Foundry Features Roadmap

Last Updated: January 2026
Status: Planning & Research

Overview

This roadmap outlines how we can leverage additional Azure AI Foundry features to enhance Engram’s capabilities while maintaining our unique strengths.

Current Implementation Status

✅ Completed

  1. Thread Management - Foundry threads for persistent conversation storage
  2. Elena Migration - Elena created in Foundry with Microsoft Graph tools
  3. Tool Endpoints - HTTP endpoints for Foundry to call Engram tools
  4. Configuration - Key Vault as source of truth

🚧 In Progress

  1. Elena Migration - Creating Elena in Foundry (script ready, needs execution)

Feature Roadmap

Phase 1: Foundry IQ (Enterprise Document Search) 🎯

Priority: High
Timeline: 2-3 weeks
Value: High

What It Is:

  • Foundry IQ powered by Azure AI Search
  • Single knowledge base for multiple data sources
  • Better enterprise document grounding

How We’ll Use It:

  • Hybrid Search: Engram tri-search (episodic memory) + Foundry IQ (enterprise documents)
  • Use Cases:
    • Enterprise document search
    • SharePoint integration
    • Knowledge base queries

Implementation:

# Hybrid search combining Engram and Foundry IQ
async def hybrid_search(query: str):
    # Engram tri-search (episodic memory, conversation history)
    engram_results = await memory_client.search_memory(query)
    
    # Foundry IQ (enterprise documents)
    foundry_results = await foundry_iq_client.search(query)
    
    # Combine using RRF
    return reciprocal_rank_fusion([engram_results, foundry_results])

Benefits:

  • ✅ Broader data access
  • ✅ Better response quality
  • ✅ Managed infrastructure
  • ✅ Automatic indexing

Phase 2: Multi-Agent Orchestration 🎯

Priority: High
Timeline: 3-4 weeks
Value: High

What It Is:

  • Foundry’s built-in multi-agent workflow orchestration
  • Agent coordination and handoffs
  • Shared context management

How We’ll Use It:

  • Complex Workflows: Use Foundry for multi-agent tasks
  • Simple Requests: Keep Engram router for single-agent requests
  • Use Cases:
    • Requirements → Project Planning (Elena → Marcus)
    • Project → Documentation (Marcus → Sage)
    • Comprehensive Analysis (All three agents)

Implementation:

# Foundry workflow for complex multi-agent tasks
foundry_workflow = {
    "agents": ["elena", "marcus", "sage"],
    "handoff_rules": {
        "requirements_complete": "elena → marcus",
        "timeline_created": "marcus → sage",
    },
    "shared_context": True
}

# Engram router for simple requests
if is_complex_task(query):
    return await foundry_workflow.execute(query)
else:
    return await engram_router.route(query)

Benefits:

  • ✅ Built-in agent coordination
  • ✅ Automatic handoff handling
  • ✅ Workflow visualization
  • ✅ Shared context

Phase 3: Foundry Tools Integration 🎯

Priority: Medium
Timeline: 2-3 weeks
Value: Medium

What It Is:

  • Prebuilt Foundry Tools for common capabilities
  • Translation, vision, speech, content understanding

How We’ll Use It:

  • Hybrid Tools: Foundry Tools + Engram custom tools
  • Use Cases:
    • Multi-language support (translation)
    • Image/document analysis (vision)
    • Enhanced voice capabilities (speech)

Implementation:

# Register both Foundry and Engram tools
elena_tools = [
    # Engram tools (Microsoft Graph, memory, BA tools)
    send_email_tool,
    search_memory_tool,
    analyze_requirements_tool,
    # Foundry tools (translation, vision, etc.)
    foundry_translation_tool,
    foundry_vision_analysis_tool,
]

Benefits:

  • ✅ Additional capabilities
  • ✅ Production-ready tools
  • ✅ Less code to maintain

