Jonathon's AI Wiki

Claude Partner Network — Eligibility, Tiers, and What Anthropic Actually Asks For

9 min read

Source: ai-research/anthropic-claude-partner-network-announcement.md, ai-research/claude-partners-landing-page.md, ai-research/powered-by-claude-directory.md, ai-research/claudelab-partner-network-guide.md, ai-research/bswen-partner-network-requirements.md, ai-research/development-partner-program-distinction.md

Anthropic launched the Claude Partner Network in March 2026 with $100M of initial funding. The program is free to join, open to “any organization bringing Claude to market,” and pitched at consulting firms, services partners, and technology builders helping enterprises adopt Claude. The most common operator question — how many production sites/apps must I show Anthropic? — has no published quantitative answer. Multiple secondary sources converge on “have at least one live Claude-powered thing you can demo” and reviewers focus on production work, not prototypes.

Key Takeaways

  • No published minimum number of public-facing production sites. Anthropic’s own announcement and claude.com/partners do not list a quantitative threshold.
  • Strong soft requirement: at least one demonstrable Claude-powered artifact. Third-party guides converge on “live prototype, GitHub repo with real API usage, or documented client work” as the bar that meaningfully improves approval odds.
  • Membership is free. No fee, applications open via partnerportal.anthropic.com/s/partner-registration.
  • Anthropic Academy completion is a hard prerequisite. All applicant team members must complete 4 modules (Agent Skills, Claude API, MCP, Claude Code) at ~2-4 hours each ^[ambiguous] — only the BSWEN guide states this and it isn’t on Anthropic’s own page.
  • Application processing: a few business days to ~2 weeks.
  • Claude Certified Architect — Foundations is bundled. Network members get priority access to current and forthcoming certifications. See CCA-F.
  • NOT a reseller or affiliate program. Anthropic explicitly excludes referral-only models. Partners are expected to “build and deliver actual AI outcomes for clients.”
  • The Powered by Claude directory is a separate curated showcase, not a quantitative submission process.
  • The Development Partner Program is a different thing entirely — it’s a Claude Code data-sharing opt-in, not consulting/services membership.

How Many Sites Do You Need to Share?

Direct answer to the original question: Anthropic doesn’t publish a hard number.

What’s actually documented:

SignalSourceConfidence
”Concrete evidence of existing Claude work significantly improves your approval odds”Claude Lab guideHigh (verbatim)
Suggested evidence: “live prototype, GitHub repository with real API usage, or documented client work”Claude Lab guideHigh (verbatim)
3-person agency was accepted (no enforced floor on company size)BSWENMedium
3 client references minimum for Consulting Partner tier; 1-2 page case study per “qualifying deployment”Aggregated search summary, not verifiable in any single fetched sourceLow
Small agencies accepted with a single live demo (Teebo Studio cited example)Aggregated search summary, not verifiable in any single fetched sourceLow

Practical floor for a small agency: one live, publicly accessible Claude-powered application that you built (not a prototype, not a demo video — a working URL or shipping product), plus completed Anthropic Academy modules for all team members on the application.

Pragmatic floor for a consulting firm targeting enterprise work: at least 3 named client deployments with case studies, plus that single demonstrable showcase artifact. This combination clears every documented bar across all sources.

What If You Have No Portfolio Yet?

The honest answer: applying with nothing reduces your odds, but Anthropic does not hard-gate at the entry tier. Their stated criterion is “any organization bringing Claude to market.” A 3-person agency was accepted per the BSWEN guide; small shops have reportedly been accepted with single demos.

If you have literally zero:

  1. Recategorize what you already have. Most agencies underestimate their Claude footprint. Inventory: chat widgets, AI-generated content workflows, internal copilots, automated client deliverables, scheduling/triage bots, support automations. Anything that calls the Claude API or uses Claude.ai meaningfully counts.
  2. Use internal tooling as evidence. A self-maintained LLM wiki, an internal Claude-Code-built operations system, a CLAUDE.md-driven team workflow, an onboarding course site — Anthropic recognizes operational use, not just shipped products. Document it publicly via a README, a blog post, a deployed site, or a public GitHub repo.
  3. Build one tiny artifact before applying. Even a 2-3 week project clears the bar. Floor: one publicly accessible Claude-powered tool (custom chatbot, copy generator, research assistant, vertical-specific agent — anything with a URL and visible Claude API usage).
  4. Apply anyway, lead with intent + capability. The application asks for plans alongside current state. A clear “we will deploy Claude for [specific use cases] over [timeline]” combined with execution evidence (Anthropic Academy in progress, CCA-F booked, prior non-Claude AI work, named clients in delivery) often clears the entry tier even with thin Claude-specific portfolio.
  5. Build during the process. Application processing is a few business days to ~2 weeks. Use that window to ship the missing artifact. It does not need to be polished — it needs to exist.

What does NOT work: pure aspirational language with no execution evidence (“we plan to use Claude for our clients” with no Claude work, no Academy completion, no deployment plan timeline). That gets rejected or stalled.

WEO Marketly–specific framing trap. Thinking the partner application requires external client deployments. Internal operational use of Claude — this wiki, the onboarding course, the intermediate course, internal tooling like Clawdbot / Blog-Agent-Worker / OmniPresence — is itself the demonstrable Claude work. The dental-practice-website day-job likely contains AI-powered surfaces (chat widgets, content generation, scheduling bots) that count even if they are not the headline product.

