Backflip: A Foundation for Building With AI
The real untapped potential in software isn't consumer apps - it's internal tools that never get built. Here's how AI-assisted development changes that equation.
Listen to under 2-min AI recap •TL;DR ↓

This article contains affiliate links. If you sign up through them, I may earn a commission at no extra cost to you.
How I came up with "Backflip"
I've been watching the software industry closely over the past year, and one pattern keeps standing out: there are so many good ideas that never get built - not because they can't be, but because they aren't.
Building a website or platform has become affordable and accessible for individuals. But the biggest untapped potential? It's inside businesses and companies, internally. This is where most organisations are leaving value on the table.
The Internal Workflow Problem
Think about it. There are processes and workflows that people grind through every day. In the age of AI, data gets copied across platforms, different AI tools get used in different ways, and while AI can handle a lot of the heavy lifting, the ways of working have to change - and are changing.
Individuals and entire teams could be forming their own workflows with custom automation steps, tailored solutions for their specific work. But development departments are perpetually short on capable hands. Sound familiar?
Now, with AI, many things have become simply doable. But a new problem surfaces within established businesses: how do we allow people to build - and get them to build - in a sensible and safe way? Especially when we're talking about non-technical individuals without developer backgrounds.
Who is the Backflip built for
Firstly this is not SaaS, this is the platform baseline I created and open sourced for staring projects from a meaningful foundation. Backflip targets exactly that audience: people who want to build by themselves but also need to retain a certain level of quality and structure. It's designed so DevOps teams can embed it into company infrastructure, and the project itself can be gated behind SSO - in this case, I've built it around Google Login.
This doesn't just allow you to own it. The platform self-documents when you use Agentic Coding (I use Claude Code), and it holds to a standard that developers can pick up at any point - most likely using the same AI tools. With self-documenting code and clean commits, decisions don't get lost.
The Message Matters as Much as the Product
Backflip isn't just a product I built. It's a message I want to broadcast: you can build something like this within your company for your specific needs, enable people. Or you can simply fork or clone Backflip and adjust it to yours, building on top of what's already there.
You save all the project bootstrapping time. There are DigitalOcean Droplet configuration scripts and deployment scripts included. With a bit of extra effort configuring deployment pipelines, you get a fully development-enabled loop you can control from your phone using the Claude and GitHub mobile apps.
Local development requires Docker and Git. And since I built it with Claude Code in mind first, that's the recommended agent - but you can try other agents and reuse the same agentic development skills and rules already defined within Backflip.
What You Get Out of the Box
AI provider integrations - the project is prepared for them. Enter an API key into Backflip's admin panel and tell your Claude Code: "Build me a new admin section that generates a quote of the day using the Claude API, saves it to the database, and displays it on the public homepage." Swap Claude for Gemini or GPT - it's ready.
Authentication flexibility - the platform comes with user/password login. Configure Google Login keys and it enables Google SSO as well. If a user's email isn't entered in the admin panel, Google SSO won't let them in. The owner decides who gets access. Want to change that behaviour? Ask your AI agent. Want to disable user/password login entirely? Ask your coding agent.
Zero-friction start - once you've cloned the repo and have Docker installed, you can ask Claude Code to start it for you. Same applies to deployments.
Where to Find It
- Repo (clone or fork): github.com/dev-geddy/backflip
- Live demo (no public admin login): backflip.dev-geddy.com
- Getting started wizard: backflip.dev-geddy.com/getting-started
I'm confident this is already being cloned. No official forks yet - some adaptations are probably happening outside the GitHub ecosystem entirely. And that's exactly the point.
This Is for Everyone
While Backflip is an enabler for non-developers, it can save a lot of time for developers too - especially when kicking up a Next.js-based project of their own.
It all comes down to your imagination and how you structure your vision for what you're about to build.
Now go build something. Then come back and tell me on LinkedIn how it worked for you.
What's the one internal workflow at your company that you've always wished someone would just build a tool for - and what's actually stopping you from building it yourself?
TL;DR
- The biggest untapped opportunity in software is internal company tools and workflows, not consumer apps.
- Development teams are perpetually understaffed, so non-technical employees need a way to build their own solutions safely.
- Backflip is an open-source platform designed to let non-developers build within guardrails that DevOps teams control, including SSO and self-documenting agentic code.
- It ships with AI provider integrations, Google SSO, Docker-based local dev, and DigitalOcean deployment scripts - ready to fork and customize.
- The real product isn't the platform itself - it's the proof that any company can stand up something like this for their own specific needs.
Stop waiting for your dev team's backlog to clear - the tools to build it yourself already exist.
Geddy
Senior Web Engineer / Lead
Engineering leadership • AI innovation • Product thinking. 20+ years of web engineering, from independent contractor to engineering leader. Passionate about developer experience and product engineering.