Partner at Kleiner Perkins
Check size: Not publicly disclosed. Kleiner Perkins invests from early stage through selective growth rounds; isolated portfolio check disclosures should not be treated as Josh's standard range.
Josh backs founders applying AI to consequential workflows in both software and the physical economy. His portfolio shows a consistent preference for products that turn abundant but fragmented information into trustworthy decisions or actions, automate labor-constrained expert work, or combine software with hardware to reverse structural productivity declines. He remains open beyond named sectors when a founder is determined to reinvent an industry.
Start with the industry or workflow being reinvented and quantify why the status quo fails. Show the AI leverage, proprietary data or technical edge, product reliability, auditability and security, customer integration, measured economic outcome, and why the team can move unusually fast. For physical AI, connect autonomy to real field deployment and unit economics; for enterprise agents, show end-to-end ownership rather than a demo layered over a model.
Founders dead-set on reinventing an industry, especially teams pairing customer obsession, technical excellence, speed, and grit with measurable workflow value. His recent writing highlights systems that are production-ready, auditable or evidence-grounded, and capable of turning AI from a clever model into a dependable product.
No personal pass list is public. His published investment cases implicitly screen against thin AI features without workflow ownership, products that cannot reach production reliability, outputs that are not trustworthy or auditable in high-stakes domains, technical novelty disconnected from a large problem, and teams lacking execution velocity or customer focus.
Test whether the system handles the physical, workflow, reliability, security, and economic constraints required for production—not just whether the model can produce an impressive candidate output.
Connect a broken industry baseline to the enabling technical shift, an end-to-end product, measured customer or field outcomes, and a team capable of compounding execution fast enough to define the category.
“I love listening to and learning from any founder who is dead-set on reinventing an industry.”
— https://www.kleinerperkins.com/people/josh-coyne/
“CuspAI designs for what can be built, which is what turns a clever model into a real product.”
— https://www.kleinerperkins.com/perspectives/cuspai-a-search-engine-for-materials-that-dont-exist-yet/
AI materials discovery becomes a real product when it searches enormous design spaces while accounting for physical performance and whether proposed materials can actually be synthesized.
A vertically integrated stack spanning planning software, remote operations, and autonomous heavy equipment can reverse construction's labor shortage and long productivity decline.
Enterprise AI should synthesize fragmented operational data into validated, permissions-aware and auditable decisions rather than merely generate more dashboards.