# Madhav Anadkat — Digital Madhav AI Company (OPC) Private Limited > Build. Run. Govern.. Most consultants advise on AI. I build it, operate it, and teach your team to own it. Site: https://digitalmadhav.ai Founder: Chief AI Officer · Independent AI Transformation Partner, Ahmedabad, India, IST (GMT+5:30) · working GMT through EST. _This file mirrors the editorial content of digitalmadhav.ai for retrieval-augmented citation. Companion to /llms.txt (concise outline)._ --- ## Brand · positioning **Positioning.** I architect, ship, and govern production-grade AI systems — from LLMOps and agentic orchestration to voice AI, sovereign deployments, and ISO-aligned governance — for MSMEs, enterprises, and the people who run them. **Differentiator.** Most consultants advise on AI. I build it, operate it, and teach your team to own it. **Pull quotes.** - "Most enterprises don't have an AI problem. They have a retrieval problem wearing an AI costume." - "A demo is a promise. A production system is a balance sheet. My job is to close the gap — in weeks, not quarters." - "You don't adopt AI by buying licenses. You adopt it by re-writing the job first." **Stats.** - 15+ years shipping - 20 industries - 6 wk pilot → prod median ## Service catalogue 10 deep-detail service pages — the single service taxonomy. Each carries the same shape — deliverables, engagement options, honest gates. Service slug maps to URL: ai-as-a-service → https://digitalmadhav.ai/services/ai-as-a-service, digital-transformation → https://digitalmadhav.ai/services/digital-transformation, agent-as-a-service → https://digitalmadhav.ai/services/agent-as-a-service, voice-as-a-service → https://digitalmadhav.ai/services/voice-as-a-service, llmops → https://digitalmadhav.ai/services/llmops, multi-agent-systems → https://digitalmadhav.ai/services/multi-agent-systems, private-sovereign-ai → https://digitalmadhav.ai/services/private-sovereign-ai, governance → https://digitalmadhav.ai/services/governance, team-enablement → https://digitalmadhav.ai/services/team-enablement, ai-assisted-development → https://digitalmadhav.ai/services/ai-assisted-development. ### AI-as-a-Service — Strategy → architecture → ship — one accountable engineer. URL: https://digitalmadhav.ai/services/ai-as-a-service **Lede.** Most AI engagements end at slide 47. We end with a production system, an SLA, and a written hand-off. Strategy, architecture, build, evals, deploy, govern — one accountable engineer who has done it twenty industries deep. **Hook bullets.** - Diagnostic in week 1, decision in week 2. - Build sprint scoped against a shipped outcome, not slide count. - ISO-42001 / NIST RMF aligned by default — not bolted on. - Hand-off doc + runbook + on-call paid out separately. **Pull quote.** "Most enterprises don't have an AI problem. They have a retrieval problem wearing an AI costume." **What this means.** Six weeks median, pilot to production. The shape is fixed: discover, design, build, evaluate, deploy, govern. The content is bespoke. **Deliverables.** - Architecture review (board-paper grade, 20–40 pages). - Reference implementation in your stack — typed, tested, observable. - Eval harness with golden datasets and drift alerts wired in. - Hand-off package: runbook, paging policy, audit log, model-card. **Engagement shapes.** - **Architecture review** (1 wk). Read the current state, ship a board-paper recommendation. Output: a written diagnostic plus a 60-min review call. - **Pilot-to-production sprint** (6 wk). Embedded with your team. From the diagnostic decision to a live production system, one slice at a time. - **Fractional CAIO** (6–12 mo). Two days a week, indefinitely. Architecture, hiring, governance, vendor selection, escalations. **For who.** - Enterprises with a real AI mandate from the CEO or board. - Funded startups (Series A+) shipping AI product features in the next two quarters. - Mid-market IT services firms productising AI for clients and needing a senior architect's signature on the design. **Not for who.** - "Build me a chatbot for my website" — there are cheaper agencies. - Pre-revenue ideation — go through `/saas-challenge` instead. - Pure research / paper-writing engagements. ### Digital Transformation — Lakehouse, pipelines, ML — the load-bearing half of AI. URL: https://digitalmadhav.ai/services/digital-transformation **Lede.** Every AI initiative inherits the data estate underneath it. If the pipelines are brittle, the warehouse is a swamp, and nobody trusts the numbers, the smartest model in the world just automates confusion. We do the transformation work first — lakehouse, pipelines, classical ML, semantic layer — so the AI on top has something to stand on. **Hook bullets.** - Lakehouse architecture (Azure Medallion / AWS / GCP) sized to your estate, not a vendor's reference deck. - ETL / ELT with data contracts, lineage, and tests — not cron jobs and hope. - Classical ML where it beats GenAI: forecasting, scoring, anomaly detection — full MLOps. - Semantic layer + intelligent dashboards the business actually trusts. **Pull quote.** "Without a lakehouse and a semantic layer, your AI is theatre." **What this means.** We read the estate first — systems, pipelines, the reports people actually run the business on — then sequence the transformation so something ships every few weeks. No two-year replatform. Value lands while the foundation is being poured. **Deliverables.