Pilot Session Training Agenda
Full curriculum for the 3-hour virtual pilot session — April 8, 2026.
Date
April 8, 2026
Format
Virtual · ~3 hours
Audience
10–15 super users
Objective
Build a workflow agent independently
Audience Archetypes
| Archetype | Background | Kindo Focus |
|---|---|---|
| Full-stack devs | Strong software engineering; may not know AI/cyber deeply | Agent construction, API steps, prompt engineering |
| Cyber professionals | Deep cybersecurity domain knowledge; less AI/ML exposure | Use case mapping, knowledge stores, integration config |
| AI/ML engineers | Some existing Kindo or AI platform exposure | Advanced patterns, multi-agent coordination, Canvas |
Block 1 — Platform Orientation
1
Platform Orientation
Goal: Everyone understands what Kindo is, how it fits their environment, and the core building blocks.
| Time | Topic | Content Source | Delivery |
|---|---|---|---|
| 0:00–0:10 | Welcome & Context | — | Facilitator intro. Why Kindo, what Deloitte is building toward, session objectives. |
| 0:10–0:20 | Platform Overview | Module 01 | What is Kindo, deployment models (SaaS vs self-hosted), key capabilities (Chat, Agents, Integrations, Governance). |
| 0:20–0:30 | Agent Types Deep Dive | Module 03 | Chatbots vs Workflow Agents vs Trigger Agents — when to use each. LLM Steps, Action Steps, API Action Steps. Live walkthrough of Agents tab. |
| 0:30–0:37 | Integrations Overview | Module 04 | How integrations work (Nango + MCP). Deloitte-relevant stack: CrowdStrike, Jira, ServiceNow, Splunk. |
| 0:37–0:45 | AI Chat First Steps | Module 02 | Live demo: ask a question in AI Chat, show model selection, show file upload. Participants follow along. |
Key takeaway: Participants understand the three agent types and can navigate the Kindo UI.
Block 2 — Hands-On Workshop
2
Hands-On Workshop: Build a Workflow
Goal: Guided construction of two real cybersecurity workflow agents using Deloitte's own use cases.
Exercise A — Firewall Rule Optimizer (45 min)
| Time | Activity | Details |
|---|---|---|
| 0:45–0:50 | Use case briefing | Given a firewall rule report, build an agent that categorizes rules as risky, over-permissive, unused, or redundant — with confidence levels and metadata. |
| 0:50–1:00 | Step 1: Create the agent | Create Agent → Workflow Agent → "Firewall Rule Optimizer". Set instructions and configure the Knowledge Store with sample firewall rule report. |
| 1:00–1:10 | Step 2: Build LLM analysis step | Add an LLM Step that reads the uploaded rule report and categorizes each rule. Craft the prompt — emphasize prompt engineering patterns (role, output format, constraints). |
| 1:10–1:20 | Step 3: Add output formatting | Add a second LLM Step that formats output into a structured report: Rule ID, Category, Confidence, Requestor, Application, Timestamp, Recommendation. |
| 1:20–1:30 | Step 4: Run and iterate | Run the agent on sample data. Review output together. Discuss edge cases, prompt refinement. |
Exercise B — Pathfinder: NIST CSF Compliance Mapper (45 min)
| Time | Activity | Details |
|---|---|---|
| 1:30–1:35 | Use case briefing | Given policies, standards, and configuration evidence, build an agent that maps documents to NIST CSF controls and evaluates compliance status. |
| 1:35–1:45 | Step 1: Create the agent | Create Agent → Workflow Agent → "Pathfinder — NIST CSF Mapper". Upload NIST CSF framework reference and sample policies/standards to Knowledge Store. |
| 1:45–1:55 | Step 2: Build the mapping step | Add an LLM Step that maps evidence documents to NIST CSF controls. Prompt specifies: control ID, control name, evidence document, mapping rationale, coverage level. |
| 1:55–2:05 | Step 3: Build the compliance assessment | Add a second LLM Step that evaluates compliance status per control: Compliant / Partially Compliant / Non-Compliant / Insufficient Evidence — with justification. |
| 2:05–2:15 | Step 4: Run and review | Run the full agent. Review compliance report. Discuss accuracy, ambiguous evidence, extending to ISO 27001, SOC 2. |
Key takeaway: Participants have built two real workflow agents end-to-end and understand prompt engineering, knowledge stores, and multi-step agent construction.
