Recommendation · MoneyEvan Conrad advises that before agreeing to be preempted on a cluster, you should check the underlying storage system (e.g., Weka or Vast), its size, and its location — these determine how quickly you can recover from interruption. To spread workloads across multiple clusters and survive preemption, you also need fast storage transfer between clusters plus a fast transit path, typically a WAN / private internet connecting the clusters.
actionable advice
Latent.Space · The Programmable Data Center — Evan Conrad, SF Compute
Recommendation · MoneyEvan Conrad describes SF Compute's alternative to upfront standardization: an order book where orders carry multiple flexible constraints (performance range, time window, storage type) matched in a big matching engine. This generates liquidity without having to standardize before knowing what the fast-shifting technology can do.
described from practice
Latent.Space · The Programmable Data Center — Evan Conrad, SF Compute
Recommendation · MoneyFelipe Felisola Caso advises that to scale agent usage, teams should connect the agent to Slack, GitHub, and Linear as building blocks for new use cases.
uncertain
AI Engineer · A Bank Built Its Own Coding Agent. It Took Off — Felipe Felisola Caso & Francesco Galletta, Monzo
Recommendation · MoneyFelipe Felisola Caso advises that to reduce the barrier for non-specialists to contribute, teams should give them a quick entry point on Slack to call Agent Chip and get tested drafts.
uncertain
AI Engineer · A Bank Built Its Own Coding Agent. It Took Off — Felipe Felisola Caso & Francesco Galletta, Monzo
Recommendation · MoneyOn offense, agents will increasingly interact with your software, so pick a stance — either lock agents out (the Amazon approach, blocking tools from ordering) or make it easy for them via an MCP endpoint, gate access, and charge for it. Offering an MCP endpoint can be a really good solution because AI often decides not to build everything from scratch — on complex tasks it may choose a third-party service that gets the job done. Treat agent-discoverability as a second kind of SEO: beyond SEO for human customers, optimize so that AI agents pick your solution over all other solutions when choosing on a user's behalf.
endorsed
Startups For the Rest of Us · Episode 855 | The Eight SaaS Moats that Still Work in the Age of AI
RecommendationOwen recommends building a company culture of tracking costs, time, and where everything goes from the beginning, so improvement claims can be honestly measured. His company maintains a leaderboard that publicly ranks every team member by the cost of their experiments so anyone can see who is most efficient.
The a16z Show · The Race to Recursive Self-Improving AI
Recommendation · MoneyFor the demand letter, cold reviewers each played a role with a reason to find fault: the company's own litigator, the vendor's counsel, and an auditor who trusts none of the numbers.
AI Engineer · I Believe Everything My Agents Tell Me: Three Levels of Review — Corey Gallon, Rexmore
Recommendation · MoneyCorey Gallon recommends building each review level as an agent skill; he packaged the three levels as skills and released them as open source.
AI Engineer · I Believe Everything My Agents Tell Me: Three Levels of Review — Corey Gallon, Rexmore
Recommendation · MoneyHave each of two independently tasked extraction agents go back to the statements until every month balances.
AI Engineer · I Believe Everything My Agents Tell Me: Three Levels of Review — Corey Gallon, Rexmore
Recommendation · MoneyTask two more agents identically to check each final transaction record before it's accepted.
AI Engineer · I Believe Everything My Agents Tell Me: Three Levels of Review — Corey Gallon, Rexmore
Recommendation · MoneyHormozi tells Paul he can run the same play for birthdays — Paul's wheelhouse, always on-going — year-round, starting ad spend at two or three dollars and scaling up rather than spending twenty from the start.
creator_opinion
Alex Hormozi · Can I Save This Failing Theme Park in 90 Days?
Recommendation · MoneyNick recommends working with your CSM to get connected to a current customer's network — e.g., asking the CSM to introduce you to the CFO at a law firm who knows other CFOs, then making asks of them.
