Amanda: AI far better used for editing marketing content that *you* are the author for versus "Make me a piece which…" If you start using AI for "good enough" content, you will forget what "good" actually is.
@patio11
-
LLMs Defer to Official Web Surfaces When Describing Brands
By
–
Amanda: An increasing portion of customers are getting their first exposure to you through a chat with an AI agent you don't control. Me: Generally speaking my experience is leading LLMs are *strikingly* deferential to one's official web surfaces in describing one, by default.
-
Meta Internal Data Shows Platforms Suppress External Link Traffic
By
–
Meta's internal data: 97.3% of views are to posts which do not have an external link. "Platforms do NOT want to send you traffic." Amanda represents Meta tells business ads users to put link in comments. (I have no reason to doubt this representation.)
-
AI Rapidly Gaining Ground in Global Search Market Share
By
–
Traditional search engines (Google) down to about an 80% share of esearches. Commerce platforms (Amazon etc) 10%, social networks 5%, AI 2% and growing rapidly.
-
AI Token Costs Drive Outcome-Based Pricing in SaaS
By
–
In response to a question: Anthony: Given how tokens work (and become COGS), expect more outcome based pricing in SaaS companies. Though, even the AI labs have fixed-price subscriptions for some of their products.
-
AI Tooling Will Boost Software Reliability for Small Teams
By
–
Me: I actually expect that software reliablity goes up rather than down over the next 2-3 years as the tooling for this gets just radically better and weirder than anything we've had before. Small teams are going to get BigCo levels of reliability. Not perfect but Really Good.
-
Operational Excellence as Moat in AI-Written Code Era
By
–
Anthony: "Operational excellence" is a defensible moat. As more code is AI written, quality inevitably goes down. Most people using one-off apps are going to get burned, repeatedly, and they will try to offload that to someone whose only job is This Can't Blow Up.
-
Vibecoding Clashes With Team Collaboration in Software Development
By
–
Anthony: Software which has collaboration within a single user (e.g. multiplayer mode in a company) is intrinsically resistant to "vibecoding" one's own solution on one's own desk. Me: Everyone loves vibecoding; nobody likes taking a meeting to synchronize vibecoders.
-
AI Will Crater the Cost of SaaS Feature Development
By
–
Anthony: We once considered feature development to be a core part of running a SaaS business. Required both some amount of taste and relatively large time/money due to engineers. Cost of features will *crater.*
-
SaaS Founders Strategies for Thriving in the LLM AI Era
By
–
Next up at MicroConf: @aeden of DNSimple (I use them for DNS and love them) on how SaaS founders can thrive in the LLM age of AI-generated apps.