Autonomous Go-green★ FinalistDeep-reviewed
by Irina Vélez
What they're building
I’m building Landscape AI, an AI-powered construction and landscaping assistant that turns site photos into design concepts, material suggestions, scopes of work, and visual recommendations. I’m working with Daniel Gutierrez who have been already accepted
AI code reviewrepo: real
Real, well-engineered Next.js + Claude Agent SDK build: deterministic qualify->price->escalate->book engine, query() loop with in-process MCP tools, canUseTool escalation gate, Anthropic vision for yard photos, Stripe checkout, full test suite. A topic-matching demo video exists (YouTube title 'Go Green Landscape Maintenance AI'). BUT zero sponsor SDKs are used: no Tavily/Nebius/Composio import or call anywhere in src/app/lib — they appear ONLY in spec.md, where the team itself writes 'Nebius: Dropped from V1' and lists Tavily/Composio as optional 'demo beats (if time).' Stack is Anthropic + Stripe + Airtable + Telegram. ⚑ Sponsor stack named-only (planning doc), not in code -> sponsor_usage floored. site_url deltanova.io is a generic AI-consulting landing page, not the product. build_description pitches 'Landscape AI' image-design tool that differs from the shipped Telegram intake/pricing operator.