Wei Lien Dang

During yesterday's LibreQoS APNIC webinar, I posed the question: won't more bandwidth solve these problems? As Herbert Wolverson said, yes, more bandwidth is good, but, still these problems remain if queueing is done incorrectly. Here's my take on why bandwidth alone is not as good as bandwidth + good queueing policies: Generally it was believed that 'data is important, so don't throw it away; hang on to it and send it later'. In practice, this has proven to be suboptimal as two issues may emerge: 1) latency increases for some flows due to heavy demand from other flows using the same bottleneck link, 2) even a single flow can have excessive latency due to aspects of typical TCP behavior (referred to bufferbloat) when the buffers grew large enough that the data being buffered was retransmitted anyway. It turns out that not all data is equally important. Active Queue Management is the art of deciding priorities, both in deciding what data to throw away, but also in allowing some later arriving data to be transmitted ahead of data in another connection that arrived before it. These problems have been studied, and good solutions have been found by using certain queueing policies in routers and switches, referred to as “fq_codel’ and “cake”. These track the different flows not by classifying the data, but by watching the behavior. Flows that send relatively little data (DNS lookups), or at a measured pace (video chat) have priority over flows that send a lot of data as quickly as possible (App and System updates, Video and ISO downloads).

10

5 Comments

Let’s geek out on name matching algorithms in the midst of a world full consent orders and burned out compliance teams.  An example to illustrate: There are lots of people named Jose Lopez or John Smith in the world, so when you are trying to automate names to watchlists, common names can yield a lot of complexity. Socure uses many novel approaches for name matching algorithms.  One example is the use of Siamese Neural Networks with Long Short-Term Memory (LSTM) layers. These models are designed to learn and compare sequential data, such as text or time series, and are particularly useful for tasks like name matching, sentiment analysis, or speech recognition, where the order of the input sequence matters. We use Siamese LSTM models because they are: 1️⃣ Well-suited for name matching from their ability to capture and learn the sequential nature of names. 2️⃣ Able to effectively capture long-term dependencies and contextual information within names, such as the order of the characters or the presence of specific patterns. 3️⃣ Enabled to learn robust representations for names and accurately compare them, even when dealing with variations, misspellings, or different name orders. This is one of thousands of examples of what innovation looks like in an industry that has not seen innovation in far too long.  Joshua Linn, Debra Geister, and Ganesh Kumar- we appreciate your leadership and dedication for always innovating.

72

2 Comments

Founderville believes one of the most significant opportunities is emerging in legacy system applications. We look for manual tasks to leverage AI for efficient automation. Rather than chasing massive TAM, which will be addressed by larger VCs, we focus on smaller market domains that won't be chased as hard. Think of "boring" processes that require human intervention or manualized tasks. Those are opportunities. Keep your eyes open if you have deep domain expertise in a non-automated and fragmented industry. #entrepreneurs #venturecapital #startups #saas

17

7 Comments

Today, GTMfund’s is excited to announce to launch of Operator, its first incubated company. Max Altschuler (founder of GTMfund) teamed up with co-founder Adam M. (design engineer and 0-1 builder) and co-founder Mark Kosoglow (Outreach employee #1 and former SVP of Global Sales) to create Operator.ai alongside an insanely talented team Pleasant Middelhof, Jeremy Jonas, Carl Gunderson. We've entered a period called “The Great Ignore”. Spam and irrelevant, poorly-automated outbound campaigns are running rampant.   Operator gives you an easy-to-use Growth Engineer that allows you to build and enrich lists with accurate data, and run experiments and queries using AI to generate highly targeted outreach.    Check out their website to get updates, early access, and see how they are taking the world back to 2 touches meant to attract buyers into conversations that convert, instead of 20 touches meant to make sure you hit your activity number: https://www.operator.ai/

31

1 Comment

Insightful breakdown from my great Partner Omri Amirav-Drory on the simple factors that really separate the top 1% of investors from the rest, and how more operators should become investors! https://lnkd.in/gEiZdT2h

16

3 Comments

AI memory is fundamental for conversational AI and AI agents. It is critical for building personalized experiences and improving reasoning. Check out Letta’s course to understand how to use AI memory to enhance the customization and quality of your AI app.

