viva_engage_posts.txt

============================================================
THE CH(E)AT CODE — VIVA ENGAGE COMPANION POSTS
============================================================

Post these to Viva Engage each week alongside the Outlook
Newsletter edition. Schedule for Monday afternoon or Tuesday
morning (after the Monday AM newsletter drop).

Attach the corresponding PDF to each post.
Tag the builder so they can engage in comments.

============================================================
ISSUE #001 — MARCH 23, 2026
============================================================

🎮 The Cheat Code — Issue #001: Code-First Agent Delivery

Your customer stores hundreds of large PDFs in SharePoint — regulatory filings, compliance docs, vendor contracts. Standard Copilot handles short docs well, but large PDFs need a different approach.

This week's pattern: Code-First Agent Delivery — build a production agent in VS Code, deploy to Azure in one command, and land it in Copilot Chat. No portal clicking. No low-code workarounds.

Built by: @Cristiano Almeida Gonçalves · Validated with Coca-Cola

🔑 "Once you've built this for one customer, the next one takes minutes. The agent is a configurable asset — the only thing that changes between customers is the system prompt."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

▲▲▼▼◄►◄► B A START

💬 Have you landed a code-first agent with a customer yet? What was the use case?

============================================================
ISSUE #002 — MARCH 30, 2026
============================================================

🎮 The Cheat Code — Issue #002: Scoped Multi-Source Search

Your customer deployed Copilot and the results feel shallow or noisy. The problem isn't the model — it's that the search is too broad.

This week's pattern: Scoped Multi-Source Search — teach the agent to analyze the input, generate precise queries, and search only the right sources. Better search starts with better questions.

Built by: @Raghav BN · Validated on live ADO work items

🔑 "The first step in better search isn't improving the search engine. It's improving the query. When the agent analyzes the input before searching, the queries it generates are dramatically more precise than anything a human would type."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

↑↑↓↓←→←→ B A START

💬 Where are your customers hitting noisy search results? What sources are they trying to query across?

============================================================
ISSUE #003 — APRIL 6, 2026
============================================================

🎮 The Cheat Code — Issue #003: Prompt-Chained Triage + Playbook Orchestration

Last issue we fixed noisy search. This week: getting an agent to reason through a multi-step process instead of trying to do everything in a single prompt.

This week's pattern: Prompt Chaining + Playbook Orchestration — break complex workflows into sequential steps where each step narrows the problem space for the next. Then match to the right playbook automatically.

Built by: @Raghav BN · Part 2 of 3 from the Work Item Triage Agent

🔑 "Each step narrows the problem space for the next. The search in step 2 is more precise because step 1 extracted structured fields. The classification in step 4 is more accurate because step 3 found relevant history."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

⇑⇑⇓⇓⇐⇒⇐⇒ B A START

💬 What multi-step workflows are your customers trying to automate that keep breaking when done in a single prompt?

============================================================
ISSUE #004 — APRIL 13, 2026
============================================================

🎮 The Cheat Code — Issue #004: Secure In-Boundary Processing

"We can't use AI — our data can't leave our tenant." Legal, healthcare, financial services, government — you've heard this objection. It kills deals.

This week's pattern: Secure In-Boundary Processing — keep every byte inside the customer's tenant boundary. Azure OpenAI in their subscription, no external calls, full audit trail. Zero data leakage by architecture, not by policy.

Built by: @Raghav BN · Part 3 of 3 from the Work Item Triage Agent

🔑 "This isn't just about security — it's about removing the biggest objection to AI adoption in regulated industries. When you can demonstrate that zero data leaves the tenant, the conversation shifts from 'can we?' to 'where do we start?'"

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

△△▽▽◁▷◁▷ B A START

💬 Which regulated industries are you working with? How are you handling the data boundary conversation today?

============================================================
ISSUE #005 — APRIL 20, 2026
============================================================

🎮 The Cheat Code — Issue #005: Human-in-the-Loop Approval Gates

Your customer loves the AI demo. Then someone asks: "But who checks what the AI produces before it goes out?" And the conversation stalls.

