Kindo × Deloitte
From Strategic Framework
to Execution Plan

Connecting the May 7 strategy session to the delivery engine — framed for Ron, adaptable for Kush/Krishna next week.

Source May 7 In-Person Updated May 21 v1 Ron v2 Kush/Krishna
1

Executive Summary

Deloitte Cyber Operate is rebuilding its entire delivery model around Kindo — not evaluating, committing. Krishna mapped Kindo into every delivery model and service line.

$5.5M
Current /yr
$6.5–7M+
With Net New Rev
40→80%
EBITDA Target
100
Installs by Feb '27
$800M
Cyber Operate Rev

Why Now

  • May 31: all non-contract MXDR → Kindo (Krishna directive)
  • HP (F50): first production client install — happening now
  • Alliance agreement in progress — same tier as Google, SAP
  • Mythos vulnerability wave = urgent demand for these exact agents
The Delivery Engine Tony + Joana + Victor + Dukane + Charlie's team + Warren. Tony as CDO — accelerating compound learning loops that make each agent better every month. The moat: institutional knowledge encoded in production agents.
2

EBITDA Strategic Framework

40% → 80% improvement

Cost Elimination ~25–35%

LOW IK
  • Alliance contract — not controllable
  • Swimlane + CrowdStrike sunset
"Sunset Swimlane in every which capacity" — Krishna

Scale Efficiency ~25–30%

MED IK
  • Same pool → more clients
  • Triage: 21→5 min (76% ↓)
  • Human effort: 70–85% ↓
  • Audits: sample → 100%

Net New Revenue ~35–45%

HIGH IK
  • A6–A13: each a revenue event
  • $5.5M → $6.5–7M+ growth
"Every agent is a net new revenue goal" — Krishna
~60–75% of EBITDA improvement flows through institutional knowledge Three compound learning levels: ① User (analysts improve monthly) ② Agent ("Week 10 vs week 6?" — Kush) ③ Organizational (cross-agent learning). Speed to production = compound learning velocity = EBITDA acceleration.
3

Revenue Structure — Contracted vs. Net New

A1–A5: Contracted ($5.5M) — No Net New

Fulfill existing license. Cost us to deliver (IK transfer) but no incremental revenue.

IDAgentStatusModel
A.1Threat MonitoringPRODMXDR
A.2Threat IntelPRODMXDR
A.3Threat HuntPRODMXDR
A.4Detection EngPRODMXDR
A.5CTEMBUILTMXDR

A6–A13: Net New Revenue — Growth Engine

Each is a revenue event. Push $5.5M → $6.5–7M+. Justifies CDO role.

IDAgentStatusModelPh
A.6Vitals DashboardPLANCross2
A.7Quality AuditPLANCross2
A.8Cloud SecurityPLANDed/Sh3
A.9IR AgentPLANDed/Sh3
A.10IoT/OTPLANDed/Sh3
A.11Custom ClientPLANBespoke3
A.12Identity→IdaaSPLANNew SL4
A.13GRC→GRC aaSPLANNew SL4
42 total items: 11 Contracted · 10 Alliance Revenue · 12 Alliance Institutional · 9 Ops. 23 of 42 (55%) depend on institutional knowledge. See full scope matrix →
4

Agent Packaging by Service Line

1. D&RaaS

Krishna · ACTIVE
  • A.1–A.5 (4 PROD + 1 BUILT)
  • A.6 Vitals, A.7 Audit, A.9 IR
  • Serves: MXDR, Shared, Dedicated

2. CaaS

Nathan Ellis · PH 2–3
  • Custom CaaS agents (TBD)
  • A.13 GRC crossover
  • Nathan owns first 5–7 deploys

3. Identity aaS

Tim Corder · PH 4
  • A.12 Identity Agent
  • J&J team (Adelina)

4. Cloud & Infra

Bhargav · PH 3–4
  • A.8 Cloud Security
  • Firewall provisioning on Kindo

5. GRC aaS

Nathan · PH 4
  • A.13 GRC Agent
  • Compliance workflows

6. App Security

No owner · FUTURE
  • TBD — Phase 4+
SOAR Flow (Kush, May 7) Triage agent → calls DE + CTI sub-agents → context returns → containment loop. Kindo Eng building (beta). See operational map →
Revenue per client: ① Base D&RaaS bundle ② Service-line add-ons ③ Bespoke custom agents (A.11) ④ Private MCP integrations. Each layer = incremental revenue.
5a

Platform Priorities — Critical

May 7 asks → current status

1. SMK Instance Stability

1ST INSTALL DONESANDBOX TESTING

Ask: Click-click-click installs (was 3–5 days).

