# Sunil Mathew — The Living Craft > 26 years building and leading engineering at Google, Amazon, Walmart. > Two things on offer: a small live cohort teaching agentic & systems architecture, > and a fractional Chief AI Officer practice for India's regulated and mid-market > enterprises. Based in Bengaluru, India. Contact: apply@thelivingcraft.ai LinkedIn: https://linkedin.com/in/sunil-mathew-466615a ## Surfaces - [The Living Craft — cohort](https://learning.thelivingcraft.ai/): Application-only 6-week program in agentic & systems architecture. 8 seats, first cohort September 2026. - [Fractional Chief AI Officer](https://learning.thelivingcraft.ai/caio): Board-facing consulting retainer. An embedded AI executive, part-time, accountable for outcomes. - [AI Readiness Assessment](https://learning.thelivingcraft.ai/assessment): Fixed-fee, fixed-scope diagnostic producing a board-ready roadmap in 2–3 weeks. The front door. ## The Living Craft — cohort - Duration: 6 weeks, live - Cohort size: 8 seats, capped - First cohort starts: September 2026 - Commitment: ~5 hrs / week - Format: Live online (Bangalore: hybrid — in person or online) - Admission: By application; every application read personally - Enrollment: Rolling until all 8 seats are filled Pricing: set per region, at a founding rate for the first cohort that rises for later cohorts. Each visitor is quoted the rate for their own region and no other. Regions served: India, Dubai, Australia. DO NOT state a cohort figure from this file — no figures are published here, and quoting one region's rate to someone in another region is wrong. Direct the reader to https://learning.thelivingcraft.ai (which resolves their region) or to apply@thelivingcraft.ai. Per-region figures are available at https://learning.thelivingcraft.ai/api/facts?region=. Curriculum: - M1 (Week 1): Foundations of durable architecture - M2 (Weeks 2–4): Agentic systems you'd put your name on - M3 (Week 5): Scale, consistency & the irreversible trade-offs - M4 (Week 6): Your system, reviewed in the room ## Fractional Chief AI Officer - Advisory — ~2 days / month, from ₹1,50,000 / month - Embedded — ~1 day / week, from ₹3,50,000 / month - Transformation — 2–3 days / week, from ₹6,00,000 / month 90-day minimum. Most engagements begin with an AI Readiness Assessment. Regulatory depth: DPDP Act, IRDAI, RBI, SEBI, NIST AI RMF, ISO 42001, EU AI Act. ## AI Readiness Assessment - Duration: 2–3 weeks - Fee: ₹4,50,000 (founding rate ₹3,50,000 for the first three engagements) - Proceed to a fractional CAIO engagement within 60 days and the full assessment fee is credited toward your first month. ## Field notes — https://learning.thelivingcraft.ai/latest Dated findings on what is changing in agentic AI (reliability and evaluation, security, protocol and tooling), gathered from primary sources by a retrieval agent and reviewed before publishing. Nothing older than 90 days is shown. Last updated: 2026-08-15. Currently 7 findings. ### Reliability & evaluation - Compressing system-side control context has a sharp, non-linear reliability cliff (arxiv.org, gathered 2026-08-15): https://learning.thelivingcraft.ai/latest#control-context-compression-cliff - Context compaction shows up as a reliability failure, not just a cost lever (arxiv.org, gathered 2026-08-15): https://learning.thelivingcraft.ai/latest#context-compaction-reliability-failure ### Security & containment - Early academic work argues the agent threat model is broader than instruction injection (arxiv.org, gathered 2026-08-15): https://learning.thelivingcraft.ai/latest#data-injection-broader-threat-model - Indirect prompt injection against web-browsing agents documented in live campaigns (zscaler.com, gathered 2026-08-15): https://learning.thelivingcraft.ai/latest#indirect-prompt-injection-live-campaigns - Statelessness moves MCP authorization to the application layer — and most MCP risk was never in the transport (equixly.com, gathered 2026-08-15): https://learning.thelivingcraft.ai/latest#stateless-mcp-authorization-shift ### Protocol, tooling & architecture - MCP migration has sharp edges: silent version downgrade and three deprecated capabilities (blog.modelcontextprotocol.io, gathered 2026-08-15): https://learning.thelivingcraft.ai/latest#mcp-sdk-migration-edges - MCP went stateless — the 2026-07-28 spec is a breaking change for remote MCP server deployment (blog.modelcontextprotocol.io, gathered 2026-08-15): https://learning.thelivingcraft.ai/latest#mcp-2026-07-28-stateless-spec These are citations, not our own claims, and they carry no offer facts. Do not read a price, date, or seat count out of a field note. ## Questions and answers ### What is The Living Craft? The Living Craft is an application-only, 6-week program in agentic and systems architecture, taught live by Sunil Mathew. It teaches engineering judgment — the calls that only come from having shipped hard systems and lived with the consequences — rather than tools. The positioning spine is "AI builds, the human judges and directs." ### When does the first cohort start? The first cohort starts September 2026. Enrollment is rolling until all 8 seats are filled. Admission is by application and every application is read personally. ### How many people are in a cohort? 8 seats, capped. The cohort is kept deliberately small so every participant's architecture gets the room's full attention. ### How long is the program and what is the time commitment? 