This week's pattern: Human-in-the-Loop Approval Gates — the AI does the work in seconds, a human confirms it in one click, nothing goes live without that confirmation. Speed AND trust.

From: Pete Puustinen's Eggis Pharma engagement (Budapest Hackathon)

🔑 "The AI does the work in seconds. A human confirms it in one click. Nothing goes live without that confirmation. That's how you get speed and trust."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

^^vv<><> B A START

💬 How are your customers thinking about AI governance and approval workflows? What's the biggest trust gap you're seeing?

============================================================
ISSUE #006 — APRIL 27, 2026
============================================================

🎮 The Cheat Code — Issue #006: Meeting-to-Knowledge Pipeline

A pharma customer had 20 project managers running multiple meetings per day — every single one required manual meeting minutes. Two agents. 2.5 days to build. Full pipeline from transcript to searchable knowledge base.

This week's agent: Meeting-to-Knowledge Pipeline — Scribe Agent captures and structures meeting output, Library Agent indexes it into a searchable knowledge base. Both use the approval gate pattern from last issue.

Built by: @Pete Puustinen · Validated with Eggis Pharma (Budapest Hackathon)

🔑 "What if every meeting your team runs this week was automatically documented, approved by the project owner, and searchable by leadership by end of day — and we could build that together in two days?"

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

▴▴▾▾◂▸◂▸ B A START

💬 Meeting minutes are a universal pain point. Where else have you seen this pattern land — or where could it?

============================================================
ISSUE #007 — MARCH 31, 2026
============================================================

🎮 The Cheat Code — Issue #007: Holographic Memory

"We have AI tools everywhere now, but none of them remember what the others did." Your customer's knowledge is scattered across chats, repos, docs, and meetings — and no single agent can connect them.

This week's pattern: Holographic Memory — a knowledge architecture where every node carries enough redundant context to reconstruct meaning even when other nodes are missing. Think of it like a hologram: cut it in half and you get a lower-fidelity version of the entire picture, not half of it.

Concept by: @Tyson Dowd · Discussed in 1:1 (March 24, 2026)

🔑 "What if your team could ask one question and get an answer that connects decisions across every project — even when those projects were never explicitly linked?"

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

⬆⬆⬇⬇⬅➡⬅➡ B A START

💬 Where are your customers hitting the "fragmented AI memory" wall? What's the most common scenario where knowledge lives in too many places?

============================================================
ISSUE #008 — MAY 19, 2026
============================================================

🎮 The Cheat Code — Issue #008: Adaptive Guardrails 🧠

Your customer deployed an agent, and within a week security wanted it shut down. The guardrails were too rigid — blocking legitimate queries — or too loose — letting sensitive data through. Static rules don't scale.

This week's pattern: Adaptive Guardrails — a three-layer system (input filtering, output validation, contextual gates) that adjusts safety boundaries based on query context, not hardcoded blocklists.

🔑 "The best guardrail is invisible when it's not needed and unmissable when it is."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

⏶⏶⏷⏷⏴⏵⏴⏵ B A START

💬 How are your customers handling agent safety today? Static rules, prompt engineering, or something else? What's breaking?

============================================================
ISSUE #009 — MAY 26, 2026
============================================================

🎮 The Cheat Code — Issue #009: Building Adaptive Guardrails in Copilot Studio 🔧

Last week we named the pattern. This week we build it — AI Builder classifier + conditional topics + Power Automate escalation, all in Copilot Studio.

This week's build: A three-tier safety layer that classifies every query as general, sensitive, or restricted — and routes each one appropriately. No custom code required.

🔑 "What if your agent could tell the difference between 'What's the PTO policy?' and 'What's John's salary?' — and handle each one appropriately, automatically?"

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

◀◀▶▶▲▼▲▼ B A START

💬 Has anyone built guardrails in Copilot Studio yet? What approach did you use — topic routing, prompt engineering, or external classifiers?

============================================================
ISSUE #010 — JUNE 2, 2026
============================================================

🎮 The Cheat Code — Issue #010: Multi-Agent Handoff 🧠

Your customer has an HR agent and an IT agent. An employee asks the HR agent about laptop provisioning for a new hire. The agent says "I can't help with that." The employee gives up. Sound familiar?

