Mini-Agents
Focused assistants for sales ops, lead triage, content drafting, customer support, and more.
Ship a custom agent or micro-app in days, not months.
We design task-specific AI agents and tiny web apps, then plug them into your workflows with a built-in CMS and dashboards.
What you get
Mini-Apps
Lightweight web apps with auth, simple dashboards, and CMS pages for your team or customers.
Ship Fast
We move in small scopes and iterative sprints—quick wins first, polish next.
How I build web experiences
Start small, ship fast
We identify the smallest valuable slice of your product and launch it quickly. Expect tight feedback loops and visible progress every week.
Composable architecture
Modern TypeScript stacks, component-driven UIs, and infrastructure you can actually maintain. Think clean APIs, schema-first data, and batteries-included auth & billing—without the bloat.
Agent-first mindset
We design features around what agents can do for people: qualify leads, summarize docs, draft content, or orchestrate workflows. Humans stay in control with clear approvals.
complext Web Service — coming soon
A unified dashboard to spin up mini‑agents and mini‑apps with a built‑in CMS. One place to configure tools, routes, prompts, and deployments.
Phase 1 · Private Beta
- Login & orgs
- Agent presets + chat
- CMS pages & blocks
Phase 2 · App Builder
- Routes, forms, and webhooks
- Deploy to Vercel/CF Pages
- Per‑app API keys
Phase 3 · Integrations
- Sheets, Calendar, Gmail
- Databases & vector stores
- Observability & analytics
R&D: Containerized Data Processors for Agents
We’re exploring containerized data processors that give agents safe, local sandboxes to transform data, run tools, and generate artifacts—without leaking secrets.
Isolated sandboxes
Each task spins up a minimal, policy‑guarded container with only the permissions and datasets it needs. Results are signed and traceable.
Deterministic pipelines
Processors are versioned like code. Inputs/outputs are typed, cached, and reproducible, so you can trust what the agent did—and do it again.
Local‑first privacy
Run sensitive workloads on your own machines when needed. The same abstractions work locally or on our managed runners.
Compute roadmap: scaling with NVIDIA Blackwell
We’re actively expanding our GPU fleet—with an eye on NVIDIA Blackwell—to unlock heavier models, longer contexts, and faster iteration on research workloads.
Why it matters
More compute lets us experiment with larger multi‑agent systems, richer retrieval, and simulation‑heavy pipelines normally hidden behind enterprise walls.
Access for builders
Our goal is to surface these capabilities in approachable tools so startups and solo devs can build with the same firepower.
Developer Docs (preview)
We’re documenting everything as we build—APIs, agents, blocks, and deployment. Here are tiny previews:
Create an Agent (REST)
POST /api/agents
{
"name": "Lead Intake",
"orgId": "org_123",
"config": {
"model": "gpt-4o",
"tools": ["calendar.create", "sheets.append"],
"system": "Qualify leads and book meetings."
}
}
CMS Page Block
{
"type": "hero",
"props": {
"title": "Welcome to complext",
"subtitle": "Mini‑agents & mini‑apps in days",
"cta": { "label": "Contact", "href": "/contact" }
}
}
Server SDK (TypeScript)
import { agents } from "@complext/sdk";
const agent = await agents.create({
name: "Support",
orgId: org.id,
config: { model: "gpt-4o", tools: ["search", "email.send"] }
});
FAQ
Can you build custom features for my team?
Yes. We start with a small scope (agent or mini‑app) and expand as we learn. The placeholder site you’re viewing is the public face while we onboard early partners.
Will I own my data?
Yes. You choose where it lives. We favor local‑first and bring your own cloud, with clear data boundaries per app and per organization.
How soon can we start?
Reach out on the contact page with your use‑case. We’ll scope a first milestone and begin within days.