An Anthropic partnership practice, built from zero
Built and led an Anthropic partnership practice at a U.S. AI and enterprise software consultancy. Seven concurrent enterprise engagements, a multi-tenant Claude platform for private equity portfolio companies, AI cost observability, value realization tooling, and a certification that has trained 500+ practitioners.
Role
AI Offering Lead, Anthropic partnership
Outcome
25,000+ employees across 10 portfolio companies on one governed Claude platform. $20M+ services pipeline in the first 90 days.
Year
2026
Stack
ClaudeMulti-tenant platformCost observabilityValue realizationCertification program
The mandate
An Anthropic partnership is a promise that a client will be able to run Claude in production:
governed, observable, and at a cost they can forecast. At a U.S. AI and enterprise software
consultancy I built and led that practice from zero: strategy, roadmap, pricing, delivery
model, and the partner relationship itself.
The work ran across seven concurrent enterprise engagements in financial services, software
and energy, including a global financial institution, a private-equity-backed software company,
an energy producer, and a scientific research firm. In each one the executive conversation was
the same four questions: where does AI create value, what will it cost to run, how do we govern
it, and how do we know it is working.
Scope workflow and disposition
→
Platform multi-tenant Claude
→
Build governed workloads
→
Instrument cost · value tracking
→
Certify practitioner program
→
Run client-owned
Fig. 1 · Enablement to run-state
What I built and led
The practice. Strategy, roadmap, pricing, and delivery model for an Anthropic partnership
practice, plus the partner relationship on both sides. The practice reached more than $20M in
services pipeline in its first 90 days. Alongside the engagements I delivered executive
workshops, co-led a Claude webinar series, and contributed to a public financial services agent
repository.
A multi-tenant Claude platform for private equity portfolio companies. Designed and
operated a platform serving 10 portfolio companies and more than 25,000 employees, with
per-company value tracking so sponsors see realized value rather than projections. Multi-tenancy
was the forcing function: once two companies share a platform, isolation, cost attribution, and
ownership stop being optional.
An AI cost observability platform. Created to make enterprise LLM spend legible per
workflow rather than as a monthly invoice, and later commercialized as a licensable product
targeting a 30%+ reduction in enterprise LLM spend. It is the measurement half of
Forecast, Measure, Optimize applied in practice.
Value realization tooling. The instrument executives use to fund AI programs on honest
numbers, so a business case is something finance can audit a year later rather than a slide.
A Claude architect certification. Authored the program, which has trained more than 500
practitioners. The goal of certification, workshops, and handoff is the same: the client’s own
people own the system after we leave.
Decisions worth surfacing
Governance before the first workflow. Accountability, data boundaries, and audit trails are
designed into the platform before a portfolio company ships a workflow on it. The scrutiny on
each new case goes to what is new about that case, and nothing else.
Price in units, not tokens. Cost attribution per workflow changes the budget conversation
from “what is the AI bill” to “what does a resolved ticket cost with AI in the loop.” That is
the number a CFO can forecast.
Enablement that outlives the vendor. Certification and handoff are designed so a client is
not dependent on the consultancy or on a single model vendor. A program that needs its
consultants forever is a program that failed quietly.
Note
This describes program and platform design. Employer and client identities, figures that are
not mine to share, and partner-confidential details are intentionally omitted.