This week's pattern: Multi-Agent Handoff — three patterns (cold, warm, hot transfer), trust boundaries, and shared context design for when one agent needs to delegate to another.

🔑 "The moment your customer has two agents, they have the handoff problem. Solve it and you own the expansion conversation."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

⮝⮝⮟⮟⮜⮞⮜⮞ B A START

💬 How many agents does your biggest customer deployment have? Have they hit the handoff problem yet?

============================================================
ISSUE #011 — JUNE 9, 2026
============================================================

🎮 The Cheat Code — Issue #011: Building Multi-Agent Handoff in Copilot Studio 🔧

Last week we named the handoff patterns. This week we build a two-agent HR/IT handoff with Dataverse session state and conversation summary passing.

This week's build: Out-of-scope detection + agent transfer + shared session context in Dataverse — so the receiving agent picks up where the first one left off.

🔑 "What if your employees could ask any agent any question and always get routed to the right one — without restarting the conversation?"

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

↟↟↡↡↞↠↞↠ B A START

💬 What's your experience with multi-agent architectures? Any war stories on context passing or misrouting?

============================================================
ISSUE #012 — JUNE 16, 2026
============================================================

🎮 The Cheat Code — Issue #012: Persistent Agent Memory 🧠

Every conversation starts from zero. The agent helped configure Project Falcon last Tuesday — but today it has no memory of that interaction. Users repeat themselves constantly. This is the gap between "chatbot" and "assistant."

This week's pattern: Persistent Agent Memory — three memory tiers (conversation, session, long-term), the summarization trap, and structured memory types that actually scale.

🔑 "Memory is the feature that makes users say 'I can't go back.' Land this and the agent becomes sticky."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

⇧⇧⇩⇩⇦⇨⇦⇨ B A START

💬 Are your customers asking for agent memory yet? What's the most common "I wish it remembered..." request you're hearing?

============================================================
ISSUE #013 — JUNE 23, 2026
============================================================

🎮 The Cheat Code — Issue #013: Building Persistent Memory in Copilot Studio 🔧

Last week we named the memory pattern. This week we build it — Dataverse memory table, "Remember" plugin action, "Recall" greeting topic, and memory hygiene flows.

This week's build: An agent that greets returning users with "Welcome back! Last time we were working on Project Falcon — want to pick up where we left off?"

🔑 "What if your agent remembered every user's preferences, project context, and past decisions — so no one ever had to repeat themselves?"

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

⤊⤊⤋⤋⬱⇻⬱⇻ B A START

💬 What would you want an agent to remember about you? What should it forget?

============================================================
ISSUE #014 — JUNE 30, 2026
============================================================

🎮 The Cheat Code — Issue #014: Proactive Agent Triggers 🧠

Every agent today is reactive — it waits for the user to ask. But the most valuable behavior is proactive: "Your compliance cert expires in 14 days." "Three items from last week's meeting are still open." Users don't have to remember to ask.

This week's pattern: Proactive Agent Triggers — scheduled, event-driven, and threshold-based triggers with severity routing and notification management.

🔑 "Proactive agents save 10x the time of reactive agents — because users don't have to remember to ask."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

⏶⏶⏷⏷⏴⏵⏴⏵ B A START

💬 What's the most valuable proactive alert you could imagine an agent sending your customers? What would make them say "I can't turn this off"?

============================================================
ISSUE #015 — JULY 7, 2026
============================================================

🎮 The Cheat Code — Issue #015: Building Proactive Agents with Copilot Studio 🔧

Last week we named the proactive trigger patterns. This week we build a weekly health check that sends itself — Power Automate scheduled flow, Azure OpenAI summarization, Teams Adaptive Cards with action buttons.

This week's build: A project health check that scans SharePoint for overdue items every Monday, generates a natural-language summary, and sends a personalized Adaptive Card to each project owner.

🔑 "What if your team got a personalized project health check every Monday morning — without anyone having to ask for it?"

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

ꜛꜛꜜꜜ‹›‹› B A START

💬 Have you built any proactive flows for customers yet? What triggers worked best — scheduled, event-driven, or threshold?

