Portfolio

The work speaks.
Here it is.

We've shipped more than twenty products for clients and for ourselves. These six are our own: voice agents, a conversational CRM, agent orchestration, a native macOS browser, and more. Each card opens the full case study of how it was built.

VarioBrowser

Developer tools

VarioBrowser

Native macOS browser that renders one URL at desktop, tablet, and phone sizes side by side. Real WebKit views with synced scrolling and input, vision-deficiency filters, bug capture, and a CLI built for AI agents.

SwiftSwiftUIWebKitAgent CLI
Read the case studyvariobrowser.com
Ergofon

Voice AI

Ergofon

Intelligent voice agent that transforms customer interactions with natural speech-to-text, text-to-speech, and LLM-powered conversations. Real-time voice processing at scale.

PythonNode.jsLLMSTT/TTS
Read the case studyergofon.com
Founders Kit

Sales software

Founders Kit

Your AI sales co-pilot. A conversational CRM that lets you manage deals, contacts, and pipelines through natural language commands. Just tell Kit what you need.

Next.jsNode.jsAI AssistantLLM
Read the case studyfounders-kit.com
Gremium.ai

AI agents

Gremium.ai

AI-powered workspace where intelligent agent teams execute complex knowledge work. Automate tasks at scale with SOC 2 compliant agent orchestration.

JavaScriptPythonLLMAgent Orchestration
Read the case study
SafestMenu

Restaurant software

SafestMenu

SaaS platform for restaurants to design, manage, and update digital menus with ease. Streamline menu operations with intelligent suggestions and analytics.

Node.jsLLM IntegrationSaaS
Read the case study
TF Convert
Downloadable

Data tools

TF Convert

Offline tool for converting structured data files with intelligent field mapping and anonymization capabilities. Download, run locally, keep your data private.

Node.jsLLMData ProcessingOffline
Read the case study

Tech stack

What these are built with

We pick technology per project, not by habit. This is what's running in production across the portfolio right now.

LLM IntegrationAI AgentsVoice AINext.jsNode.jsPythonTypeScriptSwift / macOSReal-time SystemsSaaS
20+
products shipped
client work and our own ventures
6
in the portfolio
our own products, shown publicly
4-6
weeks per MVP
the typical sprint behind each first version

Yours could be next.

AI product, customer-facing platform, or internal system - tell us what you need and we'll tell you what it takes.

No obligationConfidentialReply within one business day