Creator perspective · MoneyGive realtors a VIP/standard tier structure as the offer to present to their clients. The VIP tier costs about $250 more but feels significantly more premium than discounting $250 or giving $250 back to the realtor. If all movers are already master movers, there is no real cost difference between standard and VIP. Keep pushing the VIP price level up until people say no, since the standard tier is always a backup, and the people who have money will spend more. The word 'upgrade' is preferred over 'white glove,' and the VIP upgrade offer is described as the biggest lever.
Alex Hormozi · Give Me 42 Minutes, I'll Show You How To Sell Like The Top 1% | Scale or Fail Episode 1
Creator perspective · MoneyOffer a free VIP upgrade (or 'free white glove') to anyone a realtor refers, so the realtor's customers think good things about them — the realtor needs to gain from referring you. Arm the referrer with a brag script ('my guy, I've got an amazing moving company, they give all my customers free upgrades') that makes the realtor look great. People go to realtors because realtors are supposed to know the area better and have done it more times — expertise is why the referral relationship exists.
Alex Hormozi · Give Me 42 Minutes, I'll Show You How To Sell Like The Top 1% | Scale or Fail Episode 1
Creator perspective · MoneyTo create search-agent training data, use a deterministic algorithm to create seed data — run a seed collection algorithm repeatedly over Wikidata to generate many different hops, then insert those hops into the prompt as seed entities and ask the model to create a question using any or all of them. Design search riddle questions the model cannot readily answer from the internet. Execute the searches via a search tool — Tavily (Exa is an alternative provider, or you can write your own search tool). Tavily returns URLs with titles and content for each result.
AI Engineer · Teaching Agents to Search with NVIDIA Data Designer — Dhruv Nathawani, NVIDIA
Report · MoneyThe presenters demonstrate asking the coding agent to use the Langfuse skill to find issues in traces; the agent accesses live traces to look at what questions are asked, what scores exist, and where all-caps frustration or user disagreement occurs, and also checks for issues beyond the configured evaluators — such as latency problems and tool call failures — to find where the system is silently failing in production.
AI Engineer · Continuously Improving Agents with Langfuse — Annabell Schäfer & Lotte Verheyden, ClickHouse
Report · MoneyWhen Signal finds a problem, its report lets you add the failing trace to a dataset so you can run evals on it, or create an evaluator based on the problem. It creates a GitHub issue completely describing the problem — which trace went wrong, how to find it, and where to fix it. Connected to your repo, Signal can open PRs directly with suggested fixes; instead of manually reading traces or evals, you review a queue of auto-generated PRs and approve or reject them.
AI Engineer · From Vibes to Production: Evaluating and Shipping AI Agents That Work 201 — Laurie Voss, Arize AI
Report · MoneyTo close the loop from observability to automatic fixing, the speaker asked the agent to fix the specific identified problem. When the agent asked for broad permissions, hitting yes-to-all would upset his head of security, so the agent's proposed actions were reviewed before approval.
AI Engineer · From Vibes to Production: Evaluating and Shipping AI Agents That Work 201 — Laurie Voss, Arize AI
Report · MoneyDORA change failure rate impact is volatile. AI has changed the amplitude of this volatility, with some companies seeing up to a 2% increase (potentially 50% more defects). This volatility exists without AI but is amplified by it, influenced by release pipelines and testing.
moderate
AI Engineer · The State of AI in Software Development: Data from 400+ Orgs — Justin Reock, DX
Recommendation · MoneyThe process involves taking written material, often a skill MD file with scripts, references, and assets, and moving it into an actual graph structure. Giving the agent access to a graph structure with nodes representing skill decisions, routing points, and scripts to be executed, and forcing the agent to run that graph, results in increases in task reward and performance. Compiling prose down to deterministic scripts is part of the process that leads to increased task reward and performance.
AI Engineer · Make Your AI Agents Predictable With Graph Engineering - Zachary Blumenfeld, Neo4j
Recommendation · MoneyYou can take a context graph and distill a skill from it, keeping it connected in the same graph to all past decision traces and data. Distilling a skill from a context graph gives you performance benefits plus provenance back from every step, versioning, tracking, and governance over time where you can assess drift. You can understand where certain tool calls are coming from and their history when using graph-based skills.
AI Engineer · Make Your AI Agents Predictable With Graph Engineering - Zachary Blumenfeld, Neo4j
Method · MoneyZachary Blumenfeld describes compiling the information from successful agent runs into a graph YAML shape inside a skill, and giving the agent a CLI to run that programmatic skill compiled from documentation or the memory graph, rather than building the loop inside the agent.
