Company profile · 3 min read
Jarmin lets teams hire role-based AI employees for technical and business work
Jarmin provides role-based AI employees and managed AI/ML engineering that learn company workflows, operate in existing systems and return reviewable work.
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
Published · Updated
Jarmin is for teams that want to assign a bounded role to an AI employee rather than assemble a collection of prompts and tools. The current product spans ML engineering, financial analysis, account executive work, compliance, data analysis, project management and paralegal tasks.
What it does
Jarmin says it learns the role and the company’s standards, works inside the systems and data the role uses, produces reviewable output and earns more responsibility over time. The public site separates role-based AI employees from managed AI and ML engineering, where Jarmin can handle data, models, experiments, evaluations, agents and ongoing improvements. Jarmin homepage
The value is ownership of a workstream. A company can hand over an initiative or individual task, keep the team visible through status updates and reserve judgment calls for people. That is more ambitious than a coding copilot, but it also means the buyer must define a job description, access boundary, review cadence and escalation path as carefully as it would for a human hire.
Jarmin’s pricing page is quote-led. It lists a company-specific AI employee, system connections, a full audit trail, SSO, role-based access control, integration support and a dedicated customer-success manager. Jarmin pricing The public pages do not publish a standard fee, so the comparison is against the cost and time of staffing a role or project—not a guessed seat price.
The founding team’s technical background is central to the pitch. The YC profile describes Chalmers Brown and Zack Leman as former Meta Superintelligence Labs staff engineers, Shirsendu Halder as a former Apple and Lockheed Martin ML researcher, and Leman as a former AWS engineer. Those facts explain the ML-automation wedge; they do not prove performance in a new role.
My editorial take
I would evaluate Jarmin for an initiative with clear inputs, artifacts and acceptance criteria—such as a bounded ML pipeline, recurring analysis or a defined compliance research task. Start with approval at major decision points and inspect the audit trail before widening scope. For ambiguous executive work or a role with high customer or financial authority, a human team should remain accountable even if Jarmin performs much of the execution.
Quick facts
| Field | Sourced detail |
|---|---|
| Buyer | Companies hiring AI capacity for technical and business roles |
| Product | Role-based AI employees plus managed AI/ML engineering |
| Controls | Audit trail, SSO and role-based access are advertised |
| Pricing | Quote-led; no standard public price found |
| Main fit question | Can the role be bounded by reviewable outputs and explicit approvals? |
Sources checked
Jarmin’s YC profile, homepage, pricing page and about page were checked on 2026-09-19.
Cohort context
Jarmin is listed in Fall 2025. In our 2026-09-18 directory snapshot, 90 of 146 listed companies in that cohort have YC’s primary industry label B2B (61.6%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.
Public website snapshot
Observed 2026-09-19T16:15:04.331Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.
| Signal | Homepage observation |
|---|---|
| Product description metadata | Observed |
| Canonical link | Observed |
| H1 or H2 heading | Observed |
| Typed structured data | Observed |
| Docs/developer link | Not observed in this response |
| Pricing link | Observed |
| llms.txt link | Not observed in this response |
| Markdown alternate | Not observed in this response |
Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.
About the author
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.
First1000 ↗ · X ↗