CTO Activity Report — Sep 2 to Sep 29, 2026
⚡ Window Highlights — Sep 2 – Sep 29, 2026
On the 1:1 side, the proposed agenda went back and forth with GDS's Shqipe Bunjaku and Thomas Wigan through mid-September; on Sep 21 Uzi escalated when three confirmed meetings (Tanita Siripanich, Adel Bekhiet, Axel Ramm) were cancelled without explanation — GDS restored them, and the event closed with 9 confirmed 1:1s. Post-event, six personal follow-ups went out Sep 24–25: Axel Ramm (Trane Technologies) — SDD for embedded code; Nathaniel Rollins (Community Health Choice) — context engineering & medallion architecture, slot booked for Oct 9; Fred Calil (Philips) — SDD and Code2Doc; Chiranjeevi "Jeevi" Kancherla (Sodexo) — agent validation & context engineering, slot booked for Oct 7; Catherine "Cassie" Janis (Chico's) — ProductArt overview and demo, with Hamza; and Vorachat Tamarree (ST Bank) — a live discussion on modernizing check processing, the strongest of the six.
Naresh's own account adds more depth than the email trail alone shows. Beyond the 9 scheduled 1:1s, he made direct connections with: the government of Georgia's ABO; an author who has written a book on AI adoption; a VP of Technology who has worked on multiple agentic solutions; and a director at Cardinal Health, now a named target account — plus four to five more leads he judges qualified and good. Separately, the Philips meeting went well: the connection was with their Director of Product, and Naresh has already handed it to Hamza to follow up and take forward, on top of the SDD/Code2Doc thread Uzi opened by email.
cibc_demo, rab-demand-forecast, ml-assistant-donna, and others) with mixed sign-off and one unilateral deletion by Mehedi when nobody responded — surfaced the underlying problem: excessive Owner/Contributor access and no clear resource ownership. On Sep 29, Kunal proposed a formal Azure IAM & Governance Documentation Suite covering the full access lifecycle — request, approval, RBAC assignment, periodic review, and offboarding — plus a least-privilege re-baseline of existing access.⚡ Window Highlights — Aug 5 – Sep 1, 2026
Post-demo, procurement issued six final clarification questions with a split deadline — written answers Aug 21 5pm EST, revised proposal/timeline/pricing Aug 24 5pm EST. Both were delivered. The substantive asks and our positions:
• Resource location — strong client preference for GTA, or at least Ontario. Impacts staffing mix, timeline and price; answered by Siddiqui/Naresh.
• Expo framework — 407 ETR has decided to proceed with Expo. We recommended it in the original proposal, so no architecture, methodology, resourcing, schedule or cost impact. Known gap: no prior Expo delivery evidence — Bell, Freedom, Indigo and Fido apps predate the framework. References offered: Bell, FIDO, Freedom Mobile.
• QA Automation — added to scope as a dedicated automation team separate from developers, gating every build: Jest + React Native Testing Library at a 75% enforced / 85% target coverage, Maestro E2E against the full feature-parity backlog, performance-regression benchmarking against 407 ETR's own targets (99%+ crash-free, 60fps, 2–3s startup, 10% regression gate), automated WCAG 2.2 AA validation, and SAST + SBOM on every build. Tooling is open source; the automation build/maintain effort is already in the bid sheet at CAD 74,000; ongoing support is explicitly excluded and negotiable.
• Financial overview — Jennifer drafted the privately-held FY2025 position; Hussain pushed to add company size, revenue range, years in operation and business-performance indicators to build credibility.
Sep 1 brought two escalations on the same day. Imran, on the SDK next-steps thread: "can we review and discuss next steps? I need to get back to the customer. It's been more than 2 weeks." He is now trying to book a slot with Kai to deliver an RFI update. Separately, Kashif Ullah raised a hard dependency: validating mobileLIVE's proposed solution requires one physical Android and one iOS device. Siddiqui's response — "I don't think we have any inhouse", then after checking the lab team, "most likely the device will not be available tomorrow."
