initializing system...

github.com/geopard-studio/nightshift · 672 tests passing

The AI firm
that never sleeps.

Give INSOMNIQ a problem. It assembles a team of autonomous AI agents — Claude Sonnet, Haiku, Gemini, or local Ollama — designs the pipeline, executes in parallel, and self-improves with every run. Not a wrapper. An AI firm.

6 autonomous AI agents ready · Type a problem, watch the team form

SYSTEM ONLINE AGENTS READY KB LOADED
Scroll

How it works

One problem. One autonomous AI team.

Describe the problem. INSOMNIQ recruits the right autonomous AI agents (LLM-powered), designs a custom pipeline, runs it in parallel, scores the output, and self-iterates — until quality is met.

📋

01

Problem In

YAML or natural language. Set budget, acceptance criteria, constraints.

🧩

02

AI Strategy

Coordinator reads KB for proven patterns, designs the team, assigns LLM models per node.

03

Parallel Execute

Agents run concurrently. Auditor monitors. Predictor flags risks in advance.

⚖️

04

Evaluate

Evaluator scores output 0–100%. If can_improve → Coordinator replans and retries.

📚

05

Learn

Librarian writes insights to KB. Run 50 is fundamentally smarter than run 1.

The AI Agent Team

Six autonomous AI agents.
One unstoppable team.

Each agent is an LLM-powered autonomous worker — Claude Sonnet, Claude Haiku, Gemini Flash, or local Ollama. Agent Resources (UCB1) learns which model excels at each task type and improves with every run.

🧩

Planner · Orchestrator

Coordinator

Plans the entire pipeline. Reads KB for proven strategies, designs agent teams, assigns LLM models per node, handles replanning when something fails. The brain of every run.

Active — planning next run Claude Sonnet
👁️

Observer · Watchdog

Auditor

Watches every node run. Manages the inbox — users and Investor send signals here. Delivers consolidated summaries to Coordinator at each replan. Never misses an anomaly.

Monitoring — 0 anomalies Claude Haiku
📈

Risk-Taker · Explorer

Investor

Breaks safe-play loops. When quality stagnates, calls a board meeting with concrete goals. Reads quality signals and forces the system to take calculated risks or reboot entirely.

Idle — watching metrics Claude Haiku
⚖️

Judge · Quality Gate

Evaluator

Scores output quality 0–100%. Spawns parallel sub-evaluators for complex output. Decides if can_improve — triggering the next iteration cycle. Your quality bar, automated.

Ready — awaiting output Claude Sonnet
📚

Memory · KB Keeper

Librarian

Consolidates the Knowledge Base after every run. Merges duplicates, drops noise, surfaces actionable insights for the next Coordinator. LanceDB + ModernBERT, 3ms semantic search.

Processing — last run KB Claude Haiku
🔮

Forecaster · Risk Scout

Predictor

Before each node runs, queries KB for past failures on similar tasks. Flags risks in advance — so the pipeline doesn't repeat history. Future failure prevention, automated.

Scanning — risk model ready Claude Sonnet

Meta moment

This page was built
by INSOMNIQ itself.

Right now — attempt 19. Pipeline: web_spark → html_architect → qa_polish. Budget: $22.52 of $80.00 spent. Not a demo. Actual autonomous execution, live.

Node 1 · Researcher

web_spark

✓ DONE

Node 2 · Frontend Engineer

html_architect

✓ DONE

Node 3 · Reviewer

qa_polish

RUNNING

19

ATTEMPT

$23

SPENT OF $80

3

PIPELINE NODES

Live session

Watch the AI team solve a real problem

A real INSOMNIQ pipeline session. Parallel LLM execution — Claude Sonnet for strategy & code, Claude Haiku for speed & memory, Ollama for local runs. Autonomous AI agents, unfiltered.

nightshift · SWOT analysis for EV startup Poland expand

Why INSOMNIQ

Every other AI tool
forgets everything.

Devin costs $9/hour and resets after each session. Cursor doesn't remember last week. INSOMNIQ gets fundamentally better with every run — because it has a persistent, searchable memory of what worked.

Feature
Other Tools
INSOMNIQ
Learns between runs
Stateless
Knowledge Base
Strategy from experience
No
AR + UCB1
Non-code tasks
Code only
Any task
Smart retry on failure
Full restart
Blame-driven
User intervenes mid-run
Not possible
Auditor inbox
Quality self-evaluation
Basic
Sub-evaluators
Community learning
No
Coming soon

672

Tests Passing

17.5K

Lines of Code

3ms

KB Search

9

Core Pillars

Community & Roadmap

This grows with
every run, every user.

INSOMNIQ isn't just another AI wrapper. The Investor Center, shared Knowledge Base, and public patterns make the system smarter for everyone — while your local data stays completely private.

Coming Soon

Investor Center

Cross-project knowledge aggregation. Your Investor learns from every project in your portfolio. Public Investor Center lets the community share collective AI team patterns and strategies.

Coming Soon

Public Knowledge Base

Opt-in shared KB. Strategies, agent patterns, domain knowledge — contributed across the community. Your local KB stays private. Public KB is read-only unless you choose to contribute.

Coming Soon

Server Mode

Run INSOMNIQ as a service. Submit problems via API, get autonomous AI agent results asynchronously. Deploy on your own infrastructure or use the hosted version.

Your first AI agent team
ships tonight.

Start on GitHub →

AGPL-3.0 · Python 3.11+ · Claude · Gemini · Ollama · uv sync and go