(About)Tel Aviv ↔ San Francisco

We build the tools we wished existed.

Donkit is a small, science-driven team turning AI agents from a weekend demo into production software. We’ve shipped, scaled and sold both B2B and B2C products — and we know which corners not to cut.

We’ve launched B2C products from zero to millions of users, architected systems that moved revenue at household-name retailers, and won competitive programming championships back when JavaScript still had jQuery in it.

HQ · researchTel Aviv
GTM · partnersSan Francisco
Disciplines
Science · R&DEngineeringProductDesignMarketing & growthEnterprise sales
(01)Mission

Make building agents as easy as writing a sentence.

For most of software history, the people who knew what to build were not the people who could build it. Agents change that — if the platform underneath is honest. Our job is to make that platform: typed, observable, deployable, and friendly enough that anyone with a clear idea can ship one.

(02)Advisors & TAB

The people we listen to.

GTM · investing · enterpriseAdvisors

Senior operators from category-defining tech companies — people who’ve scaled developer tools, sold AI into the enterprise, and helped early-stage teams find their footing.

Ex-VP marketingEx-GoogleInvestorsStartup coachesEnterprise AI
AI · ML · infraTechnical advisory board

We’ve assembled a TAB of professionals from across the AI industry — researchers, infrastructure engineers, applied ML leads. They pressure-test our architecture and keep us honest about what works in production.

Foundation modelsRetrieval & memoryAgent runtimesDistributed systemsApplied MLAI safety
(03)Backed by
AWSNVIDIA InceptionMicrosoft for StartupsGoogle for StartupsAlchemist AcceleratorIsrael Innovation AuthorityErasmusTel Aviv University
Alchemist Accelerator portfolio company

Donkit’s founders are Alchemist Accelerator alumni. The team carries a strong engineering and R&D background, deep product expertise, and extensive experience shipping enterprise-grade software — the kind that has to survive a security review, not just a Twitter thread.