15

ISP Strategic Leadership, Engineering, and Operations As an organization scales, there is often a need to separate strategic technology leadership under the CTO from internal execution under engineering and operations management. This allows the CTO to focus on outward-facing duties while enabling streamlined operational efficiency. However, it's crucial to recognize the differing motivations and needs of these distinct technical roles. On one hand, engineers thrive on tackling interesting, challenging projects that push their abilities and knowledge. They value working alongside capable peers, genuine recognition of technical accomplishments, and contributing to products/services that reach the market. In contrast, operations teams should strive for repeatable processes, continual improvement, and driving efficiency at scale. Their priorities revolve around standardizing workflows, minimizing disruptions, and incrementally optimizing existing systems and procedures to maintain consistent service delivery as the company grows. An effective organizational structure acknowledges this dichotomy, fostering an environment that fuels engineering innovation while providing operations with the stability and process-orientation they require. Striking this balance enables the company to stay agile and innovative through its engineering efforts, while simultaneously scaling operational execution in a reliable and repeatable manner. Competitive compensation remains an important motivator across both domains. By properly aligning structural elements, motivational factors, and leadership priorities, a rapidly growing ISP can position itself to attract and retain top talent across engineering and operations, driving sustainable growth and success in the dynamic technology landscape.

45

10 Comments

In overconnected environments dominant suppliers rapidly emerge and many become monopolies — think Google, Facebook, Microsoft, and Apple. When that happens, it is hard to wrestle free of a situation. http://ow.ly/YR5f30prqAF #CloudComputing #data #internet

13

1 Comment

Key takeaways from VC Goat - Vinod Khosla podcast. - The firm cares more about working on interesting problems with large technical solutions rather than just maximizing IRR. The team is there because they believe in the mission. - Khosla assumes they've lost the money the day they invest, and then maximizes for the upside opportunity. He calls it "option value investing" rather than IRR investing. - Khosla has contrarian bets in AI (neuro-symbolic computing, probabilistic approaches), biotech, robotics, crypto, and more. - He believes aviation fuels, fusion energy, new transit systems, and other contrarian areas ignored by most investors will be huge opportunities. - Most large innovations come from outsiders, not industry insiders. Khosla looks for founders who can learn a business rapidly rather than have deep domain expertise. The Future Impact of AI - AI will be deflationary and increase productivity growth to 4% annually vs the typical 2% forecast. This will cause great abundance but also increasing income inequality. - Bipedal robots will take over most manufacturing and manual labor jobs within 20-25 years. This will free humans to be more creative and pursue their passions. - Education will shift from job training to creativity. Uniquely human elements like taste and curation will be most valued in an AI-enabled world. https://lnkd.in/gqBsmbks

66

5 Comments

At Forum we look at thousands of deals each year across our Accelerator and Pre-Seed funds but only end up investing in a little over 1% of those startups.  As an early stage VC, I am in the business of saying no to nearly every startup I meet, and while there is nuanced rationale for many of the startups I pass on, a vast majority of businesses that don’t meet the qualification to get funded typically fall into one of two buckets. ▪ First time founders have a tendency to go out to market too early to raise pre-seed capital ▪   I’m sure most founders have heard some iteration of this from VCs that ultimately pass on their business.  In some cases it's a blanket excuse covering up for other pass reasons- perhaps a VC doesn’t believe in a team or isn’t bullish on the market opportunity - but in many cases founders simply haven't proven enough to the market to warrant an investment. I meet many teams with an articulate vision for the product that they want to build, but they either haven't performed enough customer discovery to prove demand for their initial wedge product, or need to raise pre-seed capital to build out an MVP.  In those cases, pre-seed investors are unwilling to underwrite those risks.  To be default fundable, teams need to be able to keep costs low, perform extensive customer discovery, and ship some version of an early product - pilots and early paid customers are key. A huge piece of this is having founding engineering talent on the team. ▪Not every startup is a venture opportunity ▪ By definition, a venture backed business is solving a problem within a market opportunity so massive and so desperately needed, that its growth rate is off the charts.   It spreads like a weed, absorbing almost unlimited amounts of capital to fulfill demand within its customer base.   Each year there are maybe 20 companies launched that ultimately have venture scale outcomes. The bar to reach this rarified air is unfathomably high and the reality is the vast majority of founders simply aren’t building for opportunities that have the potential to be one of these generational companies. Many startup founders view VC as the only viable option for capitalizing their business, but not every great business is a venture business, and most successful founders never take VC dollars.  In order to reach an IPO/acquisition large enough to generate venture returns, a startup needs a credible path to $100s of millions in revenue. This is the standard by which early stage opportunities are evaluated. The flip side to this, and I share this feedback often with founders I meet that are building interesting business in non venture markets, if you are able to scale to $10- $20M in revenue that can be a massively successful outcome for a founder.  There are investors who focus on these type outcomes, namely lower middle market PE shops, and the typical path to those deals is a bootstrapped business or a friends/family capital raise.