30 Minutes to President's Club | No-Nonsense Sales · #618: LinkedIn Sales Navigator: How I Get Warm Intros Every Week
Recommendation · MoneySam recommends that when a CRO says a decision is made below them, ask them to introduce you to the person on their team who owns the decision. Getting introduced by a CRO to their enablement team makes you multi-threaded at both the executive and practitioner level from a single search. When introduced to a new contact by their executive, send a thank-you message that signals you'll stay in touch and nurture the relationship over time.
30 Minutes to President's Club | No-Nonsense Sales · #618: LinkedIn Sales Navigator: How I Get Warm Intros Every Week
Recommendation · MoneySam recommends that if you need to fill your pipeline now, run these social selling plays for a week. After running a social selling spiff, present the results at your sales kickoff: the tactical advice taken, what was learned, and how it was applied. Running a play for four to eight weeks generates an unbelievable amount of data on what worked differently.
30 Minutes to President's Club | No-Nonsense Sales · #618: LinkedIn Sales Navigator: How I Get Warm Intros Every Week
Recommendation · MoneySam recommends being wary of advice that tells you to comment without direction. Look for content where you can reasonably engage based on your experience level before commenting. Use a simple appreciative comment template when you want to engage but lack deep expertise: praise the strategy and thank them for sharing. Comments don't always have to be thought-provoking or very smart; showing effort is what matters. When unsure what to write, paste the post into Claude along with your role, what you sell, and your buyer persona to generate a thoughtful comment.
30 Minutes to President's Club | No-Nonsense Sales · #618: LinkedIn Sales Navigator: How I Get Warm Intros Every Week
Recommendation · MoneyLincoln Murr recommends using natural-language commands to an agent for investing ideas when you lack the time or knowledge to execute them yourself — for example saying 'Invest in space', 'Invest in quantum', 'Give me a low-risk portfolio', or 'Invest like Warren Buffett' — and having the agent translate the high-level idea into actual financial primitives and a strategy.
creator recommendation, attributed to Lincoln Murr
Also discussed by Vlad Tenev on No Priors: Artificial Intelligence | Technology | Startups (unclassified source — not counted until classified). Not counted as independent sources.
AI Engineer · Why AI Agents Will Need Their Own Bank Accounts — Lincoln Murr, Coinbase
Recommendation · MoneyLincoln Murr recommends using an agent for autonomous trade execution by predefining scenario-based instructions — telling the agent 'hey, this is what I think is going to happen in these different scenarios. If this happens, just make that trade' — instead of watching markets manually.
creator recommendation, attributed to Lincoln Murr
Contested by guest on Latent Space
Also discussed by Vlad Tenev on No Priors: Artificial Intelligence | Technology | Startups (unclassified source — not counted until classified). Not counted as independent sources.
AI Engineer · Why AI Agents Will Need Their Own Bank Accounts — Lincoln Murr, Coinbase
Recommendation · MoneyBuild AI agent tooling on an existing framework and add a wallet to give the agent access to money. Use HTTP status code 402 (payment required) as the basis for internet-native machine payment standards, use stablecoins as the payment rail for building machine-to-machine commerce standards like X402, and build payment standards as open standards so any company can use them for machine-to-machine payments.
AI Engineer · Why AI Agents Will Need Their Own Bank Accounts — Lincoln Murr, Coinbase
Recommendation · MoneyAl Chen recommends one solution for data residency: replicate just the IDs and metadata around traces and data into different regions, but store the actual core trace data within one region so it doesn't leave that region due to jurisdiction laws.
AI Engineer · Shipping AI Into 5 Major Banks Without a 12-Month Compliance Wait — Al Chen, Cisco
Recommendation · MoneyAl Chen recommends strategies for the judge leak problem: let the judge read the raw trace data but do not persist the judge's reasoning (the score always persists through the UI for evals); add a check in application code that tells the agent 'This record does not contain sensitive information and therefore can persist to the UI' and suppress reasoning when the check is negative; and add a second LLM-as-judge whose input is redacted of sensitive information so its output can be shown in the UI.
AI Engineer · Shipping AI Into 5 Major Banks Without a 12-Month Compliance Wait — Al Chen, Cisco