Make with Notion 2025: The startup stack: AI-first founder playbooks (Karan Goel, Marty Kausas)

AI-first startups are scaling faster by treating their operating system as a single source of truth—collapsing “software sprawl” into one workspace where product, engineering, go-to-market, and customer-facing teams can all move with the same context.

The session opened with a problem statement: modern startups often juggle 112+ tools just to run day-to-day work, creating constant context switching, slower execution, and missed opportunities. The proposed fix is operational consolidation—using Notion early so teams capture customer notes, product requirements, and fundraising progress in one place. That shared context is framed as a direct driver of lower costs, faster shipping, and a longer runway.

Marty Kausas, CEO and co-founder of Pylon, tied that operating principle to a company built for speed. Pylon focuses on B2B customer support and related functions—customer success, solutions, professional services, and account management—using an AI-native platform. After a period of “pivot hell,” the company found traction quickly: entry into YC, first customer within weeks, and rapid ARR growth, followed by multiple funding rounds including General Catalyst and later Andreessen Horowitz. Today, Pylon is shipping at a high cadence—about 100+ PRDs per week and two feature launches every day—while publicly tracking progress on LinkedIn.

The operational backbone behind that pace is a Notion-based feature and launch system. Pylon organizes every shipping item in a single database table with fields for internal and external launch dates, engineering and marketing owners, and the feature details themselves. Multiple views let different teams slice the same information: engineering tracks ownership and timelines, sales and customer success answer “when is it coming?” questions, and product marketing separates internal launches from public ones for coordinated campaigns. The team also uses Notion’s Ask AI to answer ad hoc questions with the right launch context, reducing the need to hunt across tools. Pylon’s content engine—roughly 750 posts in the past year—runs on the same alignment process, and the company even shares a replicated “feature update playbook” template.

Karan Goel, CEO and co-founder of Cartisia, described a different startup built from research to production: voice AI systems and interactive conversational models. Coming out of Stanford’s Hazy Research Lab, Cartisia aims at long-lived AI collaborators and companions that can retain context over time. The company launched its first model product, Sonic (text-to-speech), and later expanded with Sonic 2, while also building audio understanding and transcription capabilities.

Cartisia’s Notion usage emphasizes knowledge centralization and workflow flexibility for a mixed team: about 40% research, 40% engineering, and 20% go-to-market, recruiting, and bizops. Notion supports daily standups via a company-wide page that evolved into team-level standup docs as the company grew. It also powers onboarding directories so new hires can quickly find answers in their first week—using Notion AI to navigate links, pages, and “who to talk to” guidance. Beyond internal coordination, Cartisia uses Notion templates to package model evals and benchmark collateral for customers and partners, keeping documents “living” as feedback arrives.

Across both companies, the shared takeaway is practical: Notion becomes the system that turns AI-driven speed into repeatable execution—providing scale, clarity, and accountability without adding more tools to manage.