Skip to content

Agent Builder

Configure AI agents to automate content workflows, campaign creation, and advisory conversations with Brand Voice integration.

Overview

The Agent Builder lets you create AI-powered agents that turn simple briefs into ready-to-publish content. Describe the outcome you want — a product launch campaign, a series of LinkedIn posts, or a co-branded one-pager — and the agent handles research, drafting, and channel adaptation, all aligned to your Brand Voice.

Why Use Agent Builder?

Without Agent Builder With Agent Builder
3–5 days per campaign Ready in minutes
Coordinate writer, designer, SEO One unified workflow
Inconsistent brand voice Always on-brand
Manual repurposing across channels Auto-adapted per channel

How It Works

  1. Brief it like a teammate — "Launch our Q1 feature with a blog and social campaign"
  2. It works — pulls from your Brand Voice, researches keywords, drafts content
  3. You ship — review, tweak, publish directly

Agent Types

Choose the execution mode that fits your use case:

Smart Mode (Recommended)

AI determines the optimal execution strategy based on your goal. Best for most use cases — just describe the outcome and let the agent figure out the steps.

Workflow

Multi-step content creation and campaign automation with explicit control. Best when you need a defined sequence of steps, such as research → draft → optimize → publish.

Assistant

Interactive Q&A and advisory conversations. Best for brainstorming sessions, strategy reviews, or when you want to iterate through dialog rather than automated execution.

Creation Wizard

The Agent Builder walks you through four steps:

Step 1: Choose Agent Type

Select Smart Mode, Workflow, or Assistant. Optionally expand Advanced Settings to configure:

Setting Description Default
AI Model Claude 3 Sonnet, Claude 3 Haiku, or Claude 2.1 Claude 3 Sonnet
Max Workflow Steps Maximum steps the agent can execute (1–50) 10
Timeout Execution timeout in seconds (30–3,600) 300
Bedrock Guardrails Enable AWS content filtering guardrails Disabled
Agent Memory Enable session continuity for 40–60% token savings Disabled

Agent Memory

When enabled, Agent Memory stores context across sessions using AWS Bedrock. Choose between Session Summary (condensed recap) or Semantic (concept-level recall), with retention from 1 to 365 days.

Step 2: Agent Details

Configure the agent's identity and context sources:

Basic Fields:

  • Agent Name (required) — A descriptive name like "Product Launch Agent" or "GTM Intelligence Agent"
  • Goal / Outcome — Describe what the agent should achieve. An AI-assisted helper provides example prompts to get you started

Context Sources:

Connect your company profile for consistent, on-brand output.

  • Select a company from your account
  • Optionally select a product for product-specific context
  • The agent pulls tone, audience, guidelines, and keywords automatically

Ground the agent in your company's proprietary data.

  • OpenAI Vector — Local RAG with vector embeddings
  • Bedrock Knowledge Base — AWS-managed knowledge base

Provide source material from outside the platform.

  • URL — Fetch content from any web page
  • External Connections — Link S3, Box, Dropbox, or Google Drive sources

Step 3: Select Capabilities

Choose from 40+ capabilities organized into four categories:

Generation

Create content across formats:

  • Generate Text — Blog posts, articles, documentation
  • Generate Content Series — Multi-piece campaigns from a single brief
  • Generate Image — AI-powered visuals via AWS Bedrock
  • Generate Meta Tags — SEO-optimized titles and descriptions
  • Generate AWS Profile — AWS partner profile content

Publishing & Distribution

Get content to market:

  • Publish to Social Media — Direct posting to social channels
  • Marketplace Listing SEO — Optimize AWS Marketplace listings for search
  • Marketplace Listing Optimizer — Improve listing quality and conversion
  • Sync Marketplace Listings — Keep listings synchronized
  • RSS Post Scoring — Score and prioritize content for distribution

Analysis & Intelligence

Research and optimize:

  • SEO Intelligence — Keyword research and search analysis
  • Content Gap Analysis — Identify missing content opportunities
  • Competitor Analysis — Benchmark against competitors
  • Partner Intelligence — Research potential partners
  • LinkedIn Graph Analysis — Map professional network insights
  • Content Readiness Analyzer — Assess content quality before publishing
  • Predict Performance — Forecast content performance metrics

Workflow & Integration

Connect and enhance:

  • Query Knowledge Base — Pull answers from your KB
  • Enrich Brand Voice — Enhance context from company data
  • Fetch External URL — Import web content into workflows
  • Co-Sell Matching — Find partner co-sell opportunities
  • Joint GTM Planner — Plan go-to-market campaigns with partners
  • ACE Opportunity Sync — Sync with AWS ACE pipeline
  • Funding Application Writer — Draft AWS funding proposals

AI-Powered Suggestions

As you enter your agent name and goal, the builder suggests capabilities that match your intent. Selected capabilities also trigger companion suggestions — capabilities that work best together.

Step 4: Review & Create

Review your full configuration before creating the agent:

  • Agent type, name, and goal
  • Connected context sources (Brand Voice, Knowledge Base, External Content)
  • Selected capabilities
  • Confirm and create

Example Use Cases

Product Launch Campaign

"Launch our Q1 feature with a blog post, social campaign, and hero image"

Recommended capabilities: Generate Text, Generate Content Series, Generate Image, Publish to Social Media, SEO Intelligence

Content Repurposing

"Turn our latest webinar into 10 LinkedIn posts and a summary blog"

Recommended capabilities: Generate Text, Generate Content Series, Publish to Social Media, Content Readiness Analyzer

Partner GTM

"Create a co-branded one-pager and email sequence for our AWS partnership"

Recommended capabilities: Generate Text, Generate Content Series, Enrich Brand Voice, Co-Sell Matching, Joint GTM Planner

Marketplace Optimization

"Audit and improve our AWS Marketplace listing for better search visibility"

Recommended capabilities: Marketplace Listing SEO, Marketplace Listing Optimizer, SEO Intelligence, Generate Meta Tags, Content Gap Analysis

Advanced Configuration

Bedrock Guardrails

Enable AWS content filtering to enforce responsible AI usage:

  1. Check Enable Guardrails in Advanced Settings
  2. Select a guardrail from your AWS account
  3. Choose a version (Draft or Version 1)

Guardrail interventions are tracked per execution for audit purposes.

Agent Memory

Enable session continuity to reduce token usage and improve context:

Memory Type Description Best For
Session Summary Condensed recap of previous sessions General workflows
Semantic Concept-level recall across sessions Complex, ongoing projects

Set retention from 1 to 365 days (default: 30 days).

Best Practices

  • Be specific with goals — "Write a 1,200-word blog post about our new analytics feature for DevOps teams" produces better results than "Write a blog post"
  • Enable Brand Voice — Connecting your company profile ensures consistent tone and messaging across all outputs
  • Start with Smart Mode — Let AI determine the best execution strategy; switch to Workflow when you need explicit step control
  • Use companion suggestions — The capability suggestions are tuned to common workflows; trust them as a starting point
  • Enable Memory for recurring tasks — If your agent runs regularly on similar topics, Agent Memory saves tokens and improves context

Getting Started

  1. Navigate to Content Manager → Agents in your dashboard
  2. Click Create Agent
  3. Choose your agent type (Smart Mode recommended)
  4. Enter a name and describe your goal
  5. Connect Brand Voice and any other context sources
  6. Select capabilities (or accept AI suggestions)
  7. Review and create your agent

Support

  • Email: support@vell.io
  • Documentation: FAQ
  • In-App Chat: Available in dashboard