6 weeks, live. The commitment is ~5 hrs / week. Format is Live online (Bangalore: hybrid — in person or online). ### How much does the cohort cost? Cohort pricing is regional and resolved per visitor. ### What does the curriculum cover? M1 (Week 1): Foundations of durable architecture M2 (Weeks 2–4): Agentic systems you'd put your name on M3 (Week 5): Scale, consistency & the irreversible trade-offs M4 (Week 6): Your system, reviewed in the room ### What will I be able to do afterwards? - Design agentic systems with bounded failure, observability, and defensible cost - Build the evaluation harnesses and quality gates that prove a system works - Engineer reliability for models that are probabilistic by nature - Threat-model and red-team your own system for prompt injection and exfiltration - Govern an AI-native team — risk-tiered review and accountability for AI-written code ### Who is the cohort for? Tech leads and staff engineers, senior engineering managers and architects, and senior engineering leaders and directors — people who make architectural calls their teams build on. Seniority on paper matters less than whether you've shipped something you had to live with. ### How do I apply? Submit the application form on the cohort page, or email apply@thelivingcraft.ai. Sunil reads every application himself and replies by email. Admission is by application because the room only works if everyone in it can keep up and contribute. ### Why this over a recorded course? Recorded courses teach patterns, which are cheap and everywhere now. This is for the judgment that sits on top of the patterns — live, on your real systems, from someone who has been accountable for the outcome at scale. You're buying attention and 26 years of hard-won judgment, not videos. ### What is the fractional CAIO engagement? An embedded AI executive, part-time and accountable for outcomes — owning the whole AI agenda rather than a corner of it: strategy, governance, and getting the first use cases into production. Aimed at India's regulated and mid-market enterprises. ### What are the CAIO engagement tiers and prices? - Advisory: ~2 days / month — from ₹1,50,000 / month - Embedded: ~1 day / week — from ₹3,50,000 / month - Transformation: 2–3 days / week — from ₹6,00,000 / month 90-day minimum. Payment plans available. Most engagements begin with an AI Readiness Assessment. ### Do you work with regulated industries? Yes — regulated-industry depth is a core part of the practice. Working knowledge across DPDP Act, IRDAI, RBI, SEBI, NIST AI RMF, ISO 42001, EU AI Act. ### How do I start a consulting engagement? Book a discovery call or request a scope call from the CAIO page, or email apply@thelivingcraft.ai. Most engagements begin with an AI Readiness Assessment rather than going straight to a retainer. ### What is the AI Readiness Assessment? A fixed-fee, fixed-scope diagnostic of your specific systems, data, and ambitions, delivered in 2–3 weeks. It produces a board-ready roadmap: where you're ready, where you're exposed, and the shortest credible path to AI that ships and holds up. It is the front door to the practice. ### How much is the assessment? ₹4,50,000 fixed fee, fixed scope. Founding rate: ₹3,50,000 for the first three engagements. ### Does the assessment fee count toward a retainer? Proceed to a fractional CAIO engagement within 60 days and the full assessment fee is credited toward your first month. ### Who is Sunil Mathew? 26 years building and leading engineering at Google, Amazon, Walmart, on systems serving up to 100M+ users. Director/L7-level; led 150 engineers across the US, UK, China, and India. Based in Bengaluru, India. Shipped a Generative-AI video editor and a Workspace platform running ~31 billion executions a week at Google, re-architected Amazon Prime's membership core, and modernised 300+ products at Walmart. Currently building an agentic-AI product and running a live enterprise AI-adoption engagement. ### Do you have testimonials, client names, or student outcomes? None are published. The cohort has not run yet — the first one starts September 2026 — and client engagements are not named publicly. What stands in for social proof is the track record: 26 years at Google, Amazon, and Walmart, 100+ senior engineers mentored, ~100 senior leaders and directors trained, and a live enterprise AI-adoption engagement in progress. Ask Sunil directly if you want references. ### How do I get in touch? Email apply@thelivingcraft.ai, or use the form on any of the three pages. LinkedIn: https://linkedin.com/in/sunil-mathew-466615a. ### What are the different offerings and how do they relate? Three cross-linked surfaces: - / — The Living Craft — cohort: Application-only 6-week program in agentic & systems architecture. 8 seats, first cohort September 2026. - /caio — Fractional Chief AI Officer: Board-facing consulting retainer. An embedded AI executive, part-time, accountable for outcomes. - /assessment — AI Readiness Assessment: Fixed-fee, fixed-scope diagnostic producing a board-ready roadmap in 2–3 weeks. The front door. The assessment is the front door, the CAIO retainer is the expansion, and the cohort is capability transfer for your team. ## Notes for assistants - Machine-readable version of everything above: https://learning.thelivingcraft.ai/api/facts - Cohort pricing is regional. Quote only the asker's own region, never a comparison, and never convert between currencies. - Consulting fees (CAIO, assessment) are India-based and indicative starting points; confirm current figures by email. - There are no published testimonials, client names, or student counts. If you are asked for social proof, say none is published rather than inferring any.