96

13 Comments

Excited to announce our investment in Alta Resource Technologies alongside Leonardo Banchik (Voyager Ventures) with participation from Orion Resource Partners, Overture Ventures, and WovenEarth Ventures. Rare earths are critical in products from magnets to ceramics and it's important for the US to develop its own supply chain. While the US has some of the best rare earth resources, separating them has historically proven expensive and environmentally unfriendly. Alta through it's computationally designed and optimized proteins allows the US to bring back rare earth production with a cleaner, lower capex, and lower opex solution - and that's just the beginning. Congrats to Nathan Ratledge and the entire Alta team. Read more here: https://lnkd.in/gbGnAJn3

60

4 Comments

ex Google CEO Eric Schmidt spoke at Stanford this week. Lots of nuggets here, but most interesting for the near term, Schmidt believes in the next year AI is set to make a big leap forward. He predicts AI will combine three important features: 1. Vast knowledge retention (1 million token models) 2. Text-to-action capability 3. Personal AI agent fleets While the full impact is unknown, everyone may soon command their own AI teams. Here's an example: Imagine you're developing new quantum computing algorithms. Your AI assistants could: - Continuously scan and synthesize global quantum research papers - Generate novel algorithm ideas based on this knowledge - Simulate and test these algorithms across various quantum architectures - Automatically document findings and prepare draft papers - Identify potential collaborators and draft outreach emails All initiated by a command like "Advance our quantum algorithm research." This could dramatically accelerate R&D cycles and scientific breakthroughs.. https://lnkd.in/eJbg6cnD #frontiertech #AI #futurefrontier

22

8 Comments

I couldn't agree less with this article. In my view, AI will change the world as much as, if not more than the internet did...which is a lot. '60s computer scientist Roy Amara said it best: “we overestimate the impact of technology in the short-term and underestimate the effect in the long run.” Both Altman and the author are falling victim to this blind spot, but to me, it seems far more dangerous to underestimate the impact and get blindsided by the change than it is to be slightly overzealous on estimating the rate of change. #AI #ArtificialIntelligence #Technology #Innovation #cdntech [Read more](https://lnkd.in/gJWwyFZR)

9

Kettleborough VC has been active for ~6 years, invested in ~30 companies, gone through ~80 funding rounds, witnessed ~12 exits including ~4 instances of partial principal erosion, seen ~4 seed to $100M journeys, tracking ~29% XIRR across the portfolio. A couple of IPOs from the portfolio surely seem very much in line ahead.