============================================================
ISSUE #016 — JULY 14, 2026
============================================================

🎮 The Cheat Code — Issue #016: Agent Evaluation & Trust Signals 🧠

"How do I know the agent is actually working?" The pilot shipped three weeks ago. Leadership wants ROI. The agent owner wants quality metrics. Users want confidence signals. Nobody has any of these.

This week's pattern: Agent Evaluation & Trust Signals — three evaluation layers (usage, quality, business), hallucination detection, regression alerts, and trust signals for end users.

🔑 "Without evaluation infrastructure, the 30-day conversation becomes anecdotal. This pattern turns anecdotes into dashboards — and dashboards into expansion funding."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

▵▵▿▿◃▹◃▹ B A START

💬 How are your customers measuring agent success today? Pure usage numbers, or something deeper?

============================================================
ISSUE #017 — JULY 21, 2026
============================================================

🎮 The Cheat Code — Issue #017: Building Agent Analytics in Copilot Studio 🔧

Last week we named the evaluation pattern. This week we build the dashboard — Application Insights telemetry, Power BI visualizations, regression alerts, and user feedback capture.

This week's build: A 3-chart Power BI dashboard (daily users, confidence distribution, satisfaction ratio) with automated regression alerts when quality drops.

🔑 "Right now, can you tell me how many people used your agent yesterday and whether the answers were good? This dashboard answers both in real-time."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

⏫⏫⏬⏬⏪⏩⏪⏩ B A START

💬 What metrics would convince your customer's leadership to expand from pilot to production? What number would they need to see?

============================================================
ISSUE #018 — JULY 28, 2026
============================================================

🎮 The Cheat Code — Issue #018: Custom Connector Patterns 🧠

The agent needs to call an internal API — a legacy ERP, a custom CRM, a proprietary compliance system. There's no pre-built connector. Graph API doesn't cover it. The project stalls at "we can't reach our data."

This week's pattern: Custom Connector Patterns — when to build a connector vs. a plugin action, OpenAPI specifications, authentication patterns, and error handling for resilience.

🔑 "Custom connectors unlock the long tail of enterprise use cases — the 20% where the highest-value, most differentiated scenarios live."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

⇑⇑⇓⇓⇐⇒⇐⇒ B A START

💬 What's the most common "we can't reach our data" blocker you're seeing in customer deployments? Legacy APIs? Auth complexity? Something else?

============================================================
ISSUE #019 — AUGUST 4, 2026
============================================================

🎮 The Cheat Code — Issue #019: Building Custom Connectors for Copilot Studio 🔧

Last week we named the connector patterns. This week we build one — OpenAPI spec, Power Platform connector, Copilot Studio plugin action, and Azure API Management gateway.

This week's build: Wrap any internal REST API in a custom connector and expose it as a Copilot Studio action in one afternoon. Includes APIM gateway for production-grade resilience.

🔑 "Your customer's most valuable data lives behind APIs that Copilot can't reach today. One custom connector changes that — and you can build it in an afternoon."

📎 Full issue attached as PDF
📬 Subscribe to The Cheat Code in Outlook Newsletters for weekly delivery

⥣⥣⥥⥥⥢⥤⥢⥤ B A START

💬 Have you built custom connectors for customers yet? What was the hardest part — the OpenAPI spec, the auth, or something else?

============================================================
NOTES FOR THE POSTER
============================================================

1. ATTACH THE PDF: Each post should have the corresponding
   issue PDF attached. The teaser drives interest; the PDF
   delivers the full experience.

2. TAG THE BUILDER: Use @mentions for the builder named in
   each post. They can answer follow-up questions and share
   additional context.

3. TIMING: Post Monday PM or Tuesday AM, after the Outlook
   Newsletter has been sent Monday morning.

4. COMMUNITY: Post to the AIWF Team community. Consider
   cross-posting to broader CSA or AI Business Solutions
   communities for wider reach.

5. ENGAGE WITH REPLIES: The discussion question at the end
   is the whole point of the Engage cross-post. Reply to
   comments to keep the thread alive.

6. KONAMI CODE: Each post uses the same unique glyphs as
   the corresponding HTML issue. Keep them consistent.