Albert Cui

LLMs are trained on internet-scale data. But to make AI work for work, models need to be grounded in your enterprise data -- a much harder problem than meets the eye! How do you connect to all your data sources and ingest them frequently enough while managing compute/storage costs? How do you make sure that you only show users results from documents they have access to? How do you rank results without internet-scale click and link data? How do you process the content of documents that are often hidden in PDFs and presentations? These are tricky problems, and it's why enterprise search didn't work for nearly a decade, until Glean came along. Turns out it's the exact same set of challenges needed to give LLMs the right context to work their magic on your enterprise content. Glean spent 5 years building the ingestion, search, and ranking infra needed to solve this problem well. This has given them the right to run the "Work AI" race and delight users at work. CEO Arvind Jain shares more on the latest episode of Training Data. Listen here, or wherever you get your podcasts: YouTube: https://seq.vc/ytd Apple: https://seq.vc/pit Spotify: https://seq.vc/jnz 

203

10 Comments

Gathering consumer insights used to be painfully slow. But what if you had AI personas built on the preferences of real humans? Check out this next-level tech from Vurvey. What if you could quickly design some concepts, then run them by a team of AI personas? What kind of product creation acceleration could you create by quickly landing on your best concepts with the best chance of appealing to a wide or specific group of consumers? Share your thoughts in the comments.

3

How to monetize AI is a tricky and important decision for any company, and there's no quick fix that works for everyone. The good news is there are frameworks that can help! This fantastic write up from Abde and Sam is a must read for anyone trying to monetize an AI-based product.

5

1 Comment

London folks! It's starting to become a cliché that there are thousands of AI pilot projects and relatively few that make it to real life. Some of that is because the products just don't deliver. But some is just because founders and sales teams haven't learned how to sell AI yet. Come learn what actually works out there! #ai #aistartups #founder #founders #startupssales #gtm #startupgtm #founderledsales #startup #startups #aistartup

12

1 Comment

Impactful Biz, Ops & Investment Roles at Early Startups & VCs: • Chief of Staff to Ex. Head of Product at Google • Founding GTM at the most YC product used by other YC companies in the past four years • Chief of Staff to a Partner at a Top-Tier VC Fund • Head of Platform of a NY State Backed Accelerator • Investment Associate at a Seed Stage Fund Investing in Category-Defining Companies • Partner at a Defense-Focused VC Fund • Chief Revenue Officer of a New Space Company in El Segundo • Venture Capital Analyst at a fund investing in overlooked geographies • Investment Associate at a Spirit-Focused Fund • Strategy & Ops Lead at a Company Focused on Helping Improve Autism • Senior Product Manager at a Company Focused on Building Shopify for Content Creators • Program Manager for a Space Security Company • Open-ended Investment Position at a Top-Tier Accelerator • Biz & Ops Lead for One of the Most Loved Digital Consumer Products 💸 All companies are well-funded and backed by top-tier investors. ⬇ You can find links to the roles above, plus many more, in the comments.

53

2 Comments

118 VCs (one hundred eighteen!) are gathering in San Diego next week - to meet with 30 SD startups. But it's so MUCH bigger than just those 30 co's. It brings VCs back - when they see quality companies. It helps those who are next-in-line and adjacent. It enables us to send deal flow all year 'round. It attracts talent and other companies. It inspires entrepreneurship. Connect is working for SD. Together, we're building the best damn innovation ecosystem on the planet. Year over year. Five.Ten.Thirty.

263

25 Comments

One way to get a VC to invest in your company is to make them fall in love with your product. Here's a story about one startup who did exactly that 👇🏻 Fin Capital first engaged with our sponsor (Harmonic) after Fan Wen, their Head of Data Science, had multiple friends in the VC space recommend it. He immediately noticed Harmonic's coverage of pre-seed and seed-stage companies, and how that coverage included: ⚫ Legal filings ⚫ Business registries ⚫ Social platforms ... and more. After just a few months of using it, the firm loved the platform so much that they decided to invest in Harmonic’s Series A. It's been a minute since their investment, but even today Fan says "I still haven’t found anything that competes in terms of data coverage.” If you'd like to go deeper on how Fan and Fin Capital are using Harmonic to get an edge in sourcing --> https://lnkd.in/eCYiVisH