Phase 4: Microsoft 365 Integration 🎯

Priority: High
Timeline: 3-4 weeks
Value: High

What It Is:

  • Native integration with Teams, Outlook, SharePoint
  • Publish agents to Microsoft 365

How We’ll Use It:

  • Teams Bot: Elena available as Teams bot
  • Outlook Add-in: Elena helps with email
  • SharePoint: Direct document access

Implementation:

  1. Publish Elena to Microsoft Teams
  2. Create Outlook add-in
  3. Connect to SharePoint
  4. Test user experience

Benefits:

  • ✅ Native Microsoft 365 integration
  • ✅ Better user experience
  • ✅ Simplified Graph API usage

Phase 5: Agent Catalog & Discovery

Priority: Low
Timeline: 1-2 weeks
Value: Low

What It Is:

  • Foundry’s agent catalog for discovery and management

How We’ll Use It:

  • Register Engram agents in catalog
  • Version control for agents
  • Share agents across projects

Phase 6: Fine-Tuning (Future)

Priority: Low
Timeline: TBD
Value: Medium

What It Is:

  • Custom model training for Engram-specific tasks

How We’ll Use It:

  • Fine-tune for requirements analysis (Elena)
  • Fine-tune for project management (Marcus)
  • Fine-tune for technical documentation (Sage)

Integration Strategy

Use Foundry For:

  • ✅ Thread management (already implementing)
  • ✅ Enterprise document search (Foundry IQ)
  • ✅ Multi-agent orchestration (complex workflows)
  • ✅ Microsoft 365 integration (Teams, Outlook)
  • ✅ Additional tools (translation, vision, etc.)

Keep Engram For:

  • ✅ Tri-search (episodic memory, conversation history)
  • ✅ Custom tools (Microsoft Graph, GitHub, etc.)
  • ✅ LangGraph agent logic (flexibility)
  • ✅ Temporal workflows (durability)
  • ✅ Knowledge Graph (relationships)

Why Hybrid:

  • ✅ Best of both worlds
  • ✅ Leverage Foundry’s strengths
  • ✅ Maintain Engram’s unique capabilities
  • ✅ Gradual migration path

Next Actions

This Week

  1. ✅ Complete Elena migration (run script when dependencies available)
  2. ✅ Store Foundry configuration in Key Vault
  3. ✅ Research Foundry IQ capabilities

Next 2 Weeks

  1. Foundry IQ POC:
    • Create knowledge base
    • Test with enterprise documents
    • Compare with Engram tri-search
  2. Multi-Agent Orchestration Research:
    • Study Foundry workflow capabilities
    • Design multi-agent use cases
    • Plan integration approach

Next Month

  1. Foundry Tools Evaluation:
    • Identify useful tools
    • Test capabilities
    • Plan integration
  2. Microsoft 365 Integration Planning:
    • Research Teams bot creation
    • Plan Outlook add-in
    • Design user experience

Success Metrics

Foundry IQ

  • ✅ Document search accuracy improved
  • ✅ Response quality enhanced
  • ✅ Search latency < 200ms

Multi-Agent Orchestration

  • ✅ Complex workflows executed successfully
  • ✅ Agent handoffs working smoothly
  • ✅ Shared context maintained

Microsoft 365 Integration

  • ✅ Teams bot adoption rate
  • ✅ Outlook add-in usage
  • ✅ User satisfaction scores

Summary

Immediate Focus:

  1. ✅ Complete Elena migration
  2. 🎯 Research Foundry IQ
  3. 🎯 Plan multi-agent orchestration

Short Term (Next Month):

  • Foundry IQ integration
  • Multi-agent orchestration POC
  • Foundry Tools evaluation

Medium Term (Next Quarter):

  • Microsoft 365 integration
  • Agent catalog
  • Fine-tuning evaluation

Strategy: Hybrid approach - use Foundry for infrastructure and orchestration, keep Engram’s unique capabilities.

Last Updated: January 2026