has raised £2.5M in Seed funding. The London-based company transforms receipts into an interactive media channel for personalized marketing. 🤖 Ariadna (Ari) Font Llitjós, PhD, Co-founder of Alinia AI, has raised €2.2M in Pre-seed funding. The Barcelona-based startup enables enterprises to govern and optimize Gen AI applications. 🌍 Holly Simmons, Co-founder of Niya, has raised £1.5M in Pre-seed funding. The startup offers an AI-powered career management platform for refugees and migrants. 🧠 Anna Muñoz Farré, Co-founder of Punto Health, has raised €645K in Pre-seed funding. The London-based digital health company focuses on managing Alzheimer’s and dementia. Congratulations! Who else needs to be added to this list of recent #funding rounds for female (co)-led #startups? 🍾 🔔 Follow me for more funding news on female-led startups! #hypewomen