AccessFox

Healthcare account-readiness intelligence from live web signals.

AccessFox helps enterprise GTM and strategy teams identify which hospitals and health systems are becoming ready to adopt advanced therapies — before that readiness shows up in CRM systems or static market reports.

Live demo: https://accessfox.vercel.app

---

What It Does

Hospitals send clear signals when they're preparing to adopt an advanced therapy: job postings, clinical trial registrations, facility investments, regulatory filings, partnership announcements. These signals are all public — but scattered across dozens of sources.

AccessFox pulls live public web signals via Bright Data's SERP API, processes them through an AI enrichment layer (AIML API / GPT-4o), and returns ranked account intelligence cards:

Output	Description
Account Readiness Score	Signal-weighted 0–100 score per organization
Evidence Ledger	Sourced signals with dates and links to original pages
Why Now	Urgency rationale synthesized for a sales rep
Recommended Action	Specific next commercial step
Outreach Angle	The message framing most likely to land

The demo ships with a radiopharmaceutical therapy / theranostics use case, but the pipeline is query-driven and works for any advanced healthcare market: cell therapy, proton therapy, robotic surgery, hospital-at-home, medical AI, and more.

---

Hackathon

Bright Data AI Agents & Web Data Hackathon

• Track: GTM Intelligence / Enterprise AI
• Live app: https://accessfox.vercel.app
• Submission: https://lablab.ai/ai-hackathons/brightdata-ai-agents-web-data-hackathon/evidence-works/accessfox

---

Bright Data Integration

AccessFox uses Bright Data's SERP API via residential proxy infrastructure to fetch live Google search results in real time — not cached, not static.

lib/brightdata.ts

• Proxy endpoint: brd.superproxy.io:33335 (HTTPS)
• Transport: undici ProxyAgent with proxyTls and requestTls configured to handle Bright Data's TLS interception certificate
• Queries are dynamically built from the user's search term
• Returns up to 20 structured results with URLs and text excerpts
• Live badge visible in UI: ● Live Web Signals via Bright Data
• Gracefully falls back to demo dataset when BRIGHTDATA_PROXY_URL is not set

---

AI / ML Layer

lib/ai.ts

• Provider: AIML API (OpenAI-compatible endpoint)
• Model: GPT-4o
• Input: up to 15 URLs from Bright Data SERP results, filtered to hospital and health system domains
• Processing: 3 parallel batches of 5 URLs each via Promise.allSettled (~15s vs ~55s sequential)
• Output: structured Organization objects with score, evidence, reasoning, and commercial intelligence
• 20-second timeout per batch; failures are isolated — partial results always returned

---

Stack

Layer	Technology
Framework	Next.js 15 (App Router)
Language	TypeScript
Styling	Tailwind CSS
Live web data	Bright Data SERP API
AI enrichment	AIML API (GPT-4o)
HTTP proxy	undici ProxyAgent
Deployment	Vercel

---

Run Locally

Prerequisites: Node.js 18+

git clone https://github.com/dbaikov/bright-data-hack.git
cd bright-data-hack
npm install
npm run dev

Open http://localhost:3000. The app runs in demo mode with no environment variables — 8 fully-populated account cards are always rendered as a fallback.

Enable Live Mode

Create .env.local in the project root:

# Bright Data SERP API — residential proxy credentials
BRIGHTDATA_PROXY_URL=http://brd-customer-<id>-zone-serp_api1:<password>@brd.superproxy.io:33335

# AIML API key — https://aimlapi.com
AIMLAPI_KEY=your_key_here

With both variables set, the app fetches live Google search results via Bright Data and enriches them through GPT-4o. The UI badge switches from "Demo dataset" to ● Live Web Signals via Bright Data.

---

Project Structure

app/
  api/opportunities/route.ts   # POST endpoint — orchestrates Bright Data + AI
  page.tsx                     # Search UI with results grid

components/
  OrgCard.tsx                  # Account card: score, evidence, reasoning
  ScoreRing.tsx                # SVG circular score indicator
  DataSourceBadge.tsx          # Live vs demo mode badge
  SignalChips.tsx              # Signal type tags

lib/
  brightdata.ts                # Bright Data SERP proxy integration
  ai.ts                        # AIML API GPT-4o enrichment
  score.ts                     # Deterministic signal-weight scoring
  fallback.ts                  # 8-org demo dataset (always-available fallback)
  types.ts                     # Shared TypeScript types

submission/
  cover.png / cover.svg        # Cover image
  slides.pdf / slides.html     # 4-slide pitch deck
  accessfox-demo.mp4           # Demo video (78s)
  VIDEO_SCRIPT.md              # Narration script

---

How Scoring Works

When live AI enrichment is not available, lib/score.ts computes a deterministic score from signal types:

Signal	Weight
Clinical trial	25
Facility expansion	20
Equipment investment	18
Hiring	15
Regulatory	12
Partnership	8
Funding	8

Bonuses applied for recency (signals dated within 90 days) and signal diversity (multiple signal types).

---

Demo Flow

1. Open https://accessfox.vercel.app
2. Enter a therapy or market, for example:
   • radiopharmaceutical therapy
   • theranostics adoption readiness
   • cell therapy
   • hospital-at-home
3. Click Find Expansion Opportunities
4. Review ranked account cards with readiness scores, evidence, and commercial intelligence