// PRODUCTS SECTION — productized use cases Rocketberry has shipped. // Each product is "Built for" one of the clients in CLIENTS. // Click a card → lightbox with the full case study (challenge → solution → outcomes). const PRODUCTS = [ { id: 'invoice-intelligence', name: 'Invoice Intelligence', clientId: 'northwind', artKey: 'Documents', color: '#D6248C', family: 'Smart Document Processing', tagline: 'AP automation for a 12-line stamping facility', blurb: 'Auto-extract, validate, route and post 8,000+ vendor invoices/month across 3 ERPs — no human touch on the happy path.', tags: ['Document AI', 'OCR', 'ERP'], stat: '98.6% accuracy', headline: 'From 14 min per invoice to 12 seconds.', challenge: 'Northwind processed 8,000 vendor invoices per month across NetSuite, Tally and a legacy system. Average handling time: 14 minutes. Posting errors cost ~$120K/year in reconciliation work.', solution: 'Rocketberry built a multi-tenant document agent: OCR + LLM extraction tuned on 142 invoice fields, dual-validator confidence scoring, exception queue, and direct API writebacks to all three ERPs.', results: [ '98.6% extraction accuracy on production traffic', '−87% manual handling time per invoice', 'Posting errors down from 4.1% to 0.3%', '6-week pilot to full production rollout', ], }, { id: 'routemind', name: 'RouteMind', clientId: 'velosys', artKey: 'Workflow', color: '#A14BCC', family: 'Workflow Automation', tagline: 'Live route optimization across 4 hubs', blurb: "Re-optimizes 600 daily routes every 15 minutes against traffic, capacity and delivery windows. Dispatch went from manual to assistive.", tags: ['Optimization', 'Real-time', 'Logistics'], stat: '−24% miles', headline: 'Same fleet. 24% fewer miles. 19% more on-time.', challenge: 'Drivers averaged 8% empty miles. Customer ETA accuracy was ±42 minutes. Four regional dispatchers built daily routes by hand each morning.', solution: 'RouteMind re-optimizes every 15 minutes against live traffic, capacity, vehicle profile and customer windows. A separate ETA model predicts arrival within ±6 minutes; both decisions are written back into the TMS.', results: [ '−24% miles driven across the active fleet', 'ETA accuracy ±6 minutes (was ±42)', '+19% on-time delivery rate', '$2.4M fuel savings in year one', ], }, { id: 'priorauth-pilot', name: 'PriorAuth Pilot', clientId: 'helio', artKey: 'Chatbots', color: '#7027A8', family: 'Healthcare AI', tagline: 'Specialty-care prior-authorization agent', blurb: 'Drafts, predicts and submits prior-auth packets across payer portals — nurses only see the exceptions.', tags: ['NLP', 'EHR', 'HIPAA'], stat: '−68% admin time', headline: 'From 4 days to 19 hours of patient wait.', challenge: 'Prior-auth submissions averaged 18 minutes of nurse time. 37% required rework. Patients waited 4.2 days on average for approval to begin treatment.', solution: 'A HIPAA-aligned document agent reads physician notes, EHR data and live payer policy. It drafts the prior-auth packet, predicts approval likelihood, and submits via payer APIs. Nurses review only flagged or escalated cases.', results: [ '−68% admin time per case', '+41% first-pass approval rate', 'Wait time 4.2 → 0.8 days', '94% nurse satisfaction score', ], }, { id: 'forecast-pulse', name: 'Forecast Pulse', clientId: 'atlas', artKey: 'ERP', color: '#1F2754', family: 'AI-Enabled ERP', tagline: 'SKU-store-day demand forecasting for fresh', blurb: 'POS + weather + local events + competitor pricing → daily replenishment recommendations across 220 stores.', tags: ['Forecasting', 'Time-series', 'Retail'], stat: '−42% waste', headline: 'Fresh waste cut by 42%. Margin up $11.4M.', challenge: 'Atlas was bleeding 8.4% of fresh category revenue to waste. Out-of-stocks on top SKUs hit 11%. Forecast accuracy at SKU-store-day level sat at 64%.', solution: 'Forecast Pulse ingests POS, weather, local events and competitor pricing. Recommendations flow into the existing replenishment system with category-manager override controls. The model retrains weekly per region.', results: [ '+18% sell-through on top-velocity SKUs', '−42% fresh-category waste', '−61% out-of-stocks on hero items', '$11.4M margin uplift in year one', ], }, { id: 'underwriter-copilot', name: 'Underwriter Copilot', clientId: 'parallel', artKey: 'StartupOps', color: '#D6248C', family: 'Conversational AI', tagline: 'AI copilot for SMB credit decisions', blurb: 'Summarizes financials, flags risk signals, drafts the credit memo — underwriters approve or override in one screen.', tags: ['LLM', 'Risk', 'Finance'], stat: '3.2× throughput', headline: 'Underwriters do 3.2× the deals — without cutting corners.', challenge: 'Underwriters spent 90 minutes per file on data gathering and memo drafting before getting to actual decisioning. Approval cycle averaged 6 business days.', solution: 'A copilot panel inside the existing LOS pulls bank statements, tax returns and credit data, summarizes financial health, surfaces risk signals against policy and drafts a one-page credit memo. The underwriter remains the decider.', results: [ '3.2× more files cleared per underwriter / day', 'Cycle time 6 days → 1.4 days', 'Default rate on AI-assisted approvals: −18% vs control', 'Audit trail captured per recommendation', ], }, { id: 'leaselens', name: 'LeaseLens', clientId: 'kinetic', artKey: 'Documents', color: '#A14BCC', family: 'Smart Document Processing', tagline: 'Lease abstraction across 8.2M sq ft', blurb: 'Extracts 142 lease fields from PDFs of every vintage, ties them to tenant risk and surfaces upcoming actions.', tags: ['Document AI', 'Legal NLP', 'Risk'], stat: '99.1% accuracy', headline: 'Lease cycle cut by more than half. Bad debt down 38%.', challenge: 'Lease abstraction averaged 6 hours per document. Tenant default prediction was anecdotal, leading to ~$7M in annual write-offs.', solution: 'LeaseLens extracts 142 fields with 99.1% accuracy, normalizes them across formats, and feeds a tenant risk model that scores default likelihood using financials, market signals and behavioural data.', results: [ '−54% lease cycle time', '99.1% extraction accuracy across vintages', '−38% bad-debt write-offs', '14-week portfolio rollout', ], }, ]; // ── Lightbox ──────────────────────────────────────── const ProductLightbox = ({ product, client, onClose }) => { useEffect(() => { const prev = document.body.style.overflow; document.body.style.overflow = 'hidden'; const onKey = (e) => { if (e.key === 'Escape') onClose(); }; window.addEventListener('keydown', onKey); return () => { document.body.style.overflow = prev; window.removeEventListener('keydown', onKey); }; }, [onClose]); if (!product) return null; const Art = SolutionArt[product.artKey] || ServiceArt[product.artKey]; const c = product.color; return (
e.stopPropagation()} itemScope itemType="https://schema.org/Article" > {/* Header art strip */}
{Art ?
: null}
{product.family} {client && ( <> · Built for {client.name} )}