Ayaz's reply: "lets chat in a couple of weeks. Fridays are typically good for me at 9am." Athar has offered Sep 11 or Sep 18 and has committed to bringing the AI team into the discussion. A separate internal CIBC working session ran with Jigar on Aug 24, and Jigar came from the CIBC site to the ML office on Sep 1 to connect.
The audit bar is evidence-heavy and lands on oCTO: three enterprise application migration/modernization projects delivered to production in the last 12 months, plus customer evidence across assessment, solution design, modernization deployment, monitoring, testing and post-deployment operations; Azure Landing Zone implementations with repeatable deployment methodology; Azure Well-Architected Reviews and governance/security assessments; DevOps, automation and monitoring tooling evidence; customer skilling and adoption plans; and customer sign-off artefacts presentable at audit. Shalin asked Naresh, Jigar and Siddiqui for candidate projects, available documentation, sign-offs, and any known gaps needing remediation before an audit is scheduled. Sync sessions were scheduled Aug 12 and Aug 14.
• Frontier — AI in the Flow of Human Ambition (Copilot, Copilot Studio, agents); Amplify Your Intelligence (Fabric, Azure AI Foundry, grounding); Ubiquitous Innovation (GitHub Copilot, AI-assisted development, AI-native apps); Trusted AI (security, identity, governance, compliance).
• Core — AI-ready Productivity & Security; Agentify Business Processes (Dynamics 365, Power Platform); Unified Data & AI Estate (Fabric, Azure Data); Modernize with Confidence (Azure infrastructure and application modernization) — which is the same territory as the Advanced Specialization above.
The assessment also asks for FY27 industry and segment focus, and the output may include technical enablement, architecture workshops, offer development, marketplace/co-sell readiness, specialization alignment, or customer-reference development. The follow-on Tech Plan Review with Microsoft is Sep 2, 3:00–3:50 PM, with Naresh, Praveen and Hussain invited.
⚡ Window Highlights — Jul 23 – Aug 4, 2026
⚡ Window Highlights — Jul 9 – Jul 22, 2026
⚡ Window Highlights — Jun 24 – Jul 8, 2026
⚡ Window Highlights — Jun 11 – Jun 23, 2026
⚡ Window Highlights — May 28 – Jun 10, 2026
⚡ Week Highlights — May 9–12, 2026
Mohammed Farhan Baluch: Device inventory, MDM and mobileLIVE setup completed. Active contributor on Rogers agent evaluations.
Olalekan Subair: mobileLIVE Microsoft email created; onboarding for CIBC project.
⚡ Period Highlights — Apr 16 – May 8, 2026
Rogers SOW Options — Engineered for Microsoft Co-Funding
Microsoft does not fund good ideas — it funds Azure Consumed Revenue (ACR). Any Rogers deliverable we want co-funded has to do three things at once: solve a real Rogers problem, run on Azure in a way that scales consumption with usage rather than with headcount, and give Microsoft a referenceable win. Both options below are built to that test. They also double as our submission for prep areas 2, 4 and 5 of the Microsoft FY27 joint plan — AI & Frontier strategy, repeatable IP, and Advanced Specialization targets — which is due to Shalin ahead of the August QBR.
Lever 1 · Consumption Growth
The solution must generate recurring, usage-linked Azure spend — inference, capacity, storage, observability — not a one-time migration. Microsoft funds the ramp, then earns on the run rate.
Lever 2 · MACC Retirement
If mobileLIVE transacts the IP through Azure Marketplace, Rogers' spend draws down their Microsoft Azure Consumption Commitment. Rogers spends budget they have already committed — and it still counts for Microsoft.
Lever 3 · Referenceable Win
A named enterprise AI win with governance evidence is worth more to Microsoft than the revenue itself. It also closes the 3-customer-reference gate blocking our Azure AI Apps Advanced Specialization.
