Vivly Integration Roadmap

Consolidated from 6 parallel deep-research passes (May 2026). Total surface mapped: ~300 unique data sources across 6 categories. This document is the executive view — sequencing recommendations, priority tiers, critical legal/architectural rules. The 6 detailed category reports live alongside this file.

TL;DR — The Build Sequence

Phase 0 (Day 1–30): Free or near-free, low legal risk, high signal. Ship Vivly with ~25 integrations across public knowledge, dev ecosystems, filings, scholarly research, hiring signals, and federated social. Total API spend: $0–$200/mo.

Phase 1 (Month 2–4): Strategic paid layer. Add ~15 paid sources at $30–$500/mo each. Total API spend: ~$1,500/mo. Covers crypto on-chain, finance market data, scraping fallbacks (Apify/Bright Data), person enrichment (PDL, Hunter), tech-stack intelligence.

Phase 2 (Month 4–9): MCP-first distribution. Ship Vivly as an MCP server. Bundle 9 first-party remote MCP servers (GitHub, Slack, Linear, Notion, Atlassian, Google Drive, Postgres, Filesystem, Fetch). This is the most important strategic move — MCP is now the distribution channel for "internal data."

Phase 3 (Month 6–12): Internal SaaS via unified-API layer. Stand up Nango (OSS, self-host) as the OAuth + delta-sync substrate. Add Merge.dev only if customers force you to (their per-account billing is the cost trap).

Phase 4 (Year 2): Enterprise data tier. Quote-priced sources unlock at Enterprise contract size. Crunchbase, ZoomInfo, BuiltWith Team, Sensor Tower, NewsAPI.ai, Feedly Enterprise. Total API spend: $5–25k/mo blended.

Phase 0 — The "Tier Zero" Build Set (free / near-free)

These have zero or trivial cost, low legal risk, high data richness, and durable APIs. Ship every one of them before paying for anything.

Knowledge & Scholarly (the unified-intelligence backbone)

• Wikipedia REST + MediaWiki Action API — entity descriptions, summaries
• Wikidata SPARQL — 110M structured entities, CC0
• OpenAlex — 250M+ scholarly works, CC0; free $1/day credit
• Crossref REST — 180M DOI records, CC0
• Unpaywall — OA fulltext lookup
• Semantic Scholar (S2) — citation graph, embeddings
• arXiv — preprint metadata (S3 bulk mirror for fulltext)
• PubMed E-utilities — biomedical literature
• ROR + DataCite + ORCID — research entity authority
• OpenStreetMap (Overpass + Nominatim) — geo (self-host for prod)
• USPTO Open Data Portal + EPO OPS — patents (EPO free 4 GB/wk)

Public Web & News (with critical legal architecture)

• GDELT 2.0/3.0 — free global event firehose, 15-min cadence
• RSS/Atom universal — table-stakes; 80% of newsletter/blog content
• Cloudflare Radar — internet traffic stats (CC BY-NC — non-commercial only)
• Common Crawl — petabyte-scale web archive (CC0)
• Wayback Machine CDX — historical snapshots (degrading for news in 2026)

Dev Ecosystem (Vivly's GitHub-mafia ICP)

• GH Archive (BigQuery) — every public GitHub event since 2011
• GitHub REST + GraphQL + Webhooks — code/issues/PRs/releases
• deps.dev — Google's package metadata + dep graph + scorecard (BigQuery + REST)
• OSV — vulnerability DB across all ecosystems
• OpenSSF Scorecard — security health checks for 1M+ projects
• Software Heritage — universal archival source code
• ecosyste.ms suite — packages/repos/commits/funding cross-ecosystem
• npm replication feed — every package publish since registry birth
• PyPI BigQuery downloads — richest install telemetry
• Crates.io, RubyGems, Maven Central, NuGet, Go proxy, Packagist — registries
• Homebrew analytics — macOS dev-tool adoption

Forums & Communities

• Hacker News (Algolia + Firebase) — irreplaceable signal-per-byte
• Stack Exchange API + Data Dump — 180+ Q&A sites, CC-BY-SA
• Lobsters (.json endpoints) — small, elite tech community
• Discourse (all instances) — Apple, AWS, OpenAI, HuggingFace forums share one API surface
• Dev.to (Forem) — CC-licensed dev content
• Hashnode — GraphQL, public read
• Lemmy — federated reddit-alikes
• Mastodon — federated social
• Bluesky / AT Protocol Jetstream — lowest stability risk in social
• Telegram MTProto — public channels (via Telethon/Pyrogram)
• Discord (within authorized servers via bot)
• Reddit (free tier 100 QPM + scraper fallback)
• Product Hunt v2 — launch signals

Finance & Filings

• SEC EDGAR API — all US public filings, Form D = real-time private funding (US gov, free)
• sec-api.io — pre-parsed EDGAR ($69/mo, optional)
• FRED (St. Louis Fed) — 800K+ macro/economic time series
• Companies House (UK) — UK corporate filings (Open Government Licence)
• Etherscan V2 multichain key — 50+ EVM chains, free 5/sec
• Alchemy — free 30M CU/month EVM/Solana RPC
• QuickNode — free 10M credits/month, 60+ chains