Program Structure

Three partner tracks (per third-party sources, naming varies)

The exact tier nomenclature differs across third-party guides — Anthropic’s own announcement does not enumerate tiers publicly:

  • Consulting Partners — services delivery, change management, enterprise rollouts (Accenture, Deloitte, PwC, KPMG, Cognizant)
  • Technology Partners — building products with Claude as the engine; explicitly includes independent developers
  • Services Partners — implementation specialists, often boutique firms with deep platform/industry expertise

A separate naming convention from Claude Lab describes three certification tracks: Foundations / Technical Specialist / Enterprise Advisor. These appear to be individual certifications rather than organizational tiers. Cross-reference via Anthropic Academy is required to confirm.^[inferred]

Application process (best assembled from multiple sources)

  1. Pre-work: complete Anthropic Academy modules (Agent Skills + Claude API + MCP + Claude Code).
  2. Apply at partnerportal.anthropic.com/s/partner-registration.
  3. Application asks: company background, use case, planned Claude deployment, current AI capabilities, target industries, organization size.
  4. Anthropic review: a few business days to ~2 weeks.
  5. (For larger applicants only) qualification call + capability dossier.
  6. Acceptance.

Benefits

Financial / GTM:

  • Direct funding for training and sales enablement
  • Market development investment for customer deployment success
  • Co-marketing for joint campaigns and events
  • 5x scaling of Anthropic’s partner-facing team — dedicated Applied AI engineers, technical architects, localized GTM support

Resources:

  • Partner Portal access (Anthropic Academy materials, internal sales playbooks, co-marketing assets)
  • Listing in the Services Partner Directory (enterprise buyer discovery)
  • Priority access to new certifications (CCA-F today, additional seller / architect / developer certs throughout 2026)
  • Code Modernization starter kit (legacy migration / tech-debt remediation)

Named Launch Partners and Deployment Scale

PartnerScale signal
DeloitteClaude access rolled out to ~470,000 associates
CognizantClaude access across ~350,000 associates
AccentureTraining 30,000 professionals on Claude
InfosysDedicated Center of Excellence for Claude Code
PwC, KPMG, Slalom, Tribe AI, TuringConfirmed launch partners; specific numbers not published

These deployment scales are illustrative of where the program is targeted — Fortune-500-scale rollouts — not eligibility floors.

Powered by Claude (Separate Surface)

The Powered by Claude directory is a curated public-facing showcase of companies that have built products with Claude (Glide, Granola, others). It is:

  • A marketing showcase, not a quantitative application form
  • Discretionary inclusion — Anthropic curates
  • Likely fed by the broader partner application or via the Claude Marketplace partner waitlist at claude.com/marketplace-partners
  • The Marketplace waitlist explicitly seeks “companies building Claude-powered products designed for the security, scale, and compliance needs of enterprise customers”

Disambiguation: NOT the Development Partner Program

The Development Partner Program (support.claude.com/en/articles/11174108) is a different surface entirely — a Claude Code data-sharing opt-in for organizations willing to contribute training data. Easy-to-confuse name. If someone references “Anthropic’s partner program” they almost always mean the Claude Partner Network.

Try It

For WEO Marketly specifically (small dental marketing agency, multiple Claude-built systems already shipped):

  1. Inventory what’s already public. OmniPresence, Clawdbot, Blog-Agent-Worker, the Karpathy LLM wiki at karpathy-llm-wiki.7yd9nx2nds.workers.dev, the Claude onboarding course site — pick 3 with the strongest “this is in production” framing for the application.
  2. Get team Anthropic Academy completions on file. Block out 8-16 hours per applicant team member to complete Agent Skills + Claude API + MCP + Claude Code modules.
  3. Check who has CCA-F credentials. Network members get priority on future certifications, but having at least one passed CCA-F at application time demonstrably moves the needle.
  4. Apply via partnerportal.anthropic.com/s/partner-registration. Submit with the live wiki URL plus 2-3 client deployment writeups.
  5. Plan for the qualification call. Practice the pitch: “We help dental practices adopt AI through [products]. We’ve shipped X with Claude in production. Here’s how we’re investing in Claude expertise.”
  6. Watch the Powered by Claude directory. If accepted into the network, ask explicitly about inclusion — it’s a separate ask.

Open Questions

  • Exact quantitative floor for “production deployments” — neither Anthropic’s own pages nor the canonical third-party guides state one. Apply and find out, or contact Anthropic directly.
  • Tier nomenclature reconciliation — third-party sources use Authorized / Solution / Strategic; Claude Lab uses Foundations / Technical Specialist / Enterprise Advisor; Anthropic’s own announcement uses Consulting / Technology / Services. Three vocabularies for what may be the same or overlapping categorizations. Likely the certification names (Foundations / Technical Specialist / Enterprise Advisor) describe individual credentials while the partner-type names (Consulting / Technology / Services) describe organizational classification, but this needs Academy login to confirm.
  • Powered by Claude directory inclusion criteria — how does an accepted Partner Network member get added to the public showcase? Separate process or automatic at acceptance?
  • Fee structure for premium tiers — membership is free at base level; whether higher tiers carry fees or revenue-share is undocumented.
  • Geographic eligibility — international applicants face any region-specific gating? Localized GTM support is mentioned but eligibility isn’t.
  • Reapplication timeline if rejected — no published cooldown.

Graph View

Backlinks