257

17 Comments

I had an absolute blast with Michel Gagnon and the Stun and Awe crew - THANKS again for having me... Sneak peek of the substance below: > Micro SaaS companies can be just as safe and stable as larger businesses, offering lower valuations and less competition from institutional players. > Founder-product fit is crucial, as founders solving their own problems often create useful products with great retention. > Founder burnout can be a buying signal, indicating a solid business with a founder ready to move on. > The first 90 days post-acquisition should focus on visibility, documentation, transition, and validating diligence assumptions before pursuing growth. > Building transparently and maintaining open communication with users post-acquisition helps preserve the human element and avoid the pitfall of going silent. > Simplicity equals mastery - if it takes a complex explanation to justify an acquisition, you probably don't fully understand the business. > Establishing reasonable incremental growth goals based on baseline performance (systems over goals) is more effective than setting arbitrary targets. > Fostering a culture of care and respect enables honest feedback and allows teams to move faster through a shared commitment to truth-seeking.

11

2 Comments

Instacart is a market leader when it comes to leveraging AI tools to boost productivity at work. I had an amazing conversation with JJ Zhuang, Instacart’s Chief Architect, who has shepherded over 50 AI-driven initiatives to market across the org. We explored what catapulted Instacart to the forefront of AI integration, supporting radically diverse use cases across various functional areas. One striking insight (echoed by Ethan Mollick and others elsewhere) is that you often don't need specialized tools for specialized work. Frontier models like GPT-4 can seamlessly adapt to a wide range of functions and use cases. They're not just for engineers; they're incredibly versatile tools used by marketing, comms, and legal teams. From generating code to proofreading and ideation, GPT-4 empowers every department to leverage AI for enhanced productivity and innovation. By integrating LLM’s like GPT-4 into their workflows, teams can streamline processes, automate repetitive tasks, and unlock bandwidth for new creative possibilities. How can you harness the power of AI in your own work to boost productivity? Share your thoughts below 👇 Want to hear more insights from JJ? Check out the full podcast episode by clicking the link in the comments.

8

2 Comments

I've spoken to dozens of CISO's and CIO's at large enterprises over the last three months about their early internal deployments of LLM's. Three themes keep coming up: 𝟭. 𝗜𝗻𝘁𝗲𝗿𝗻𝗮𝗹 𝗲𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲 𝘀𝗲𝗮𝗿𝗰𝗵 𝗶𝘀 𝗮 𝗸𝗶𝗹𝗹𝗲𝗿 𝗮𝗽𝗽𝗹𝗶𝗰𝗮𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗟𝗟𝗠'𝘀, 𝗯𝘂𝘁 𝘁𝗵𝗲𝗿𝗲 𝗮𝗿𝗲 𝗵𝘂𝗴𝗲 𝗱𝗮𝘁𝗮 𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝗮𝗻𝗱 𝗮𝗰𝗰𝗲𝘀𝘀 𝗰𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲𝘀 𝘁𝗵𝗮𝘁 𝗻𝗲𝗲𝗱 𝘁𝗼 𝗯𝗲 𝗮𝗱𝗱𝗿𝗲𝘀𝘀𝗲𝗱 𝗯𝗲𝗳𝗼𝗿𝗲 𝗲𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲𝘀 𝗮𝗿𝗲 𝗰𝗼𝗺𝗳𝗼𝗿𝘁𝗮𝗯𝗹𝗲 𝗴𝗼𝗶𝗻𝗴 𝗮𝗹𝗹 𝗶𝗻. Giving everyone access to search every database / app sounds great until employees start finding salary data or forward looking forecasts through an LLM. Early LLM deployments are shining a light on an already existing application overprovisioning / data sprawl challenge that enterprises need to address. 𝟮. 𝗧𝗵𝗲 𝗮𝘁𝘁𝗮𝗰𝗸 𝘀𝘂𝗿𝗳𝗮𝗰𝗲 𝗳𝗼𝗿 𝗟𝗟𝗠'𝘀 𝗶𝘀 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 𝗶𝗺𝗽𝗼𝘀𝘀𝗶𝗯𝗹𝘆 𝗹𝗮𝗿𝗴𝗲 𝗮𝗻𝗱 𝗷𝘂𝘀𝘁 𝗴𝗲𝘁𝘁𝗶𝗻𝗴 𝗯𝗶𝗴𝗴𝗲𝗿 𝘄𝗶𝘁𝗵 𝗹𝗮𝗿𝗴𝗲𝗿 𝗰𝗼𝗻𝘁𝗲𝘅𝘁 𝘄𝗶𝗻𝗱𝗼𝘄𝘀 𝗮𝗻𝗱 𝗺𝘂𝗹𝘁𝗶𝗺𝗼𝗱𝗮𝗹 𝗱𝗮𝘁𝗮 𝗶𝗻𝗽𝘂𝘁. There is a need for more automated and scalable approaches to red-teaming / stress testing to ensure model robustness. 𝟯. 𝗘𝘃𝗲𝗿𝘆𝗼𝗻𝗲 𝗶𝘀 𝗮𝗹𝗿𝗲𝗮𝗱𝘆 𝘁𝗿𝗶𝗮𝗹𝗶𝗻𝗴 𝗮𝗻𝗱 𝗯𝘂𝘆𝗶𝗻𝗴 𝗱𝗼𝘇𝗲𝗻𝘀 𝗼𝗳 𝘃𝗲𝗻𝗱𝗼𝗿𝘀 𝗮𝗰𝗿𝗼𝘀𝘀 𝗰𝗹𝗼𝘀𝗲𝗹𝘆 𝗮𝗱𝗷𝗮𝗰𝗲𝗻𝘁 𝗰𝗮𝘁𝗲𝗴𝗼𝗿𝗶𝗲𝘀 𝗮𝗻𝗱 𝗳𝗶𝗻𝗱𝗶𝗻𝗴 𝗶𝗻𝘁𝗲𝗴𝗿𝗮𝘁𝗶𝗼𝗻 𝗮 𝗰𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲. Teams are buying products across the toolchain (explainability, bias & fairness, robustness, privacy, guardrails in production) and finding these tools generally do not integrate well together, despite obvious areas where output from one product might be input to another. There is a desire from buyers for a more unified end-to-end suite that continuously monitors, stress tests, and updates guardrails on LLM deployments. But in the near term, everyone expects to bolt a bunch of tools together. Welcome thoughts from others going deep in the space cc Michael Roytman Gerhard Eschelbeck Mario Duarte Joe Sullivan Casey Ellis Volodymyr Kuznetsov Joseph Thacker Emily Glassberg Sands Sid Trivedi Tobias Citron Brian Schechter #ai #cybersecurity #dataprotection #startups