Hiring Signals (public job board APIs — unauth, free)

• Greenhouse Job Board API — per-customer, fully public
• Lever Postings API — per-customer, fully public
• Ashby Job Postings API — per-customer, fully public
• Workable Public Jobs Feed — per-customer, fully public

Podcasts (free + canonical)

• PodcastIndex.org — 4M+ podcasts, free, Podcasting 2.0 transcripts
• iTunes Search API — canonical podcast ID directory

Startup Discovery

• Y Combinator directory (yc-oss/api) — daily-refreshed JSON

Company Intelligence

• TheirStack — $59–169/mo, job-posting-derived tech stack + hiring intent (only "Tier-A" pick that's strictly paid; included here because it's cheap enough to belong in Phase 0)

Phase 1 — Strategic Paid Layer ($30–$500/mo each)

Add when Phase 0 hits limits, ~Month 2.

Total Phase 1 monthly spend: ~$1,200–$1,800/mo (varies with credit usage). At this level Vivly covers ~80% of the realistic public-web data surface a paying customer expects.

Phase 2 — MCP-First Distribution Strategy

This is the single most important strategic move. MCP transferred to the Agentic AI Foundation in 2025 (Anthropic donation; stewards include GitHub, Microsoft, PulseMCP). It is the de facto distribution channel for "internal data agents."

Ship Vivly as an MCP server

• List on the official registry (registry.modelcontextprotocol.io)
• Streamable HTTP transport (SSE is sunsetting June 30, 2026 at Atlassian; others following)
• Use this as the primary distribution channel — bigger than your own SDK

Bundle 9 first-party remote MCP servers

Federate these so Vivly's "unified query" actually unifies them:

1. GitHub MCP (github/github-mcp-server) — most-installed MCP, baseline
2. Slack MCP (official) — channels/messages/threads, despite 2026 rate-limit cuts
3. Linear MCP (official) — issues/cycles/projects (best-designed PM API)
4. Notion MCP (official) — pages/databases
5. Atlassian MCP (Jira + Confluence) — official remote
6. Google Drive MCP (reference) — file search/read
7. PostgreSQL MCP (reference) — DB introspection
8. Filesystem MCP (reference) — sandboxed file access
9. Fetch MCP (reference) — URL → markdown grounded retrieval

Optional Phase 2 additions (when customers ask)

HubSpot, Salesforce, Stripe, Sentry, Vercel, Supabase, Neon, Cloudflare, Figma, Brave/Tavily/Exa search MCPs, Playwright browser MCP.

Phase 3 — Internal SaaS via Unified-API Layer

For customers whose stack predates MCP, use a unified-API substrate.

Recommended primary: Nango (OSS, code-first, self-host friendly)

• 700+ APIs, OAuth + delta syncs + unified models
• Best unit economics — no per-linked-account billing
• You keep control of sync logic

Alternative / supplement: Merge.dev

• Deepest normalization across HRIS, ATS, Accounting, CRM, Ticketing
• Per-linked-account billing escalates fast — model carefully

Domain-specific unified APIs (use only when needed)

• Finch — unified HRIS/payroll (220+ systems)
• Plaid — banking, with CFPB §1033 in flux (recheck 2026)
• Codat — accounting (Xero/QBO/Sage/NetSuite)
• Apideck — pass-through model, usage-based

Direct connectors when unified APIs underdeliver

• Slack (Marketplace listing required for non-trivial use)
• Microsoft Graph — Teams/Outlook/Drive/SharePoint/Calendar (CASA audit ~$5–20k for restricted scopes)
• Google Workspace — Gmail/Drive/Docs/Sheets/Calendar/Admin SDK (CASA same)
• Salesforce REST + Bulk + Pub/Sub (you consume customer's API allocation)
• HubSpot CRM
• Zendesk, Intercom, Front (support)
• Mixpanel/Amplitude/PostHog (product analytics)
• Zoom Cloud Recording + Transcripts, Otter, Fireflies (meeting intelligence)
• Calendly (booking data)

Phase 4 — Enterprise Data Tier (Quote-Priced, Year 2)

Unlocks when Vivly has Enterprise contract revenue to justify $5–25k/mo blended.

Company / Funding Intel ($15k–$100k+/yr)

• Crunchbase Enterprise — full funding/M&A
• PitchBook — VC/PE gold standard
• CB Insights — market maps + tech trends
• Tracxn — India/APAC strength
• Dealroom — EU strength
• HG Insights — IT spend (unique data)
• BuiltWith Team ($995/mo) — historical tech stack
• Sensor Tower / AppTopia — mobile app intelligence
• G2 Enterprise — B2B reviews + buyer intent
• Trustpilot Enterprise + Connect — consumer reviews

B2B Contact / Person Data ($50k+/yr)

• ZoomInfo — 300M profiles + intent
• Apollo Org tier — $119/user/mo
• Cognism — best EU GDPR posture
• 6sense (Slintel) — intent + technographics
• Coresignal — LinkedIn-derived datasets (residual legal risk post-Proxycurl)

News & Wire (heaviest legal licensing)