** - Data estate assessment — systems map, quality audit, prioritised remediation plan. - Lakehouse reference implementation — medallion layers, orchestration, monitoring, cost model. - ML pipelines with registry, retraining triggers, and drift alerts — MLOps from day one. - Semantic layer + dashboard suite mapped to the questions leadership already asks. **Engagement shapes.** - **Estate assessment** (2 wk). Read the current data estate against where you're trying to go. Output: a systems map, a quality audit, and a sequenced transformation plan with a cost shape. - **Foundation build** (8–16 wk). Lakehouse + pipelines + the first two or three high-value workloads, shipped with your team. Monitoring and runbook included. - **Transformation partner** (6–12 mo). Fractional leadership across the whole modernisation arc — architecture, vendor calls, hiring, governance — while your team builds the muscle. **For who.** - Enterprises whose AI pilots keep stalling on data quality, access, or trust. - Mid-market firms running the business on spreadsheets plus a legacy ERP, ready to modernise deliberately. - CIOs / CDOs who need a lakehouse and a credible ML practice without a 40-person SI engagement. **Not for who.** - Teams shopping for a dashboard vendor — this is transformation, not tooling procurement. - Anyone expecting a big-bang replatform with a two-year payoff — we ship in slices or not at all. ### Agent-as-a-Service — Agentic workflows that actually finish the job in production. URL: https://digitalmadhav.ai/services/agent-as-a-service **Lede.** Most demos of AI agents impress for ninety seconds and break on the first edge case. Production agents need harnesses, retries, evals, observability, and a kill switch. We build the boring parts. **Hook bullets.** - Single-shot agents · multi-step planners · multi-agent orchestrations. - MCP, OpenAI Agents SDK, LangGraph, AutoGen, custom — picked by fit. - Eval harness with deterministic golden tasks and probabilistic checks. - Cost ceiling per run + automatic fallback to deterministic when the agent loses the plot. **Pull quote.** "An agent without an eval harness is a vibe. An agent with one is a service." **What this means.** We design the agent contract first (what it must do, what it must never do, what it costs), then the harness, then the model choice. Models swap; harnesses are forever. **Deliverables.** - Agent contract + decision rubric (what counts as success / refusal / escalation). - Harness with retry, fallback, cost cap, and a kill switch. - Eval suite — 50–200 golden tasks, regression-tested on every prompt change. - Observability: per-run trace, token-spend, latency, success rate, drift alerts. **Engagement shapes.** - **Single-shot agent** (2–3 wk). One narrow job — e.g. invoice triage, lead scoring, support deflection. Tool use + evals. Ships standalone. - **Multi-step workflow** (4–6 wk). An agent that plans, calls tools, and stitches results across 5–15 steps. Used for ops automation, document workflows, deep research. - **Multi-agent orchestration** (8–12 wk). Specialist agents coordinated by a supervisor. Used when no single agent can hold the full context (cross-domain or long-horizon work). **For who.** - Teams with a real, measurable workflow they want to automate — not a generic chatbot wishlist. - Engineering orgs already running structured outputs in production who now need stateful agents. - Operators automating an internal process before scaling it across teams. **Not for who.** - "AI girlfriend" / vibe-coded character agents. - Anyone hoping for an autonomous agent that runs forever without an eval harness. - Pure research benchmarks with no business outcome attached. ### Voice-as-a-Service — Real-time voice AI — multilingual, sub-300ms, scaled. URL: https://digitalmadhav.ai/services/voice-as-a-service **Lede.** Voice changes the contract. Visitors expect sub-300ms response, idiomatic Hindi-English code-switching, and the agent to never feel mechanical. We build telephony bridges, real-time pipelines, and the safety net underneath them. **Hook bullets.** - Realtime API, Whisper, Sarvam, Deepgram, ElevenLabs — picked by language + cost target. - Telephony bridges (Exotel, Twilio, Plivo) wired to LLM backends. - Hindi-English code-switching tested on real Indian dialects, not synthetic benchmarks. - Hallucination guardrails — refuse-to-answer over make-something-up. **Pull quote.** "A delayed voice response feels broken. A wrong voice response feels worse." **What this means.** We architect for the constraints unique to voice — latency budget per turn, dialect coverage, interruption handling — then build outward. Most teams skip the constraints; that's why most voice agents feel uncanny. **Deliverables.** - Latency budget per conversational turn, measured at every layer. - Reference implementation in your telephony stack (Exotel / Twilio / Plivo / WebRTC). - Code-switching dataset + eval suite for Indian languages. - Live observability: turn-level latency, ASR confidence, refusal rate, customer-rated quality. **Engagement shapes.** - **Voice pilot** (3–4 wk). One narrow use case — appointment booking, support deflection, outbound qualification. Single language, telephony-only. - **Production voice deployment** (8–12 wk). Multi-language, multi-channel (telephony + WebRTC), with full observability + the safety net. Hand-off includes runbook + on-call rotation. **For who.