Block 3 — Apply & Extend
3
Apply & Extend
Goal: Independent practice — participants apply what they learned to their own use cases.
| Time | Activity | Details |
|---|---|---|
| 2:15–2:20 | Briefing | Pick your own use case (or extend one of the workshop agents) and build it independently. Facilitators available for help. |
| 2:20–2:40 | Independent build time | Participants work on their own agents. Facilitators circulate (virtually) to help, answer questions, debug. |
| 2:40–2:45 | Show & tell (optional) | 2–3 volunteers share what they built. Quick feedback from facilitators. |
Tips for facilitators:
- For cyber professionals: help with prompt crafting (they know the use case, may need help structuring prompts)
- For full-stack devs: help with use case framing (they know how to build, may need a cybersecurity scenario)
- For AI engineers: push toward advanced patterns — multi-step, API action steps, trigger agents
Block 4 — Wrap-Up & Next Steps
4
Wrap-Up & Next Steps
| Time | Activity | Details |
|---|---|---|
| 2:45–2:50 | Key concepts recap | Agent types, steps, knowledge stores, prompt engineering patterns, integrations |
| 2:50–2:55 | Ongoing resources | Docs site, training portal (LMS), interactive training assistant (chatbot), slide decks, video overviews |
| 2:55–2:58 | Scaling plan | Subsequent batches (75–80 total engineers), iteration based on pilot feedback, in-person sessions later |
| 2:58–3:00 | Q&A & feedback | Open questions. Share feedback form. Thank participants. |
Prerequisites
Share with attendees 3–5 days before the session:
-
Kindo account provisioned — Confirm you can sign in at the training instance URL
-
Browse the docs — Read the What is Kindo? and Best Practices sections (30–45 min)
-
Prepare a use case (optional) — Think of one workflow from your daily work to automate in Block 3
-
Technical setup — Modern browser (Chrome/Edge recommended), stable internet, screen large enough for split-view
Facilitator Guide
Pacing Notes
- Block 1 is presentation-heavy — keep it moving, use live demos over slides
- Block 2 is the core — allocate extra time here if Block 1 runs short
- Block 3 can expand into Block 1 time if orientation runs under 45 min
- Never cut Block 2 short — the hands-on exercises are the highest-value part
- Take a 5-min break between Block 1 and Block 2 (7 AM Pacific = late evening India)
Audience-Specific Emphasis
| Archetype | Extra attention on… |
|---|---|
| Full-stack devs | Agent step types, API Action Steps, integration mechanics |
| Cyber professionals | Use case framing, prompt engineering, Knowledge Store setup |
| AI/ML engineers | Advanced patterns (multi-agent, memory/persistence, Canvas), API reference |
Common Pitfalls
- Participants may try to build overly complex agents in Block 3 — encourage starting simple
- Prompt engineering frustration — have pre-written prompt templates ready as fallbacks
- Integration auth issues — pre-provision all integrations before the session
Pre-Session Checklist
- ☐ All participant accounts provisioned and tested
- ☐ Training instance integrations configured (CrowdStrike, Jira, ServiceNow, Splunk — at minimum stubs)
- ☐ Sample data files ready: firewall rule report, NIST CSF reference, sample policies/standards
- ☐ Backup prompt templates available for both exercises
- ☐ Training portal live and accessible
- ☐ Recording setup (if session will be recorded for later batches)
- ☐ Feedback form prepared
Content Module Mapping
| Module | Block | Role |
|---|---|---|
| 01 — Introduction & Overview | Block 1 | Platform overview presentation |
| 02 — AI Chat & First Steps | Block 1 | First hands-on demo |
| 03 — Agents & Automation | Block 1 | Agent types deep dive |
| 04 — Canvas, Models & Integrations | Block 1 | Integrations overview + advanced reference |
| 05 — Governance & Security | Reference | Admin/governance — background reading for team leads |
| 06 — Deployment Planning | Reference | Not covered (scope excludes install/admin) |
| 07 — AWS Deployment | Reference | Not covered (scope excludes install/admin) |
| 08 — Advanced Deployment | Reference | Not covered (scope excludes install/admin) |
| 09 — Operations | Reference | Not covered (scope excludes install/admin) |
| 10 — Best Practices & Reference | Block 2 | Prompt engineering patterns, agent config management |
| 11 — Admin FAQ | Reference | Background reading |