• NewsAPI.ai / Event Registry ($60–600+/mo) — enriched news
• Webz.io ($500–5k+/mo) — incl. dark web archive back to 2008
• NewsCatcher ($29–399/mo)
• Feedly Enterprise (~$1,600/mo) — Leo AI enrichment
• Reuters Connect ($25–100k+/yr)
• AP API (quote-only)
• Bloomberg B-PIPE ($50–200k+/yr)
• PR Newswire / BusinessWire / GlobeNewswire (free RSS for headlines; aggregator like Benzinga/RTPR for programmatic)

Web Traffic / SEO

• SimilarWeb API ($90–200k+/yr) — display-only license
• Semrush ($499+/mo) — keyword/SERP
• Ahrefs API (~$949/mo floor) — best backlink graph
• Diffbot Knowledge Graph ($299+/mo) — 10B+ entities

Finance (enterprise terminals)

• Refinitiv (LSEG) ($22–24k/seat/yr)
• FactSet ($12k+/seat/yr)
• S&P Capital IQ Pro (~$15k/seat/yr)
• Morningstar Direct (custom)
• AlphaSense (~$15k/seat/yr) — semantic search across filings + transcripts + broker research

On-Chain Crypto Intel

• Nansen API ($150–3k+/mo)
• Glassnode Business+ (mid-$thousands/mo) — API tier is gated
• Dune Analytics Plus ($349/mo) — SQL over indexed on-chain
• Arkham Intelligence (enterprise beta)

DO NOT INTEGRATE (sources that died, locked down, or carry unacceptable risk)

Critical Legal & Architectural Rules

1. NYT v. OpenAI Architecture

NYT v. OpenAI (allowed to proceed March 2025, focused on regurgitation of memorized training data) reshapes the news/media architecture:

• Free / scraped news → store only titles, URLs, snippets, hashes. Never permanent vector store of full text.
• Licensed APIs (Reuters/AP/Bloomberg/NewsAPI.ai) → full text only in transient cache, not embedded long-term.
• Provenance on every Vivly answer: source + license + retrieval timestamp. This is the architectural answer to regurgitation claims.

2. LinkedIn-Derived Data — Elevated Risk in 2026

Proxycurl shut down July 2026 after LinkedIn sued. hiQ v. LinkedIn protects public-data viewing but not commercial resale, fake-account scraping, or behind-auth data. All LinkedIn aggregators (Coresignal, Bright Data, ContactOut) carry residual risk. If Vivly resells profile-level data, get outside counsel.

3. DPDP / GDPR — Public Availability ≠ Lawful Basis

DPDP (in force, India) and GDPR (EU) both require lawful basis, opt-out, DSAR handling, retention policy for any EU/India personal data. Even if data is publicly scrapeable, the legal stack must be intact.

• Cleanest GDPR-defensible person data: Cognism, FullContact.
• DPDP gotcha for Vivly (Indian entity): limited "publicly available" carve-out only when individuals make it public themselves. Aggregated B2B contact data fails this test in many cases.

4. Diversify Across 2+ Providers Per Category

The Bing API death is the lesson. Stack at least two providers per critical category:

• News: GDELT + NewsAPI.ai + Webz.io
• Search/SERP: DataForSEO + Brave + (Google CSE as compliance fallback)
• Person enrichment: PDL + Hunter
• Scraping: Apify + Bright Data
• Crypto RPC: Alchemy + QuickNode + Etherscan multichain
• Stocks: Polygon + Tiingo + EODHD
• Code intel: deps.dev + ecosyste.ms + Software Heritage

5. Honor robots.txt, ai.txt, llms.txt

Publishers increasingly use per-User-Agent blocks (GPTBot, ClaudeBot, CCBot). Honor them. Vivly's crawler User-Agent should be public, identifiable, and respectful.

6. Stack Exchange CC BY-SA (Attribution + ShareAlike)

SO content carries Creative Commons obligations downstream. If Vivly surfaces SO content to users, attribute it.

Cross-Cutting 2024–2026 Changes to Track

Source Counts by Category

The "Negentropy" View

If you only ship 10 integrations, ship these. Each is high-leverage, durable, and compounds with the others:

1. GH Archive (BigQuery) — every public GitHub event since 2011 (free)
2. SEC EDGAR — all US filings + Form D = real-time private funding (free)
3. GDELT 2.0/3.0 — global news/event firehose (free)
4. OpenAlex — 250M scholarly works, CC0 (free)
5. Wikidata SPARQL — structured knowledge graph (free, CC0)
6. PodcastIndex.org — 4M podcasts, complete graph (free)
7. Hacker News + Stack Exchange — irreplaceable dev signal (free)
8. Greenhouse + Lever + Ashby + Workable — hiring signals (free, public, unauth)
9. Etherscan multichain + Alchemy + QuickNode — on-chain (free tiers cover most use)
10. MCP server federation — GitHub + Slack + Linear + Notion + Atlassian + Google Drive (this is the distribution moat, not the data)

Total cost: $0/mo. Total signal: 60% of what Vivly's "unified intelligence" claim actually requires. The next 40% is paid (Phase 1) and enterprise (Phase 4).