Recommendation · MoneyInclude a specific news and product updates category on your website, because customers search for your latest features and LLMs can retrieve the latest information from your site; whenever there are new features or small updates, push out quick press releases on the news category to get the information out.
MicroConf · SEO for AI Search: How Ajelix Grew to 350K Signups Without Ads
Recommendation · MoneyAjelix's MVP was seven AI-powered tools, each with its own landing page. They started with small, super small tools like an Excel Formula Generator and an Excel template generator to start getting traction.
MicroConf · SEO for AI Search: How Ajelix Grew to 350K Signups Without Ads
Recommendation · MoneyAjelix writes about 20 listicle articles per month per product launch, not more, because they write the stuff on their own without AI slop. They have real writers who actually go through the tools and test them out, so they give real value to users.
hedged
MicroConf · SEO for AI Search: How Ajelix Grew to 350K Signups Without Ads
Recommendation · MoneyFor every product category, define who the ICP is, how you are different, and how you are better, then plug that context into all of the listicles.
MicroConf · SEO for AI Search: How Ajelix Grew to 350K Signups Without Ads
Recommendation · MoneyAjelix made sure the interface and the product itself had zero resistance: no credit cards, you can sign up very fast, super quick value, and super clear limits and upgrade paths.
MicroConf · SEO for AI Search: How Ajelix Grew to 350K Signups Without Ads
Recommendation · MoneyAjelix hopes your web pages include a specific category for news and product updates. Whenever they have new features or small updates, they push out a quick press release on their news category to get the information out. This also helps LLMs have the latest information and context and retrieve the information from your website.
MicroConf · SEO for AI Search: How Ajelix Grew to 350K Signups Without Ads
Recommendation · MoneyIn comparison listicles, position against general-purpose tools by claiming a specific niche: 'ChatGPT and Claude are very good for general use, but we are specifically good for business users, for those who need to create reports or administrative tasks.'
MicroConf · SEO for AI Search: How Ajelix Grew to 350K Signups Without Ads
Recommendation · SocialAs part of making it about you, the host asks listeners to reflect: do I want to be with someone who leaves me on red for days? Do I want to be with someone who makes me feel anxious, insecure, or sometimes even embarrassed? Her stated answer is no.
Call Her Daddy · The Situationship Survival Guide
Recommendation · SocialThe host advises challenging the fantasy by asking for an example of what he's actually done — because you can't give one.
Call Her Daddy · The Situationship Survival Guide
Recommendation · MoneySagar Batchu discourages blocking on PII, saying it will probably grind your business to a halt because it's hard for people not to use agents with PII. When you turn on observability, there's a ton of PII going through agents. The more important thing to look at is what is happening to the PII — the agent's intent: will it post it in a public Slack channel, send it via email, or put it back into a tool call to another system? He recommends using observability to define policies about PII intent rather than just blocking or redacting all PII altogether. Customers often look at PII findings in the product even without viewing redacted data.
The claim that blocking PII will 'probably' grind the business to a halt is hedged by the speaker.
AI Engineer · A Valid Key Let an AI Agent Delete Production - Sagar Batchu, Speakeasy
Recommendation · MoneySRE workflows break down into three core areas: investigation, prevention, and remediation. On the investigation front, Sourcegraph has the tooling to find the root cause of issues in minutes instead of days. Issues created by AI-written code are much easier to investigate and find with classic deterministic search, which allows you to search across every repo, every branch, and every commit with a single query. DeepSearch, Sourcegraph's semantic search feature, allows you to easily ask what's broken, get cited answers, and be confident in the responses you're receiving.
AI Engineer · AI Writes Code Faster Than Your Team Can Review It - Andrew Norrish, Sourcegraph
Recommendation · MoneyIt is even better to prevent issues from occurring in the first place than to find their root cause quickly. Sourcegraph has built-in monitoring capabilities to alert when risky code is committed or when certain non-compliant patterns are matched. Through these code monitors, you can ensure that AI-written code follows all of the patterns and requirements your organization has in production.
AI Engineer · AI Writes Code Faster Than Your Team Can Review It - Andrew Norrish, Sourcegraph
Recommendation · MoneyAfter identifying a break or risky pattern, make sure the fix is applied quickly to all impacted areas of the code. Sourcegraph tooling can make bulk changes across all repos within an entire organization to remediate issues and prevent them from recurring across other services and teams.
AI Engineer · AI Writes Code Faster Than Your Team Can Review It - Andrew Norrish, Sourcegraph
Recommendation · MoneyKarri Saarinen describes the feature roast, an optional meeting for new features that anyone in the company can join, hosted by the team building the feature. Attendees critique the whole feature — they can be as nitpicky or confused as they want, giving raw, non-personal feedback — and the building team can interact with them and ask questions. The lead then synthesizes the feedback, groups it, and turns it into actionable issues. Because many company members see the feature for the first time, if they are confused, users likely are too, giving the team a realistic view of how users might think.
hedged
Lenny's Podcast · Context is now the product: Product leadership when software can build itself | Karri Saarinen
Recommendation · MoneyKarri Saarinen recommends asking AI agents questions like 'What do we have about this?' to surface existing company knowledge before starting new work, and 'Teach me something about the customers — what are they saying, what are they doing, how is the team responding?' to build customer understanding.
hedged
Also discussed by Randy Hunt on Design Better (unclassified source — not counted until classified). Not counted as independent sources.
Lenny's Podcast · Context is now the product: Product leadership when software can build itself | Karri Saarinen
Recommendation · MoneyKarri Saarinen describes Quality Wednesday, a practice where everyone is tasked every week to spend time looking into the product, find one quality defect, and fix it. Defects can be very small — a weird hover state, a janky animation, a copy error — a five-minute fix counts. Findings and fixes are shared in a meeting so everyone learns how others find problems; the bigger impact is not the fixes themselves but that everyone trains their eye to notice small mistakes.
Lenny's Podcast · Context is now the product: Product leadership when software can build itself | Karri Saarinen
Recommendation · MoneyRyan Cooke advises that as you do work you learn where the gaps in the plan are, so continuously keep the plan fresh by using the agent itself to determine missing pieces of the project. In Linear, define tickets that have dependencies — one ticket blocks another — as a common way of breaking down large units of work into smaller ones, and when a ticket is completed, ask TARS to reevaluate the Linear project and report if it is missing new tickets. Through the planning process, construct a map of how a plan is expected to execute so the agent can start executing on it autonomously.
AI Engineer · No, That's Not a Software Factory — Ryan Cooke, WorkOS
Recommendation · MoneyBecause WorkOS owns the infrastructure, they can point agents at sessions to find gaps in how people are using the factory: where engineers can improve their skills, which skills should be built, where the agent made a mistake, and which skills are no longer relevant — maybe a skill written six months ago is now obsolete because the code has changed. Continuously verify skills against the current codebase, since skills written months ago can become obsolete as code changes. Semi-online evaluation is where owning all the infrastructure shows a big advantage.
hedged
AI Engineer · No, That's Not a Software Factory — Ryan Cooke, WorkOS
Recommendation · MoneyRyan Cooke suggests pointing Devon or local Claude Code (Opus) at the agent-created tickets and documentation — via MCP — to implement parts of the work instead of always using the factory's coding agent.
AI Engineer · No, That's Not a Software Factory — Ryan Cooke, WorkOS
Recommendation · MoneyRyan Cooke recommends using TARS to triage support requests from customers in shared Slack channels. TARS can understand where a customer might be running into problems with the product by looking at the code.
hedged
AI Engineer · No, That's Not a Software Factory — Ryan Cooke, WorkOS