How We Work.

AI-first by default, a small set of canonical systems, and a steady cycle-based cadence across a remote, global team.

AI-first by default

When building new features, the default question is: “Can we use AI to do this?” AI-first isn't optional — it's the default. The point isn't to replace people; it's to supercharge speed and output so we spend our time on architecture, design, and high-value problem solving.

From the team, on the engineering mindset: “The expectation is that AI handles the feature end-to-end. Use Claude Code or the Cursor Agent to generate full implementations by default. What matters most is giving AI the context to succeed.”

What “giving AI the context to succeed” means in practice:

  • Write detailed product, design, and engineering requirements in Linear.
  • Spell out use cases, edge cases, and success criteria.
  • The clearer the ticket, the more AI can deliver production-ready code.

This applies beyond engineering — design, sales, and operations all lean on AI tools, and routine reporting is increasingly automated.

Where the truth lives

A small number of systems are canonical. When something is documented in one of these, link to it rather than copying it — copies drift.

NotionSource of truth for the company knowledge base — company overview, people policies, security & compliance, hiring & levels, playbooks, and SOPs. (The Outline→Notion migration completed June 2026; Outline is retired.) AGENTS.mdSource of truth for engineering conventions, in-repo. Agents and engineers read it before writing code. LinearProduct, design, and engineering requirements and tickets — the context AI builds from. AttioCRM — deal stages and customer records. this wikiinternal.perkup.com — the cross-team internal reference, linking out to the canonical systems above.

How we operate

  • Create momentum — don't sprint. We work in cycles: decide priorities, assign responsibilities, and maintain a healthy, sustained pace rather than rushing toward a finish line.
  • Remote and global. The team spans North America, Latin America, North Africa, and the Philippines. Default to public channels and writing so work is transparent and async-friendly — see How We Use Slack.
  • IC pace and ownership. People own meaningful systems end-to-end. The expectation, especially in engineering, is to own — not just maintain.
  • No fabrication. When we write docs, proposals, or customer-facing material, every fact, quote, and price traces to a real source. If a source isn't confirmed, we say so rather than inventing.

Growth & feedback

Growth runs on a few recurring rituals, documented elsewhere in People:

  • 1:1 Meetings → — the most important recurring meeting; the team member owns the agenda.
  • Function Levels → — what's expected at each stage of a role, and the shared language for growth.
  • Onboarding → — how a new hire gets set up and read in on how we work.

Source: Principles (Notion) · “AI-First Engineering Mindset” and the Outline→Notion migration announcement (#engineering, #general, PerkUp Slack) · engineering conventions in the repo AGENTS.md files.