Inspiration
Closer is the CRM that fights scope creep for you. It is an AI-powered command center built for freelancers, agencies, and consultants who lose real money every month to the same client pattern: "Hey, can you also add…" Last year, freelancers lost $47 billion to scope creep. Closer ends that, one client message at a time.
The Problem And Why I Built It
Most CRMs are passive filing cabinets. They store contacts, list deals, and wait for you to do the work. Meanwhile, the actual money leaks happen between the apps, in Slack threads, Gmail inboxes, and "quick question" messages where a five-page redesign quietly becomes a five-page redesign plus PayPal integration plus Apple Pay plus a marketing site. By the time you notice, the weekend is gone and the invoice is the same. I built Closer because I wanted a CRM that reads every client message, flags the moment something falls outside the original scope of work, and drafts the change-order reply and the invoice line item in one click, before the work is done for free.
How I Used MeDo To Build It
I structured my work in MeDo as a series of focused, narrative prompts rather than a single mega-spec. The flow was roughly: first, define the universe (a freelance agency owner named Sarah, her clients, her projects, her revenue), then layer one module at a time on top of that seed data. I started with the dashboard and the 6-stage pipeline (Lead, Qualified, Proposal, Negotiation, Won, Lost), then added contacts with active/at-risk/silent health scoring, then layered on invoices, tickets, and alerts. Once the foundation was real, I added the hero feature, the Scope Creep engine, and finally the AI Assistant, the Email Threads parser, the Reports module, the Guide, the Integrations panel, the Audit Log, and a Test Lab for validating the AI detection logic. Every prompt assumed the previous state, so MeDo kept building on a coherent, seeded workspace instead of starting from scratch each time.
The single feature I am most proud of is the Scope Creep engine itself, which MeDo generated when I told it the product should "catch the money before it leaks." The fact that the same engine also has a Test Lab with built-in test cases (Obvious Scope Creep, Subtle "While you're at it" Creep, False Positive, Overdue Invoice) so you can validate the AI's judgment, that is the part that genuinely surprised me. MeDo did not just build a feature, it built the feature plus the QA harness around it.
What You Can Try In The Live Demo
No signup, loads in two seconds. Open Closer, hit Try the Live Demo, and you land in the Demo Agency workspace with $24,834 in tracked revenue, 12 deals across the pipeline, 8 contacts with live health scores, 10 invoices, 5 alerts, 15 email threads, and 4 scope creep detections worth $15,500 in captured revenue, ready to explore. Then ask the AI Assistant "what projects are at risk?" and watch it answer from your own data, not from the internet. That is Closer. Built with MeDo. Zero code.
How I Used Plugins And API Integrations To Extend Functionality
Closer is built on a small set of focused integrations that each do one job well, wired together through MeDo. The voiceover for the demo is generated by the ElevenLabs Text to Speech API (Roger voice, Eleven Multilingual v2 model), which lets a solo builder ship a studio-quality narration without a microphone or a hire. For the AI Assistant inside the app, MeDo wires the chat surface to an LLM endpoint that is grounded in Closer's own data, so when you ask "what projects are at risk?" it answers from the live pipeline, contact health scores, and alerts rather than the open internet. The Scope Creep detector is the same pattern in reverse: every inbound email thread is passed through an LLM classification call with a structured prompt that returns is_scope_creep, suggested_charge, and a drafted reply, which Closer then renders as a Review modal with Accept, Reject, and Add to Invoice actions. On the money side, the Integrations panel is wired for Stripe and PayPal so that an accepted scope change becomes a real invoice line item, and the Email Threads module is designed to plug into Gmail and Outlook through their official APIs so the inbox itself becomes the trigger surface for the whole CRM. Slack and a generic webhook channel handle outbound alerts (overdue invoices, stalled deals, new scope creep detections) so the team is notified where they already work. Finally, the Test Lab uses a lightweight internal API to replay seeded fixtures (Obvious Scope Creep, Subtle "While you're at it" Creep, False Positive, Overdue Invoice) against the detection engine, which turns the AI from a black box into something a non-technical owner can actually trust. The point of every integration is the same: keep the human in the loop, but remove the manual copy-paste between tools that is where freelancers normally lose money.
CHECK IT OUT HERE: https://docs.google.com/document/d/1drLVP6pHcGQUSkZu5Aco3RhURB9XV6KHUzcBKYKRIck/edit?usp=sharing
Log in or sign up for Devpost to join the conversation.