19

3 Comments

🌟Innovative VC Firm Overcomes Exits Drought with Secondary Sales🌟 ⛵Navigating a challenging landscape where exits are scarce, Santa Barbara Venture Partners (SBVP) has pioneered a novel approach to sustain its growth and attract investors for its second fund: secondary sales. Instead of waiting for traditional exits like IPOs or acquisitions, SBVP opted to sell shares of its portfolio companies, demonstrating its ability to generate returns for investors and stand out in a competitive market. 🎤According to Dan Engel, founder and managing partner of SBVP, these secondary transactions have been a game-changer, sparking investor interest and bolstering the firm's credibility. By leveraging its recent successes, including a lucrative stake in sports-betting company DraftKings Inc.' acquisition of digital lottery app Jackpocket, SBVP seized the opportunity to return profits to its limited partners (LPs) and pave the way for its second fund. 💡Engel highlighted the challenges faced by young VC firms in raising subsequent funds, particularly amid a downturn in exit activity and heightened investor scrutiny. With traditional exit routes becoming increasingly elusive, the pressure is on for firms to demonstrate tangible returns and establish a track record of success. ✨"For us, secondary sales have been a game-changer. They've helped us return profits to our LPs and attract investors for our second fund," said Dan Engel. 💰For SBVP, the decision to pursue secondary sales was driven by the need to provide liquidity to LPs and validate its investment thesis in the eyes of prospective investors. By strategically offloading portions of its holdings in high-performing portfolio companies like Bark Technologies and Rad AI, SBVP not only generated substantial returns but also bolstered investor confidence in its ability to deliver results. ⚠Despite the complexities and potential stigma associated with early share sales, Engel emphasized the importance of prioritizing investor returns and seizing opportunities to unlock value for stakeholders. With a focus on profitability and transparency, SBVP remains committed to its mission of delivering sustainable growth and maximizing returns for its LPs. 🔍 "Returning profits to our investors is our top priority. By strategically selling shares, we're proving our commitment to delivering results and driving value for our stakeholders," added Engel. As SBVP continues to explore secondary transactions and expand its investor base, the firm stands as a testament to innovation and resilience in the face of market challenges. 🚀 ✅ Looking to raise capital for your #fund and increase the international pool of your LP #investors?   🤝 Need warm #LP introductions?   📝 Selling #secondaries to increase liquidity?   🧐 Looking for co-investments?   ▶ G+QUANT's link for inquiries and fund decks: https://lnkd.in/gjC_EuTE #VCInnovation #SecondarySalesSuccess #InvestorReturns #ValueCreation 

2

Batch processing for LLM inference activity: "One sign of that is the rise of batch processing. Today, popular consumer AI products like ChatGPT and Perplexity provide users with near instantaneous responses, otherwise known as real-time inference. However, businesses don't always need immediate responses and are often willing to wait hours, days, or even weeks for responses, as long as they don't have to shell out as much money. Founders of cloud and inference providers have told me there's growing pressure from business customers for this kind of flexibility, otherwise known as batch processing." -- Nathaniel Whittemore on "The AI Daily Brief" (with much deserved shout-out to his inspo from Stephanie Palazzolo at The Information) I fall into this bucket as we have a lot of backfilling to do as we're building out the underlying data in our service, and some of that could just as well be done in the wee hours of the night. The new stuff flowing in I want processed ASAFP, but these backfills—I'd mos def take more cost-efficient delayed batch processing. There's also use cases like needing (near) real-time inference for the first inference pass for timeliness / contemporaneousness value, but then a second pass slow-time inference for wider classification exercises over an ever-growing data corpus. I could see us mixing Sonnet & Haiku models in both cases. Just a heads up to Anthropic Mike Krieger as you contemplate this, along with Amazon Web Services (AWS) Jeff Barr Mark Roy Swami Sivasubramanian on the Bedrock side of things...and I don't mean the "batch" mode that doesn't seem to have any real value other than helping AWS manage its own rate limit challenges. (DeepCast is using Bedrock.)