Now: 1st prod SMK in ITS this week. Installer/upgrader/preflight in May 27 release. Observability MVP in final testing.

2. Release Parity (SaaS ↔ SM)

CLOSING MAY 27

Ask: "You keep getting this question from me" — Kush

Now: May 27 = 15+ features to SM. Chatbot APIs, Version Control, Pinned Creds, ServiceNow, MITRE, Member API Keys. Biggest parity close yet.

3. Agent Memory & Self-Improvement

NOT STARTEDKUSH'S #1

Ask: 3-layer compound learning (user → agent → org). "In Kindo, I did not see any of this stuff today."

Now: Not in May 27. Requires platform architecture. Risk: degrades compound learning in HP shadow.

4. Multi-Agent Orchestration

BETA — FF ON SMK

Ask: Supervisory triage → DE + CTI sub-agents.

Now: Agent-to-Agent FF enabled on Deloitte SMK (calibrated rollout). GA gated on resource hardening.

5b

Platform Priorities — High & Medium

May 7 asks → current status

5. Integrations — MCP Ecosystem

PRIVACY IN DEVNEW MCPs SHIPPING

Shipping May 27: ServiceNow triggers, MITRE ATT&CK, Dynamic API resolution

In review: SailPoint writes, PostgreSQL, Jira attachments

Urgent: Zscaler ZIA for May 27 demo; Swimlane fix (TEK-141)

6. Agent Reliability & DX

SHIPPING MAY 27

Done: Long-run reliability + Plan Mode, Agent Version Control (GitOps), Pinned Credentials, Error UX, Chat Actions API, Chatbot APIs

Backlog: Error messages (8798), re-run failed step (10190), resizable windows (9378), prompt filtering (9967)

7. Token / Cost Optimization

ROADMAP

"$25K/month, 80% LLM" — Nathan. Four strategies planned: auto model selection, better context, structured memory, compaction. Not in May 27.

8. GenUI / Canvas

DEPRIORITIZED

"Hold back on Canvas. We'll use TrueArch Hub." — Kush. Chat Actions API (shipping May 27) powers it. Kindo = backend/API.

6a

HP Deployment RACI — Phase 1 & 2

First production client install (Dedicated MSS)

R Responsible A Accountable C Consulted I Informed

ActivityKrishnaNathanKindoJoanaTonyNotes
Phase 1 — Installation & Document Ingestion
SMK provisioning (AEF)ARRCINathan + Brandon; AEF for prod
Security & NEC reviewARCIINathan + Harish
D&RaaS agent deploymentCARCIBrandon + Marcos: A.1–A.5
HP integrations (private MCP)CACIIRobby's team; data stays in HP
ITSM ingestion (6 mo)ACCRIWarren + platform
SOP & doc ingestionAIIRIJoana + Warren
Phase 2 — Shadow (Parallel Operation)
Ticket mirroringCARIIKindo configures pipeline
Agent monitoringCIIRIAnalysts + Warren compare
Human feedbackCICAIFeeds compound learning
Accuracy trackingCIIRIVitals Dashboard proof points
Weekly reviewACIRCJoana runs; Kush consulted
6b

HP Deployment RACI — Phase 3 & 4 + Risks

ActivityKrishnaNathanKindoJoanaTonyNotes
Phase 3 — Reverse Shadow (Agent-Primary)
Agent primary executionACRIIPlatform runs; Krishna to Kush
Human oversight (15%)CIIAIShiva's analysts; Joana tracks
Validation (acc/compl/cons)CICRIWarren + Audit Agent (A.7)
EBITDA trackingCIIRAFlows: us → Deloitte → client
Go/no-go steady stateRCCCIKrishna recommends; Kush decides
Phase 4 — Steady State (Production)
Autonomous execution (70%)AIRIIKrishna owns outcomes
100% audit coverageAIRCIAudit Agent (A.7)
EBITDA reportingCIIRAProof points for upsell
Custom agent expansionCCRAIWarren + Kindo build

