AgentBase Features

AgentBase is infrastructure: composable layers that transform agents from demos into dependable systems. The goal is simple — a one-person economy operator can ship work, retain context, and monetize outcomes with the same reliability that SaaS teams expect from cloud platforms.

1) Agent Runtime

The runtime is the execution engine: it turns a goal into tasks, runs tools, verifies outputs, and writes memory. The key difference between “chatbots” and real agents is the runtime loop — planning, action, and verification as first-class system primitives.

In practice, agents need deterministic guardrails: typed tool calls, safety policies, budget constraints, and retry behavior. AgentBase treats these as runtime configuration rather than application code sprinkled across scripts.

Runtime Loop
Inputs
  - Goal / ticket / conversation
  - Agent identity (PersonID)
  - Tools + policies
  - Memory context

Plan
  - Decompose tasks
  - Choose tools
  - Define success criteria

Act
  - Execute tools
  - Coordinate sub-agents
  - Collect artifacts

Verify
  - Check constraints
  - Validate against criteria
  - Produce final output

Write Memory
  - Summaries + decisions
  - Entities + relations (KG)
  - Embeddings (vector DB)

2) Agent Memory (Vector DB + Knowledge Graph)

Memory is what makes an agent organization compound over time. AgentBase uses two complementary systems:

  • Vector memory for semantic retrieval: “find related context”.
  • Knowledge graph for structured state: “what is true, and how does it connect”.

This structure is especially powerful for learning systems: a knowledge graph can encode prerequisites and mastery signals, which is why products like LearningNav and DeepLearnPath fit naturally into the AgentBase memory model.

Memory Architecture
AgentBase Memory
  ├─ User Memory
  │    - preferences
  │    - tone + constraints
  │    - recurring facts
  ├─ Vector DB
  │    - embeddings for notes, docs, calls
  │    - similarity search for RAG
  └─ Knowledge Graph
       - entities (people, tools, clients, services)
       - relations (works_with, booked_via, paid_by)
       - events (calls, deliveries, invoices)

Retrieval Strategy
  - vector search for context
  - KG queries for truth + state
  - merge + rank for the runtime

3) Agent Tools (Integrations)

Tools are the difference between “advice” and “work done.” AgentBase integrates high-leverage systems so agents can execute actions and close loops end-to-end:

  • Bookora for booking infrastructure: share links, embed widgets, and let clients schedule without back-and-forth.
  • Skillshop for skill monetization: package agent work into repeatable deliverables.
  • Paylair for payment rails: invoices, subscriptions, and automated settlement.

Your agent runtime should be able to call tools like APIs: typed inputs, predictable outputs, and audit logs that are easy to reason about.

Tool Orchestration
Tool Registry
  - booking.createLink  (Bookora)
  - skill.publish       (Skillshop)
  - payment.charge      (Paylair)
  - identity.resolve    (PersonID)
  - trust.verify        (SkillCredit)

Execution Guarantees
  - input validation
  - retries + idempotency keys
  - audit log + artifacts
  - policy checks (budget, scope, safety)

4) Agent Identity (PersonID + SkillCredit)

Agents need accountability. AgentBase uses an identity layer to answer: who initiated the action, who is allowed to run it, and what trust signals are associated with the agent’s outputs.

PersonID anchors identity, while SkillCredit provides verifiable trust primitives: credentials, outcome attestations, and reputation that compounds as you ship.

Identity & Trust Model
PersonID
  - person identity
  - agent identity
  - permissions + scopes

SkillCredit
  - verified skills
  - portfolio artifacts
  - reputation events

Result
  - auditable actions
  - enforceable policies
  - transferable trust signals

5) Agent Economy (Marketplace + Autonomous Income)

The one-person economy is about outcomes. AgentBase connects the runtime to monetization rails so outputs can become bookings, subscriptions, or productized services.

When your content drives demand, platforms like CreatorBase can capture the audience; agents can then route qualified leads into Bookora and collect payment through Paylair.

Economy Layer
Demand Sources
  - content + SEO (CreatorBase)
  - referrals + reputation (SkillCredit)

Conversion
  - booking links + widgets (Bookora)
  - productized services (Skillshop)

Settlement
  - invoices, subscriptions, payouts (Paylair)

Outcome
  - autonomous income with accountable agents

Infrastructure that feels like Stripe + AWS for agent teams

AgentBase is designed to be the core platform of Personal Economy Infrastructure — secure identity, durable memory, composable tool integrations, and monetization rails that make one-person companies operate like software.