๐ Pipeline Builder
Mikoshi Orchestrator
Chain AI products into intelligent pipelines
Connect Curiosity, Synapse, Turbo, Sentinel, MikoshiLang, and more into automated workflows with built-in governance โ approval gates, multi-agent quorum, policy enforcement, and full audit trails. Each node is a verified AI product โ not a generic API call.
How It Works
Three steps to intelligent automation
01
Design
Drag and drop Mikoshi products as nodes. Connect them into pipelines.
02
Configure
Set parameters for each node. What does Curiosity explore? How many Synapse agents? What does Sentinel check?
03
Run
Execute the pipeline. Watch data flow through each node in real-time. Verified at every step.
Available Nodes
Every Mikoshi product is a pipeline node
Curiosity
Explore state spaces, find anomalies, score for novelty/surprise
Synapse
Multi-agent collaboration, parallel research, consensus
Turbo
Multi-candidate generation + code-execution verification
Sentinel
Action verification, prompt injection defense, safety checks
SafeGuard
Runtime alignment verification, honesty/stability/holonomy
MikoshiLang
Symbolic computation, verified math, unit conversion
AI Provider
Mikoshi AI (free), Claude, OpenAI โ generate/summarize/classify
Data Source
Yahoo Finance, CSV upload, API fetch, database query
Output
Email alert, dashboard update, webhook, notification
Logic
If/else branching, loops, filters, aggregators
Trigger
Schedule (cron), webhook, market event, threshold alert
Governance Nodes
Human-in-the-loop control for AI agents
Every autonomous action can be gated, audited, and governed. No rogue agents.
Approval Gate
Block pipeline until a human approves. Full context provided for informed decisions.
Agent Pool
Route tasks to multiple agents. Require quorum consensus (e.g. 2 of 3 agree) before proceeding.
Policy Check
Enforce spend limits, risk thresholds, allowed tickers, and trading hour restrictions.
Escalation
Auto-escalate to human when confidence is low or risk exceeds threshold.
Audit Log
Immutable record of every decision, vote, approval, and reasoning. Full traceability.
Example Pipelines
Pre-built templates to get started
๐ Daily Alpha Scanner
Trigger (9:00 AM) โ Data Source (500 tickers) โ Curiosity (surprise scan) โ Synapse (3 agents research top 10) โ Turbo (verify claims) โ Sentinel (safety check) โ Output (email report)
๐ฌ Research Assistant
AI Provider (generate hypotheses) โ Curiosity (score novelty) โ MikoshiLang (verify math) โ Synapse (literature review) โ Output (formatted paper draft)
๐ก๏ธ Portfolio Guardian
Trigger (every 15 min) โ Data Source (portfolio positions) โ Curiosity (anomaly detection) โ SafeGuard (risk assessment) โ Logic (if risk > threshold) โ Output (alert + recommendation)
โ Governed Trade Pipeline
Trigger (9:00 AM) โ Data Source (watchlist) โ Synapse (3 agents analyse) โ Policy Check (spend limit, risk, hours) โ Agent Pool (quorum 2/3) โ Escalation (confidence check) โ Approval Gate (human sign-off) โ Execute Trade โ Audit Log
๐ฅ Multi-Agent Research with Governance
Claude (thesis A) + Mikoshi AI (thesis B) + OpenAI (thesis C) โ Agent Pool (weighted vote) โ MikoshiLang (verify claims) โ Policy Check (confidence โฅ 0.7) โ Approval Gate (publish?) โ Output (report) โ Audit Log
Comparison
Not another generic workflow tool
| Mikoshi Orchestrator | Zapier / n8n | LangChain | |
|---|---|---|---|
| AI-native nodes | โ All Mikoshi products | โ Generic APIs | โ ๏ธ LLM calls only |
| Verified computation | โ Turbo + MikoshiLang | โ | โ |
| Safety layer | โ Sentinel + SafeGuard | โ | โ |
| Curiosity scoring | โ Built-in | โ | โ |
| Multi-agent | โ Synapse | โ | โ ๏ธ Basic |
| Governance | โ Approval gates, quorum, policy, audit | โ | โ |
| Human-in-the-loop | โ Escalation + approval | โ ๏ธ Manual pause | โ |
| Audit trail | โ Immutable logging | โ ๏ธ Basic logs | โ |
Orchestrator is a Nexus premium feature
ยฃ24.99/mo
Unlimited pipelines ยท All Mikoshi products ยท Priority execution
Upgrade to Nexus โ