159

18 Comments

Big things are happening at NeuBird.ai From early access trials to general availability - Hawkeye is helping enterprises tackle their IT challenges head-on. Marina Temkin, CFA at TechCrunch has the scoop on their $22.5m funding round led by: M12, Microsoft's Venture Fund—with participation from Mayfield Fund, StepStone Group, and Prosperity7 Ventures Read on here: https://lnkd.in/eCUZpNp5 cc: Kayla Abbassi, John Vollmer, Raven Raines, Statyn Coppage, Rick Liebling, Anne Sophie Hurst

8

I like this idea of "technical taste". It gets into where software engineering is as much creative art as it is science. Four great takeaways (IMHO) for software devs in this era in particular: 1. Aspiring engineers should cultivate a sense of curiosity, experiment with different tools and technologies, and embrace a mindset of continuous learning. 2. AI has the potential to streamline processes and enhance productivity for engineers, but it may also lead to disruptions in traditional software development workflows. 3. Developing technical taste and judgment is essential for making informed decisions about which technologies and approaches to pursue. 4. Collaboration and open-mindedness are key to leveraging the full potential of AI and staying ahead of technological advancements. I also really liked these two quotes from Sam Schillace: 1. "The right time to do something is when you have that feeling in the pit of your stomach that's like, 'oh, this is a great idea and it's going to suck to build because nothing's ready yet.'" 2. "Technical taste is like, 'how well have you consolidated that set of experiences and heuristics into judgment that you can apply accurately when you see new things?'" 3. "It may be the case that very small teams can do very large projects, or like we were talking about before, it may be the case we're just going to get really ambitious about what we try to do with the same size of teams, which is kind of where I would put my money." Great job on continuing to put out great episodes, Brett Berson and team First Round Capital! #ai #engineering #softwaredevelopment

5