Agent Marketplace + Evals
What We Deliver
- Agent Marketplace on Azure — catalog, publishing pipeline, versioning, ownership, and lifecycle management for every agent built across Rogers.
- Evals harness — automated quality, safety, cost and latency gates that every agent must clear before publish, plus continuous post-publish scoring and drift detection.
- Public scorecards in the catalog so an internal consumer sees an agent's measured quality before adopting it.
- Golden datasets per domain, owned by Rogers, that become the regression suite for the agent estate.
- Agent registry with identity, entitlement and observability wired in from day one.
Azure Surface
Why It Drives ARR
- The marketplace is a multiplier: one platform build, then every new agent onboarded adds its own inference load.
- Evals create a second, independent consumption stream — each release and each scheduled re-score is itself inference, and it recurs whether or not agent traffic grows.
- Consumption scales with agent count × release cadence, not with our headcount.
- Builds on a forum Rogers already has (ADLC & Marketplace).
- Little net-new data access needed — fast to scope and sign.
- Responsible-AI evidence is exactly the story Microsoft wants.
- Clean path to co-sell IP and a Marketplace listing.
- Smaller ARR than a data-platform play.
- The ADLC & Marketplace sync has been dark since Jul 23 — needs reviving first.
- Risks being seen as tooling rather than transformation.
The Context Engine
What We Deliver
- Semantic layer — canonical definitions. What "customer" means, how "revenue" is calculated, which field is source of truth.
- Ontology — the relationship model: customer → account → product → contract → obligation → risk and compliance rule.
- Context layer — the differentiator, wrapping both with authority (who or what may act), lineage (where the data came from), policy (what rule applies here), decision memory (what was already decided and why), change awareness (what moved since the last action), and human-review triggers.
- Agent runtime that must consult the context layer before it acts, escalate when confidence or authority is insufficient, and write every decision back to memory.
Azure Surface
Why It Drives ARR
- Fabric capacity is committed monthly spend, not variable — the single largest and most durable ARR line of the two options.
- Purview scales with the size of the data estate, and Rogers' estate is large.
- It is a platform commitment rather than a project: once agents depend on the context layer, the spend is structural.
- Materially larger ARR — Fabric capacity dwarfs inference spend.
- Highly differentiated; few partners can actually deliver it.
- Positions mobileLIVE as architecture authority, not staff augmentation.
- Reusable IP across CIBC, TELUS and Manulife.
- Long sales cycle; needs data-estate access and many stakeholders.
- Rogers may not be ready — most enterprises are still stretching the semantic layer.
- Real risk of degenerating into a data project with no agent outcome.
Sign Option 1 now — but architect it as Phase 1 of Option 2
These are not competing bets. Presented as a choice they split the room; sequenced, they compound. Option 1 is fast, low-risk and immediately fundable, and it produces exactly the raw material Option 2 needs.
Sequencing also gives Microsoft two funding events instead of one — a contained pilot they can approve quickly, then a platform commitment backed by evidence from the pilot. That is a materially easier path than asking for the full context-engine envelope cold. And it lets us name Option 2 in the FY27 plan as committed future IP without having to deliver it before the August QBR.
Decisions Needed
- Which option goes into the Rogers SOW — or do we adopt the sequenced recommendation and write one SOW with a Phase 2 option clause?
- Who owns the Microsoft funding conversation — Shalin leads the ask, but we need the ACR model and the eligible funding vehicle confirmed with the Microsoft PDM before the August QBR.
- Do we list on Azure Marketplace? This is the MACC-retirement lever and it changes the commercial construct for Rogers. It needs a decision, not a discussion.
- Which Rogers sponsor carries this — Sreeraj on architecture, or higher? The ADLC & Marketplace forum has been dark since Jul 23 and needs reviving either way.
- Do we commit Option 2 as named FY27 IP in the Microsoft joint plan even if we only sign Option 1 this year?