The New Way To Build A Startup

Predictions and Expectations:

• The speaker predicts that startups mastering internal automation will "win" the market because they will be able to "beat huge incumbents" through "leanness."
• The speaker expects that companies building "AI teammates," "integrated source of truth," and "custom agents" will achieve "record-high growth rates" while "staying lean."
• The speaker believes that "compounding software businesses" offer a "much more powerful type of product market fit" that is "much harder to displace."
• The speaker anticipates that internal AI automation allows "20x companies" to "postpone hiring additional sales and ops staff for much longer."
• The speaker predicts that "AI FTE" (like Atlas) will "dramatically expand the range of what each engineer can take on," effectively doubling or tripling an engineer's scope.
• The speaker expects that "PhaseShift" and similar startups can "avoid hiring" for entire functions, such as design, by leveraging "magic patterns" and AI agents.

Timelines and Milestones:

• The speaker notes that a competitor has been "around since 2006," contrasting it with PhaseShift's current "12-person" timeline.
• The speaker references that Legion Health has "grown 4X in the past year" while maintaining flat headcount.
• The speaker mentions that a pilot with "multiple Fortune 500s, 10 plus Fortune 500s" is currently active.
• The speaker states that GigaML closed DoorDash as a customer when it had "approximately like four to five engineers."

Technology and Product Direction:

• The speaker describes a direction where AI agents like "Atlas" can "use browsers," "edit the policies," and "write code" to perform tasks previously requiring manual intervention.
• The speaker outlines a product strategy where companies build "custom internal interfaces" that pull "patient history, scheduling availability, insurance codes" into a single source of truth to manage complex healthcare workflows.
• The speaker expects companies to move from automating "one or two internal functions" to "automating all of them," including "code, support, marketing, sales, hiring, [and] QA."
• The speaker suggests a future workflow where employees "document the manual tasks they do" to enable the creation of "quick AI agents" tailored to specific needs.

Market and Industry Outlook:

• The speaker predicts that "tiny teams" utilizing internal automation will be able to "beat huge incumbents" because "leanness is their superpower."
• The speaker believes the current trend represents a "fundamental shift in how startups operate right now" driven by the rise of "20x companies."
• The speaker observes that the market includes "companies that have been around since 2006 that have hundreds of employees" facing competition from small, AI-enabled teams.

Company Plans:

• Legion Health plans to keep its "ops headcount flat even as it's dramatically scaled revenue" by utilizing a "single source of truth interface."
• PhaseShift plans to "delay hiring for entire functions" by having employees document manual tasks and deploying "custom agents."
• GigaML plans to service "dozens of accounts" with a single human FTE in tandem with an AI employee named "Atlas."
• The speaker notes that "20x companies" intend to allow employees to focus on high-value tasks rather than "boilerplate stuff."

Financial Guidance:

• The speaker cites GigaML's capability to beat competitors "20x" their size with "four to five engineers" against "100x engineers."
• The speaker states that Legion Health achieved "4X the number of patients" while "not hiring a single net new person."
• The speaker mentions that Fortune 500 clients like DoorDash process volumes of "over like 500,000 or a million calls a day."

Risks and Caveats:

• The speaker cautions that without automation, companies face the risk of having their culture "drifting" due to excessive hiring.
• The speaker notes the difficulty of the task, stating "it's hard as it's to believe" that a company with one person can service such high volumes.
• The speaker acknowledges that traditional healthcare workflows involve communications that "might've otherwise gotten lost in the sea of different communications."

Confidence and Disagreement:

• The speaker is highly confident, stating that "the startups that figure it out first are going to win."
• The speaker uses strong language to describe the shift, calling it a "fundamental shift" and asserting that "all of them" must be automated for success.
• The speaker expresses certainty in the efficacy of the approach, noting that Atlas allows each engineer's scope to be "basically doubled or tripled."