Method, not a verified protocol.
AI Engineer · Make Your AI Agents Predictable With Graph Engineering - Zachary Blumenfeld, Neo4j
Recommendation · MoneyKatie Givens recommends pulling six to eight quarters of baseline data so you can see trends at the organization rather than judging a single snapshot, and analyzing sales metrics on a 6-8 quarter trend line because that's where the most informed decisions come from. She uses the trend to establish a baseline and see where the organization has dropped off — there is no single 'right' number for metrics like sales-cycle length.
30 Minutes to President's Club | No-Nonsense Sales · #622: Sales Enablement Playbook: How To Win Your First 90 Days
Recommendation · MoneyWill Brown recommends collecting traces from using agents; the traces contain your prompts and your feedback to things models did.
AI Engineer · Your Company Should Train Its Own AI Models - Will Brown, Prime Intellect
Recommendation · MoneyWill Brown recommends that a really useful starting point for an organization going from the abstract idea of training models on their data to actually getting there is evals as essentially test cases.
hedged
AI Engineer · Your Company Should Train Its Own AI Models - Will Brown, Prime Intellect
Recommendation · MoneyMolly Graham advises never shipping AI output straight to your boss without iterating or editing it — the way you would never forward an intern's presentation untouched. Treating AI output like it came from the best employee ever leads to copy-paste-send, which produces AI slop.
Lenny's Podcast · Molly Graham: The grief, burnout, and opportunity hiding inside the AI transition
Recommendation · MoneyMolly Graham recommends acknowledging that AI is not a super-intelligent being — it is much more like an intern. Literally think of it as an intern: ask what you would give to an intern and what you would never give them. She frames it as a tool to get work done while recognizing the work is still done mostly by humans and human brains.
hedged
Lenny's Podcast · Molly Graham: The grief, burnout, and opportunity hiding inside the AI transition
Recommendation · MoneyMolly Graham says the biggest skill to build right now is asking yourself 'can AI help me with this?' before doing any task. She compares it to meditation's stimulus-response gap: it creates space between the task and your response, letting you leverage AI tools in new ways. She also says to always think about what's the best use of your time, and then what an agent or robot could do for you.
Lenny's Podcast · Molly Graham: The grief, burnout, and opportunity hiding inside the AI transition
Recommendation · MoneySagar Batchu recommends modeling the agent as a principal in its own right with its own identity. Record an operator — who is actually interacting with the agent in real time — for attribution, but never use it to widen access in real time. Also record the authorizer: the human who approved the original credential setup for the agent, fixed for the credential's lifetime. He presents this trio as what compliance really wants: a true log of what actually happened.
AI Engineer · A Valid Key Let an AI Agent Delete Production - Sagar Batchu, Speakeasy
Method · MoneySagar Batchu walks through a concrete control-layer flow: run secrets detection on user prompts and tool requests with the desired action being a block or quarantine. Before the model ever sees a key being sent, a hook fires in the harness runtime that detects it; the prompt is caught by a policy evaluated by the control layer, enforced by a detector that matches the AWS key. The interception and decision should happen in 100 milliseconds or less so the agent or human doesn't change their path. Once the interception happens it's logged, creating an audit of the event — inspected, blocked, and logged; the secret never left the machine.
The 100ms target is stated as an ideal ('ideally we want it to be, like, 100 milliseconds or less').
Method, not a verified protocol.
AI Engineer · A Valid Key Let an AI Agent Delete Production - Sagar Batchu, Speakeasy
Recommendation · MoneyKarri Saarinen recommends maintaining a shared, accessible place where customer context, product thinking, and learning live — important for both people and AI agents, who can query it to answer questions. He discourages skipping it: without it, 'What do we have about this?' has no answer. The space can also cover learning about anything else you're doing, not just customers and product.
Lenny's Podcast · Context is now the product: Product leadership when software can build itself | Karri Saarinen
Recommendation · MoneyRyan Cooke advises that output metrics — percentage of PRs, number of PRs, or amount of code generated into production — can disguise how well AI coding systems are actually working, since a rising PR count may belie an overall increase in PRs and make it hard to tell whether AI output is driving outcomes. Instead, measure outcome metrics such as whether the factory is accelerating the ability to deliver features, and think first about how the investment's success will be measured and whether it drives value for the engineering organization.
AI Engineer · No, That's Not a Software Factory — Ryan Cooke, WorkOS