11

2 Comments

I’m thrilled to be announcing that Notion Capital is leading the $15m Series A investment in Cogna today. In the industry, we like to think that software has been eating the world. But when you look at the data a better description would be more of a nibble. Depending on what source you look at the software industry represents between 5-8% of global GDP. Contrast that with the services industry that represents between 60-70% of GDP and worth around $25 trillion on a global basis. Most services have been out of reach for the software industry. They have either been too complex or too specialized for off the shelf software to access in any economically viable way. But that’s beginnning to change with GenAI leading to the emergence of ‘service as a software.’ Cogna is at the forefront of this opportunity - building precision software at scale and delivering huge productivity gains for traditional industries. We know Ben Peters & Lars Mennen well having backed their previous company Five AI that was acquired by Bosch in 2022. And we’re very excited to be backing them again together with Hoxton Ventures & Chalfen Ventures Read my full blog post on why we invested here. Radu Bozga Ben Peters Lars Mennen Bryan G. Mike Chalfen Hussein Kanji Kirsten Connell https://lnkd.in/eh8mhcJb

201

13 Comments

LinkedIn team (Juan Pablo Bottaro Karthik Ramgopal) shared a helpful deep dive into their deployment of an LLM-powered job search assistant. Highly recommend this read! Few highlights:  💡 Relatively easy to setup RAG-based pipeline. Achieved 80% of the basic experience in 1 month, then 4 more months to reach 95% mark. 💡 Evaluating the quality of application outputs is hard. I hear this from many enterprises developing LLM applications. (check out Okareo by Matthew Wyman Boris Selitser) 💡 Balancing latency with quality is important. While certain techniques like Chain of Thought could improve accuracy and reduce hallucinations, it will increase response times, which can affect user experience. https://lnkd.in/gtGGwf5P #largelanguagemodels #RAG #EBR #LLM #ml #ai #hallucinations #evaluations #

4

Is the Head of Data (or Chief Data Officer or VP data or  … ) becoming the new ICP (Ideal Customer Profile) star???💫 💫💫 In the past few months, we’ve seen a surge in startups tackling #datamanagement —covering observability, stability, cost reduction, efficiency, and AI assistance. #Data isn’t just valuable anymore—it’s expensive. Data management has become a core part of business strategy, affecting budgets, competitive ability, and decision-making. Key Trends in Data Management Startups: 𝐃𝐚𝐭𝐚 𝐎𝐛𝐬𝐞𝐫𝐯𝐚𝐛𝐢𝐥𝐢𝐭𝐲: Real-time tools giving full visibility  and insights into pipelines, reducing downtime and improving performance (open source Spark and others). 𝐂𝐨𝐬𝐭 𝐎𝐩𝐭𝐢𝐦𝐢𝐳𝐚𝐭𝐢𝐨𝐧: Platforms designed to cut the rising costs of data storage and processing (Snowflake\ Databricks etc ). 𝐀𝐮𝐭𝐨𝐦𝐚𝐭𝐢𝐨𝐧 & 𝐀𝐈: AI is automating repetitive tasks, freeing up data managers for strategic decisions. 𝐄𝐧𝐡𝐚𝐧𝐜𝐞𝐝 𝐃𝐚𝐭𝐚 𝐆𝐨𝐯𝐞𝐫𝐧𝐚𝐧𝐜𝐞: Tools ensuring data quality, compliance, and security in increasingly complex ecosystems. 𝐃𝐚𝐭𝐚 𝐈𝐧𝐭𝐞𝐠𝐫𝐚𝐭𝐢𝐨𝐧 & 𝐂𝐨𝐥𝐥𝐚𝐛𝐨𝐫𝐚𝐭𝐢𝐨𝐧: Solutions that streamline collaboration by integrating various tools into cohesive platforms. 𝐒𝐞𝐥𝐟-𝐒𝐞𝐫𝐯𝐢𝐜𝐞 𝐀𝐧𝐚𝐥𝐲𝐭𝐢𝐜𝐬: Platforms empowering non-technical users to analyze and act on data independently. Established companies are innovating with a clear focus on data engineers. Datadog enhances real-time tracking and anomaly detection. Splunk offers deeper insights into system health and performance, making proactive management easier for data engineers. New Relic supports data engineers in maintaining performance across complex infrastructures, while IBM leverages AI to automate data governance, freeing data engineers from compliance tasks. Elastic focuses on real-time observability and security, enabling data engineers to monitor, manage, and protect data environments in real time. It is argued, however, that these established companies do not fully address the needs of data engineers, overlooking the unique challenges they face in managing big data infrastructure . The future is clear, data is the driving force behind modern business strategy, and startups. We’re excited to see more Israeli innovators emerge as key players in shaping the future of data management and 𝐰𝐞 𝐚𝐫𝐞 𝐥𝐨𝐨𝐤𝐢𝐧𝐠 𝐟𝐨𝐫𝐰𝐚𝐫𝐝 𝐭𝐨 𝐦𝐞𝐞𝐭𝐢𝐧𝐠 𝐲𝐨𝐮 𝐚𝐭 𝐭𝐡𝐞 𝐧𝐞𝐱𝐭 𝐃𝐀𝐓𝐀 𝐜𝐨𝐦𝐩𝐚𝐧𝐲 :)