{product.name}

{product.tagline}

{/* Body */}

{product.headline}

— Challenge

{product.challenge}

— Rocketberry solution

{product.solution}

{product.tags.map(t => ( {t} ))}
{/* Results sidebar */}
); }; // ── Section ─────────────────────────────────────────── const Products = () => { const [open, setOpen] = useState(null); const clientMap = Object.fromEntries((window.CLIENTS || []).map(c => [c.id, c])); return (
{PRODUCTS.map((p, i) => { const Art = SolutionArt[p.artKey] || ServiceArt[p.artKey]; const client = clientMap[p.clientId]; return (
setOpen(p)} className="reveal card group relative overflow-hidden cursor-pointer transition-all hover:-translate-y-1" style={{transitionDelay: `${i*40}ms`}} itemScope itemType="https://schema.org/Product" itemProp="itemListElement" role="button" tabIndex="0" onKeyDown={(e) => { if (e.key === 'Enter' || e.key === ' ') { e.preventDefault(); setOpen(p); } }} > {/* Art strip */}
{Art ? : null}
Product · {String(i+1).padStart(2,'0')}
{p.stat}
{/* Body */}
{p.family}

{p.name}

{p.blurb}

{/* Built-for footer */}
{client ? (
{client.mono}
Built for
{client.name}
) : } Case
); })}
{open && ( setOpen(null)} /> )}
); }; window.Products = Products;