About
I am a co-founder and partner at Mosaic Ventures, a venture capital firm that backs…
Articles by Toby
Activity
18K followers
-
Toby Coppel posted thisSpecific knowledge is found by pursuing your genuine curiosity, not whatever is hot
-
Toby Coppel posted thisConsensus ... is the process of avoiding the very issues that have to be solved
-
Toby Coppel shared thisSpent a morning at The London School of Economics and Political Science (LSE) with the students of LSESU Entrepreneurs for their Zenith Forum. Thank you to Yucheng Tao Jimmy Zhang and the whole team for having me, and for putting together such a thoughtful programme. I spoke about applied AI, where durable value is likely to be built, and what I've taken from thirty years of watching technology cycles, including the one Yahoo-era call I'm still proud of (Alibaba) and the two I'm not (letting Google win and not quite buying Facebook). What struck me most was the quality of the questions. Sharper, more specific, and more honest about the uncertainty of this moment than I hear in many boardrooms. If this is the generation deciding what to build next, Europe is in good hands. Grateful for the invitation, and for the warmth of the welcome.
-
Toby Coppel posted thisThe ultimate reason for setting goals is to entice you to become the person it takes to achieve them
-
Toby Coppel posted thisWrite your obituary and try to figure out how to live up to it
-
Toby Coppel posted thisLuck is what happens when preparation meets opportunity
Toby Coppel replied to a comment
4d
Roberto Bonanzinga no I think suffering is an important part of living. We need to learn how to deal with it and learn from it
Toby Coppel commented on a post
1w
Thanks also to Lorenza Manrique, Zoe Zhou, Jacqueline Dou, Yao Wang, Aries W., William Wang, Jenn Zeng, and the entire LSESU Entrepreneurs, LSE Women in Entrepreneurship Society, Stanford ASES London
-
Toby Coppel liked thisToby Coppel liked thisData retrieval is a critical frontier for AI. Coral is a new way for your AI agents and apps to retrieve company data much more accurately, securely and efficiently than MCP tools, CLI tools or code mode. For agent read tasks, SQL has a structural advantage when query complexity exceeds what a single API call can answer. It avoids pagination through large results, has cleaner, tabular responses, can bring back specific columns, and can correlate across sources more efficiently. We benchmarked Coral with direct MCPs (Datadog, Sentry, Linear, Slack and GitHub) for a diverse set of 82 real-world AI tasks using Claude Opus 4.6. 𝐖𝐢𝐝𝐞𝐬𝐩𝐫𝐞𝐚𝐝 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞 𝐮𝐩𝐥𝐢𝐟𝐭. Across all tasks, Claude was 20% more accurate and 2x more cost efficient using Coral than using direct provider MCPs. With Coral, Claude also had 42% lower latency. 𝐇𝐢𝐠𝐡𝐞𝐬𝐭 𝐢𝐦𝐩𝐚𝐜𝐭 𝐨𝐧 𝐦𝐮𝐥𝐭𝐢-𝐡𝐨𝐩, 𝐡𝐢𝐠𝐡𝐞𝐫 𝐩𝐨𝐬𝐭-𝐩𝐫𝐨𝐜𝐞𝐬𝐬𝐢𝐧𝐠 𝐭𝐚𝐬𝐤𝐬. Across the more complex tasks that typify coding agent, devops and cybersecurity workloads, Claude was 31% more accurate and 3.4x more cost efficient with Coral. A more detailed report with methodology and the full, task-level results are available on our website. Coral’s free open source product is available now. Docs and Github details are linked in the comments below.
-
Toby Coppel liked thisToby Coppel liked thisEveryone talks about the start and the finish. The idea. The launch. The fundraise. The IPO. But most of building a company lives in the middle. The long stretches where the signal is messy. The hypothesis — about a product, a channel, a leader — isn’t proven yet. Progress is uneven. No headlines. No dopamine hits. Just the work. And plenty of uncertainty. That’s the moment to reframe things. To fall in love with the work itself. Talking to customers. Fixing what’s broken. Keeping the team moving forward. It’s not glamorous. The outside world may not notice. But your team will. And it’s where almost all the value gets created. The middle isn’t the part you survive. It’s the part that matters.
-
Toby Coppel liked thisToby Coppel liked thisI’ve recently joined Meta to lead growth across mobile gaming partnerships in Europe. It’s an exciting moment to step into this space, working alongside some of the most advanced and ambitious companies in the industry, all pushing the boundaries of how apps are built, distributed, monetised and scaled. Whether building from the ground up or scaling existing relationships, I’ve been drawn to opportunities where collaboration and execution unlock growth, and that’s where I’ll be spending my time. I’ve been lucky throughout my career to collaborate with incredibly sharp teammates, and that continues here with a group that brings depth and pace to everything they do. Meta’s trajectory has been hard to ignore, particularly in how AI is being embedded across products and internally. Beyond adopting increasingly powerful tools, it’s pushing us to improve how we operate and how we help our partners grow. Looking forward to what we build next!
-
Toby Coppel liked thisToby Coppel liked thisProud of the work we are doing with US Army, Dan Driscoll, and Dave Fitzgerald. A model armory at the tactical edge. Built for denied environments. Built for the American warfighter. https://lnkd.in/eRJBWQXVArmy’s Project ARIA seeks to accelerate AI adoption across the forceArmy’s Project ARIA seeks to accelerate AI adoption across the force
-
Toby Coppel liked thisIt is mind blowing how many different industries need the world‘s best real world model for artificial intelligence. Robotics. Warehouse automation. Mobility. Defense. The interest in BADAS is stunning.🤯Toby Coppel liked thisThirteen days ago we launched BADAS 2.0 as the best collision anticipation model in mobility. Turns out, that was greatly underselling it... The inbound since launch hasn't all been from mobility. Warehouse automation teams. Drone programs. Humanoid robotics engineers. Defense tech operators. All asking the same question: "does this work in our environment?" The answer is yes. And it shouldn't surprise anyone who understands what BADAS actually is. We trained it on 2 million real-world driving scenarios - not because roads are the only environment that matters, but because roads are the hardest environment on earth to model. Unpredictable motion. High-speed decisions. Physics that punishes mistakes immediately. A model that survives that training learns something universal: how the physical world works. 22 million parameters. 99.4% average precision. Runs on the compute already inside most robotic and autonomous systems. No environment-specific retraining required. Run BADAS 2.0 for yourself on your own videos here: https://bit.ly/42B08f0 #PhysicalAI #Robotics #BADAS #Nexar #EdgeAI
Experience & Education
-
Mosaic Ventures
View Toby’s full experience
See their title, tenure and more.
or
Recommendations received
1 person has recommended Toby
Join now to viewOther similar profiles
Explore more posts
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More