June 8, 2026. The firms that win AI work are rarely the ones with the best slide deck. They are the ones who have already put the technology into production and can prove it. On June 3, Anthropic moved to make that proof visible, adding a Services Track and a Partner Hub to the Claude Partner Network. For agencies and consultancies that build AI systems for a living, this is the clearest path yet to a credential that buyers, and increasingly AI assistants, can actually check.
Anthropic launched the Claude Partner Network in March, backed by a $100 million investment in training, technical support, and shared marketing. Since then more than 40,000 firms have applied and more than 10,000 consultants have earned a Claude certification. The largest professional-services firms are building entire practices on it. Accenture is training 30,000 people, Cognizant has rolled Claude out to roughly 350,000 associates, Deloitte is making it available to 470,000, and KPMG is integrating it across a workforce of more than 276,000. The new Services Track sorts all of that activity into something a customer can compare at a glance.
The reason behind it is a gap every operator who has shipped AI knows well. A successful pilot is not the same as a system a business can run on. The real work sits in the integration, the evaluation, and the way people's jobs change around the tool, and the firms that get that right tend to have done it before. The Services Track is Anthropic's attempt to score that experience on terms a customer can trust.
What Anthropic announced
- A three-tier Services Track. Select is where partnership begins, requiring at least 10 active certified individuals, at least 2 joint customers deployed in production in the trailing 12 months, and at least 1 public customer story. Preferred raises that to 100 certified individuals, 15 deployed customers, and 3 public stories. Global Premier, the top tier, requires 1,000 certified individuals, 100 deployed customers across three or more regions, 15 public stories, and a joint business plan with named executive sponsors.
- One scorecard for everyone. Every firm is measured on the same three things: how many of its people hold a current certification and have used Claude in the past 90 days, how many customers it has taken live in production, and how many of those customers will vouch for it in a published story. Size does not raise the tier. A ten-person shop climbs the same ladder as a global consultancy.
- A public Partner Hub. The Hub is a directory where each partner sees its standing against the requirements, refreshed daily, and where customers searching for Claude expertise find the firms most qualified for their project. Tier, certified team size, deployments, and references are all visible in the public listing.
- A way to query it inside Claude. Partners can connect the Hub to Claude through a new MCP connector, then simply ask Claude where the firm stands against the next tier, the status of a registered deal, or how many consultants hold an active certification, and act on the answer.
- A predictable cadence. Promotions are processed twice a year, on January 1 and July 1, with an extra review on October 1, 2026 in this first year. A firm can only move down at the annual year-end review, only if it no longer meets its tier, and only after 90 days notice and a chance to close the gap.
- A free on-ramp. Getting started costs nothing. New applicants begin at Registered, the entry level, with a commitment to 10 certified practitioners, and real partnership begins at Select. Certifications are earned by individuals through Anthropic Partner Academy exams, and tiered partners get discounted rates on a first attempt.
What it means for operators
For a small or mid-sized AI shop, the headline is that the entry rung is reachable. Ten certified people, two production deployments in a year, and one public case study is a realistic target for a focused team, and it lands you in the same public directory as the global names. That matters because the Hub turns expertise that used to live inside a sales call into a credential a buyer can verify before they ever contact you. It is, in effect, a new inbound channel for AI automation agencies that have done the work but struggle to prove it quickly.
There is an answer-engine angle too. A public directory that is queryable through an MCP connector is exactly the kind of structured, authoritative source that AI assistants lean on when a user asks who can build something. Being listed, with live deployments and named customer stories attached, makes a firm easier for both people and models to recommend. If your team is already shipping production systems, the practical move is to convert that work into certifications and documented case studies rather than leaving it undocumented and invisible.
The path is straightforward. Put your engineers through the certification exams, log the Claude deployments you have already delivered for clients in the trailing twelve months, and publish at least one customer story with real outcomes. If you are still building your bench, this is the moment to hire an AI engineer or bring in a partner to stand up your first production AI automation projects. Teams setting up Claude-powered agents from scratch can shorten that runway with a guided OpenClaw setup, then point the resulting deployments straight at the Select requirements.
One detail worth planning around: the higher tiers reward regional spread, with Global Premier expecting deployments across three or more regions. For a firm operating across the GCC, the US, and the UK, that geographic range is an asset rather than an afterthought, and it is worth structuring customer stories so they show breadth, not just volume. The mistake to avoid is treating certification as a box to tick. Anthropic measures 90-day activity, so a bench that certifies once and stops using Claude quietly drops off the count.
It also pays to be deliberate about what a public customer story contains, because the tier counts it but buyers actually read it. A reference that names the problem, the system you shipped, and a measurable result carries far more weight than a logo and a quote. Treat each deployment as a future case study from day one: capture the baseline before you start, agree with the client on what you will publish, and record the outcome in numbers. That discipline does double duty, satisfying the Services Track requirement while giving you the proof that closes the next deal.
Specializations for specific industries and use cases are coming next, along with rewards that grow with deployments. The direction is clear. Anthropic is building a marketplace where proven implementation, not marketing, decides who gets found. The firms that document their work now will be the ones surfaced when the buying starts.
Frequently Asked Questions
It is a tiered ranking Anthropic added on June 3, 2026 that sorts Claude implementation partners by how deep their practice runs, measured by certified staff, production deployments, and public customer stories. The tiers are Select, Preferred, and Global Premier.
Select, the entry point to partnership, requires at least 10 active certified individuals, at least 2 joint customers deployed in production in the trailing 12 months, and at least 1 public customer story.
Small firms can qualify. Every firm is measured against the same requirements and size does not raise the tier. A ten-person shop reaches Select on the same terms as a global consultancy by growing its certified bench and documenting production work.
It is a portal and public directory, refreshed daily, where partners track their standing against the requirements and customers find the firms most qualified for a project. Partners can also query their status inside Claude through a new MCP connector.
Certifications are held by individuals, not firms, and are earned by passing exams through Anthropic Partner Academy. To count toward a firm's tier, the holder must have used Claude in the past 90 days.
Promotions are processed twice a year, on January 1 and July 1, with an additional review on October 1, 2026 in the first year. Downgrades happen only at the annual year-end review, after 90 days notice.