Key Risks

RiskImpactMitigationOwner
Platform stabilityBlocks Ph 1Sandbox hardening; Nathan cleanupNathan + Brandon
Agent memory gapDegrades learningManual IK during ShadowKush
Release parityLimits visibilityMay 27 release closes gapKindo Eng
AEF env decisionDelays provisioningHP = prod = AEFNathan
7a

Current Operational Status

Week of May 19 — Three parallel workstreams toward May 31

🔧 Platform

15+
Features in May 27 release
  • Pinned Credentials LIVE
  • Agent Version Control (GitOps)
  • Chat Actions API
  • Long-running reliability + Plan Mode
  • ServiceNow triggers + Dynamic API
  • MITRE ATT&CK
  • Agent-to-Agent FF ON
  • Member API Keys (SMK)
  • Sandbox stability — final testing
  • SMK installer + Observability MVP
Build May 26 · Upgrade May 27

🚀 Deployments

1st Done
First SMK production install (ITS)
  • DONE 1st prod SMK in ITS
  • 🚨 May 31 — Swimlane migration
  • HP planning next (Ded MSS, AEF)
  • Target: repeatable install package
Krishna: MXDR → Kindo by May 31

🎓 Training

26 / 75
Engineers trained toward 100-install
  • DONE Cohort 1 (26/31 attended)
  • NEXT Cohort 2 (June 1st–2nd wk)
  • Practitioner + Technical tracks
  • 6 domains: Platform, Impl, Integrations, Governance, Use Case, Ops
  • LMS + auto new-hire paths
Ongoing: certification tracking
7b

Execution Timeline

Phase 1: Foundation — Now → May 31
  • ✅ A.1–A.4 in production · A.5 built
  • ✅ First SMK prod install in ITS
  • ✅ May 27 release: 15+ features, agent-to-agent FF
  • ✅ Training Cohort 1 (26/31)
  • 🚨 Swimlane migration May 31
  • 📋 HP deployment planning · Cohort 2 (June)
Phase 2: First Client + Scale Prep — June
  • HP Phase 1 (SMK provisioning + doc ingestion)
  • Alliance agreement draft contract
  • MXDR efficiency data → proof points
  • D&RaaS agent package finalized
  • 2nd/3rd client ID from renewal pipeline
Phase 3: Shadow + Expansion — Jul–Aug
  • HP Phase 2 (shadow — parallel operation)
  • A.6 Vitals + A.7 Audit development
  • CaaS integration planning (Nathan)
  • 10–20 additional MXDR installs
Phase 4: Production + Net New — Sep → Feb 2027
  • HP → steady state · A.8–A.11 dev
  • Dedicated MSS scaling (multi-F50)
  • Identity aaS expansion
  • Target: 100 installs by February 2027
8

Kindo Team Considerations

⚠ Internal Only — Not for Ron

Validate with Deloitte

  1. A1–A5 vs A6–A13 accurate against contract? ($5.5M = platform license)
  2. Platform blockers colliding with May 31?
  3. Krishna's directive aligns with deployment team tracking?
  4. HP timeline — when does Phase 1 provisioning start?

Resource Ask

Tony + Joana + Victor + Dukane + Charlie's team. Net new revenue (A6–A13) justifies: $5.5M → $6.5–7M+ doesn't happen without this team.

Warren's Execution Role

  • Warren: Scope decomposition, ITSM analysis, agent config, EBITDA dashboards, readiness checklists, SOP processing, audit reporting
  • Human: Product decisions, client relationships, alliance negotiation, custom agent design, go/no-go gates
Mythos Connection (Ron's frame) Agents Deloitte needs (A.1–A.9) = agents Mythos clients need. Tony's delivery builds Ron's catalog without separate funding. Ron pitches speed → Tony's team delivers speed.

Reference: Scope Matrix v1.0

42-item scope: agents, platform, delivery, service lines, operations — with status, IK dependency, focal person.

Scope Matrix v1.0 Click for full screen
Scope Matrix

Reference: Krishna D&RaaS Operational Map

Leadership, ops, service lines, agents, and Kindo integration points across Cyber Operate.

Krishna DREAS Operational Map Click for full screen
Krishna Map