Shipsy AgentFlow
AgentFlow is the visual canvas beneath AgentFleet. Drag blocks, connect logic, and deploy AI agents and logistics automations tuned to your exact workflow — without writing code.
The Problem
Every network has its own exception flows, escalation paths, and regional rules. Generic agents can't capture the nuance. Customisation either waits on an engineering backlog or never happens.
ChatGPT tabs and Zapier bolt-ons can't execute on a consignment, assign a trip, or trigger a bag dispatch. Reasoning without execution is just another dashboard.
Best-practice playbooks live in Google Docs and ops lead heads. When workflows aren't codified, every new hire relearns, and every corner case gets handled differently.
From generic AI on top → to custom agents inside your operation.
The Canvas
AgentFlow ships with every Shipsy entity and action as a first-class block. Chain AI reasoning, logic, communication, and domain operations on one canvas — then deploy to production with one click.
No code. Drop blocks, wire them together, and configure inputs with references like <agent.output>. Real-time collaboration with live cursors and presence.
Consignment, trip, vehicle, bag, hub, worker, IST — 25+ domain entities with 150+ actions. No wrappers, no REST clients. Just drop the block and configure.
OpenAI, Anthropic, Google, xAI, DeepSeek, HuggingFace, Perplexity. Pick per block based on cost, latency, or capability. Add Vision, Evaluator, Router, and Memory blocks as needed.
Manual runs, REST API, webhooks, cron schedules, or BI-report-driven triggers. When a report fires, a consignment updates, or a webhook lands — your workflow runs.
SMS and WhatsApp via Raven with DLT compliance and template support. No third-party messaging stack to configure — it's wired into the platform.
Condition branches, LLM-powered routers, for-each loops, parallel execution, memory, and semantic knowledge search. Complex workflows without the complexity.
The Layer Beneath AgentFleet
AgentFleet ships with Atlas, Clara, Astra, Nexa, and Vera — the agents most logistics operations need on day one. AgentFlow is the builder layer beneath them. Extend any agent with customer-specific logic, or design entirely new agents for workflows Shipsy hasn't pre-built.
Meet AgentFleet →What's in the box
Every major Shipsy entity is already a block. Every major AI provider is already wired up. You build workflows, not infrastructure.
FAQ
No. AgentFlow is a visual canvas. Ops leads and implementation teams drag blocks, connect them, and deploy. Engineers only step in when a workflow needs custom code via the Function block.
AgentFleet is the pre-built squad — Atlas, Clara, Astra, Nexa, Vera. AgentFlow is the builder layer beneath them. Use it to extend any AgentFleet agent with customer-specific logic, or build entirely new agents for workflows Shipsy hasn't shipped yet.
Anything that combines AI reasoning, logistics operations, and communication. Chain LLM calls with logic. Trigger consignment actions, trip assignments, bag operations. Send SMS or WhatsApp via Raven. Pull data from BI reports. Run on schedule, webhook, API, or manual trigger.
Seven providers out of the box: OpenAI, Anthropic (Claude), Google (Gemini), xAI (Grok), DeepSeek, HuggingFace, and Perplexity. Pick per block based on cost, latency, or capability.
Walk through a live workflow, built on your own logistics data.