To move fast, we combine multiple tools, AI agents, and systems. This lets us compress the product development lifecycle down to 1–2 days.

Here’s the high-level flow: Idea → Boilerplate → AI Planning Agents → Core Features (Claude / Codex / Gemini) → Deployment

Every tool includes repeatable features such as emails, payments, and marketing pages. To avoid rebuilding these each time, we created a modular internal boilerplate which can be seen here. This modular approach allows us to change the design very easily and focus only on the core features of the product. Once the boilerplate is set up, we are ready to go. The documentation can also be found here. The boilerplate features are outlined below:

Boilerplate features: Marketing pages: Home, About, Pricing, Blog, Contact, Services, Legal Pages Authentification: NextAuth & Google Auth Payment Emails Notifications Dashboard Structure Feature Gating SEO & GEO ready Database Setup

AI Planning Agents

AI Planning Agents act as our internal agile team.

When building with AI, strong planning is essential to ensure the development agent operates within clear guardrails. These agents live directly inside our codebase, making it easy to provide full context for the features we want to build.

A simple flow looks like this:

Analyst Agent → creates the Product Brief (http://brief.md) → PM Agent → creates the PRD (http://prd.md) → Architect Agent → creates the System Architecture (http://architecture.md) → PM Agent → creates the Epics & Stories (http://epics.md, http://stories.md)

Why are these so important? This process gives both us and the development AI agent a clear execution plan with strong guardrails. As a result, the agent does not hallucinate and builds exactly what is required, in the way it is required.

Here is an example of one story:

## Story 2.9: Send Email Notifications to Submitters on Status Changes As a *feedback submitter*, I want *to receive an email when my feedback status changes (e.g., Doing → Testing → Finished)*, so that *I know the team is working on my suggestion and can see progress*. ### Acceptance Criteria 1. When team member changes feedback item status (Story 2.5 drag-and-drop), trigger email notification 2. Email sent only if submitter provided email address during submission (FR17) 3. Email subject: "[Project Name] Update: Your feedback is now [Status]" 4. Email body includes: original feedback title, new status, team comment (if any), link to view on public board 5. Email sent asynchronously (doesn't block status update) 6. If email sending fails, log error but allow status update to succeed (NFR12) 7. No duplicate emails if status changes multiple times quickly (debounce or queue) 8. Unsubscribe link included (placeholder for now) 9. Test email delivery in development and production

Now that we have everything in place the boilerplate with all repeatable product features (login, dashboard, payments, emails, etc.) and the planning stage completed with clear focus, guardrails, user stories, and architecture we have all the context needed to build with AI (Claude, Codex, or Gemini).

In this phase, development happens story by story. With the full planning context in place, the AI agent implements exactly what is required. Depending on the number of features, we can deploy and have a live product ready for real user validation in 1–2 days.

Here is an example of what we manage to achieve:

https://startupkit.today https://founderspace.work