Company profile · 3 min read
Lua: an agent platform for mid-market companies
Lua gives companies and developer agencies a governed platform to build, release and operate agents across business channels and tools.
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
Published · Updated
What it does
Lua is an agent platform for mid-market companies and developer agencies. It provides a runtime, tools, skills, channels and integrations for agents that handle sales, support and operations across web, WhatsApp, Slack, Teams, email, SMS and voice. The YC profile describes the shift from SMB agents toward an agentOS; the current docs show TypeScript development, CLI releases, MCP, workflows, channels and governed operation.
The fit is a mid-market company or agency that wants to deploy several agents without rebuilding orchestration, permissions and channel delivery each time. Lua is strongest when the agent needs to do real work inside a business process, not just answer a chat question.
Why I’d look closer
The platform exposes a sensible build-to-operate path: write an agent, test locally, release it with the CLI, connect tools and channels, then monitor and roll back. The YC profile says Lua grew from 300+ SMB sales/support agents into a platform layer for custom skills and deeper integrations. It also describes distribution through developer agencies and channel partners, which is a practical route into mid-market customers.
The founders’ backgrounds fit the operating challenge. The YC biographies describe Lorcan O Cathain with 15 years building and investing in technology companies across emerging markets, and Stefan Kruger as a former VP of Engineering at Paystack with resilient fintech infrastructure experience. The current careers page says Lua serves customers across hospitality, banking, fintech, insurance and other sectors; that is company positioning, not an independent customer roster.
What I’d ask
How are credentials, permissions, approvals and rollbacks handled when an agent acts across several channels? What is included in SMB self-serve versus enterprise delivery, and how are channel partners prevented from creating inconsistent workflows? I’d start with one bounded agent and inspect logs, human handoffs, retries and data residency before expanding.
My editorial take
Lua is a strong fit for teams that want an agent platform they can operate, not a one-off bot. The docs make the build surface tangible. The platform earns trust through governed execution and predictable releases, not the breadth of its integration list alone.
Quick facts
| Field | Sourced detail |
|---|---|
| Buyer fit | Mid-market companies, developers and channel agencies |
| Product | Hosted AgentOS with SDK/CLI, workflows, channels, integrations and governance |
| Distribution | Developer agencies and channel partners; company-described |
| Public pricing | Not exposed in the sources checked |
Sources checked
Checked 2026-09-20.
| Source | Used for |
|---|---|
| YC company profile | Product, founders and evolution |
| Lua homepage | Current agent-workforce positioning |
| Lua docs | SDK, CLI, channels, integrations and governance surface |
Cohort context
Lua Global Inc 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 2026-09-19T16:15:08.696Z 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 | Not observed in this response |
| H1 or H2 heading | Observed |
| Typed structured data | Not observed in this response |
| Docs/developer link | Observed |
| Pricing link | Not observed in this response |
| 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 ↗