98

7 Comments

Congratulations to the Gentrace team on their Series A! I have now worked with Doug Safreno, Vivek Nair and Daniel C. Liem for a while through our pre-seed investment at K9 Ventures and they're an amazing team. Also excited to partner with Jett Fein at Headline and Kojo Osei at Matrix! With each new technology shift comes the need for new tools to build with that technology. We saw this with the Internet, with mobile, and now we're seeing it with AI. If you're using #AI and #LLMs at any level of scale you need tooling to be able to evaluate the performance and cost across different language models. Gentrace has built incredible tools to help large companies manage, evaluate, and improve their AI usage. If you're using AI models without this, you're operating in the dark 🕳️. Let gentrace turn the lights on for you!💡🚀

18

3 Comments

Tribe AI just raised our first external round of capital from Bryce Roberts at Indie, with significant participation from our customers, partners and network of top AI engineers. We’re excited to share this milestone in TechCrunch (thanks Rebecca Szkutak for sharing our story!) along with some long overdue updates:  💲 Raised $3.25M to keep up with demand and build the platform to enable service delivery at a tremendous scale 🚀 8 figure revenue run rate, more than doubling year over year (and accelerating) 🏛 Enterprise customers from the Fortune 20  🧠 500 top AI engineers from places like Google, Tesla, Meta and OpenAI As a refresher, Tribe is an AI-native services company that helps Enterprises build best in class AI products. Our team is augmented by a talent network of over 500 top AI engineers from places like Google, Tesla, Meta and OpenAI who bring both industry and deep technical specialization to our project deployments. These are the people who have built the infrastructure that is powering the AI wave and know it better than anyone. After bootstrapping Tribe AI since 2019, we decided to raise this capital because the market opportunity was simply too big. Even as we were reinvesting our profits back into the business, we still saw more demand than our capacity could support. We’re grateful to have met Bryce and have him believe in us and the opportunity we saw. We’re also extremely proud to have the major stakeholders in our business (our customers, partners and network of top AI engineers) invest in the round as well. We’re on a mission to help Enterprises realize the value of AI for their business, and we’re building a category defining AI company in the process. Read all about it 👇 https://lnkd.in/dYRDQDjp

348

87 Comments

Do you know how to turn your AI Pilot into actual Profit??? Last week’s session at the Generative AI World summit was all about unlocking GenAI’s potential to drive real-world ROI! Thanks to all who joined and brought your insights, questions, and energy… It is clear that innovation in AI isn’t just about the tech - it’s about how we creatively and responsibly apply it. Keep pushing boundaries and finding new ways to make these advancements -> work for us, not just impress us… For all those who messaged me for the presentation - it will be published by the organizers on YouTube in the next few months. Stay tuned… #genai #artificialintelligence #profit #entrepreneurship #roi

53

2 Comments

Goldman Sach's recent takedown of Generative AI is going to read one day like those who said the Internet is just a fad, especially if we expand to LLMs. But there are two things that can make AI more useful to more people and faster: 1) Chat interfaces are fundamentally limited because they don’t direct. Like when a friend is hangry, but can’t make a decision where to eat. We often don’t know what we are looking for, especially from general purpose tools. Tools tend to be more useful when they are opinionated and have clearer use case guardrails. Hybrid chat and dynamic personalized user interfaces as a starting point (not just an output) are promising future paths to explore. What if we could generate a directive UI using the context of how someone learns, past behavior, if they have some visual impairment such as color blindness and more? There is a UI/UX opportunity to redefine how humans and technology interact that hasn’t existed since graphical interfaces were first invented. 2) When LLMs work, it’s like magic. When they fail the fail mode can be nonsensical or downright wrong. The last 1-5% of the solution is where many products will fail. If people can’t rely on a product to provide consistent value they are less likely to form habits around it. LLMs can be paired with vector and graph databases in creative ways to reduce hallucinations and have better fail modes where the next step creates a loop with the user to sharpen their intent and offer something useful. Have your own vision for the future of how we interact with AI and technology? Drop a note in the comments! https://lnkd.in/giQjUmAk

7

2 Comments