Skip to main content

Introduction to Ariftly

Agents that act. Not chat.

Ariftly is a Vertical AI Agent Platform — a growing roster of autonomous agents that understand their domain, connect to your existing tools, and take real action. Every agent works within your approval gates — nothing is sent, changed, or posted without you saying so.

There's an agent for every domain. Whether you run a company, close deals, manage compliance, or build software — Ariftly ships agents that are already expert in your area and ready to work the day you deploy them.

What makes Ariftly different?​

Most AI products give you a chat interface. Ariftly gives you agents that initiate work, connect to your tools, and surface decisions for your approval — then execute.

Instead of building thisYou deploy this
Lead discovery workflowSales Agent — finds, enriches, and drafts outreach
AI compliance questionnaire flowAI Readiness Agent — audits and answers in hours
Custom approval logicHuman-in-the-loop built into every agent, out of the box
Tool integrationsPre-built connectors to Gmail, Slack, GitHub, and more
Agent infrastructureRemote Agent Protocol — open spec, any language

Not a workflow builder​

Zapier and Make are workflow builders — you connect nodes and write the logic yourself. Ariftly ships agents that already know what to do. You deploy them and extend their behavior by describing what you want in plain English. We do the implementation via the Skill Builder.

Not agent infrastructure​

Tools like LangChain are infrastructure — you build the agent yourself using their primitives. Ariftly is the finished product: agents with domain knowledge built in, approvals as a first-class primitive, and a Skill system that lets non-engineers extend behavior.

Available Agents​

AgentWhat It DoesStatus
AI Readiness AgentEU AI Act compliance, procurement questionnaire responses, knowledge base audits✅ Live
Sales AgentLead discovery, email enrichment, AI-drafted outreach with approval gates✅ Early Access
Finance AgentInvoice reconciliation, spend anomaly detection🔜 Coming soon
DevOps AgentIncident triage, deployment risk scoring🔜 Coming soon
Founder AgentCompetitive intelligence, board pack generation🔜 Coming soon

How it works​

You connect your tools once
         │
         ▼
┌──────────────────────────────────┐
│           ARIFTLY                │
│       Control Plane              │
│  Credentials · Tasks · Approvals │
│  Skills · Knowledge · Memory     │
└──────────────────────────────────┘
    │              │              │
┌───────┐    ┌──────────┐   ┌──────────┐
│  AI   │    │  Sales   │   │  Custom  │
│ Agent │    │  Agent   │   │  Agent   │
└───────┘    └──────────┘   └──────────┘
    │              │              │
    └──────────────┴──────────────┘
               │
         Your approval
               │
         Action executed
  1. Connect your tools (GitHub, Slack, Gmail, etc.) once via the dashboard
  2. Deploy an agent — it immediately starts working in its domain
  3. The agent surfaces findings and drafts actions
  4. You review and approve — nothing external happens without you
  5. Extend any agent with the Skill Builder — describe new behavior in plain English

Platform Primitives​

Every agent in the platform automatically has access to the same set of primitives:

PrimitiveWhat it is
TaskAn async unit of work dispatched to an agent
ArtifactA typed output from a task — reports, drafts, lead lists
ApprovalA human-in-the-loop gate — any external action waits here
SkillA user-authored behavior extension (describe it → YAML generated)
CredentialAES-256-GCM encrypted secrets — agents see proxy tokens, never raw keys
KnowledgeTenant-scoped document retrieval (pgvector + BM25)
MemoryLong-lived key-value and time-series state across tasks

Security model​

  • BYOK by default — bring your own AI provider keys (OpenAI, Anthropic, etc.)
  • No raw credentials to agents — encrypted vault + short-lived proxy tokens
  • Deny-by-default policy engine — destructive actions require explicit policy approval
  • Append-only event log — every state change is recorded for audit and replay
  • RS256 JWT + HMAC-SHA256 — mutual authentication on every Core ↔ Agent call

What Ariftly is NOT​

  • ❌ A chatbot or assistant
  • ❌ A workflow builder (Zapier / Make / n8n)
  • ❌ Agent infrastructure (LangChain)
  • ❌ A scanner or compliance checklist tool
  • ✅ A platform of vertical agents with domain expertise built in

Next steps​