Demo project by Thryve Apex. Businesses and bookings are fictional, and data resets daily. All demos · Build an app like this →
← All demos

Demo app

Mini CRM with AI lead scoring

A lead pipeline for Brightwave Solar, a fictional UK solar installer. Submit an enquiry on their website form, then watch it arrive in the CRM, scored and ranked so the team calls the best leads first.

For the sales team

Open the CRM

Pipeline, "call these first" list, KPIs, lead pages with AI analysis.

Open →

For customers

Submit a test enquiry

The business's quote form. Your lead appears in the CRM straight away.

Try it →

Every lead in one pipeline

Google Ads, Meta Ads, WhatsApp, website and referrals, each with its source kept for ROI reporting.

Instant lead scoring

Transparent rule-based score the moment a form is submitted: timeline, budget, property, bill, intent.

AI scoring on top

Claude or your own local LLM reads the enquiry and returns a score, summary, risks and the next best action.

Draft replies

A ready-to-send, on-brand reply for every lead, so the team answers in minutes, not days.

Bring your own AI

Add a Claude API key or point it at Ollama / LM Studio. Keys stay in your browser.

Tracking built in

generate_lead and AI events go to the dataLayer for GA4, Google Ads and Meta conversion tracking.

How the AI works

  1. Open AI settings and add a Claude API key, or connect a local LLM.
  2. Open any lead and click Analyse with AI, or AI-score all new leads from the pipeline.
  3. The model returns structured JSON (score, temperature, summary, signals, risks, next action, draft reply), saved on the lead.

Stack: Next.js 16, React 19, TypeScript, Tailwind CSS, Anthropic SDK (structured outputs), OpenAI-compatible API for local models.