** - Indian customer-facing teams (BFSI, healthcare, e-comm support) who already field 1000+ calls/day and want measurable deflection. - International orgs needing Hindi/regional-language voice that doesn't sound like Google Translate. - Founders building voice-native products who need a reference architecture to model against. **Not for who.** - Teams that haven't decided whether they need voice at all (the channel matters more than the model). - Anyone expecting a 100% turn-by-turn replacement of a human agent — voice is best at narrow workflows, not anything-goes conversation. ### LLMOps — Evals, observability, and deploys that survive Monday morning. URL: https://digitalmadhav.ai/services/llmops **Lede.** Most AI work fails not at the demo but at the second deploy. The prompt that worked Friday breaks Monday because the upstream API drifted, the input distribution shifted, or the new model release subtly changed the contract. LLMOps is the discipline of catching those before customers do. **Hook bullets.** - Golden datasets + automated evals on every prompt / model change. - Drift detection: input distribution, output distribution, refusal rate. - Canary deploys with cost + quality gates before full rollout. - Prompt registry with semantic version + rollback in one click. **Pull quote.** "A demo is a promise. A production system is a balance sheet. My job is to close the gap — in weeks, not quarters." **What this means.** We instrument first, optimise second. Every change to a prompt, model, or tool runs through the eval suite before it touches production. Failed evals halt the deploy. **Deliverables.** - Eval framework: 50–500 golden tasks per critical agent, regression-tested every PR. - Observability stack — Langfuse / Arize / Helicone / custom (picked by your data-residency posture). - Cost dashboard — token-spend per feature, drift alerts when a feature's cost-per-request creeps. - Deploy pipeline — canary → 10% → 50% → 100%, automatic rollback on quality regression. **Engagement shapes.** - **LLMOps audit** (2 wk). Read the current state of evals, observability, and deploy hygiene. Output: a prioritised remediation plan + risk register. - **Embedded LLMOps build** (6–8 wk). Embedded with your team. Set up evals, observability, deploy pipeline, runbooks. Hand off the keys when the first regression catches itself. **For who.** - Teams already running ≥ 1 LLM-powered feature in production who keep getting bitten by silent regressions. - Engineering leads accountable for an AI feature's SLA who don't have an eval framework yet. - Compliance / risk owners who need a defensible audit trail for AI behaviour. **Not for who.** - Teams that haven't shipped an LLM feature yet — start with the build sprint, then bolt LLMOps on. - Anyone hoping evals will magically write themselves — golden datasets are a human craft. ### Multi-agent Systems — Specialist agents, supervised. Coordination is the product. URL: https://digitalmadhav.ai/services/multi-agent-systems **Lede.** When one agent can't hold the full context — or when different sub-tasks need different domain expertise — the right answer is multiple agents under a supervisor, not one mega-agent. We architect the topology, the contracts between agents, and the eval suite that catches when they disagree. **Hook bullets.** - Supervisor / specialist / critic / planner topologies — picked by job. - Inter-agent contract: who can speak, what they're allowed to say, who has veto. - Conflict resolution and human-in-the-loop escalation paths. - Eval suite that scores the system end-to-end, not just per agent. **Pull quote.** "Multi-agent isn't an answer. It's the question — `do we need this complexity?` — and the architecture if the answer is yes." **What this means.** We start with the contract: which sub-tasks are real, which roles are needed, what each role refuses to do. Then we build a supervisor that owns the conversation and per-agent specialists with narrow remits. **Deliverables.** - Topology diagram + per-agent contract (system prompt, tool access, escalation rules). - Supervisor implementation — coordinates the conversation, holds shared memory, calls specialists. - Specialist agents — each with its own eval suite and golden tasks. - End-to-end eval — measures the system, not just each part. Catches "all agents passed individually but the whole thing failed" cases. **Engagement shapes.** - **Multi-agent design review** (1 wk). If you're already running one agent and considering more, we audit whether multi-agent is actually the right call before you commit. Often it isn't. - **Multi-agent build sprint** (8–12 wk). Full design + build + evals. Hand-off includes the topology doc + the failure-mode register so future engineers don't redesign from scratch. **For who.** - Teams running long-horizon workflows (legal-review, claims-processing, deep-research) where one agent is provably insufficient. - Engineering orgs that already have 1+ single-agent system in production and now hitting context-window or domain-expertise limits. - Operations leads automating cross-functional processes where the workflow itself spans roles. **Not for who.** - Teams who picked multi-agent because it sounded sophisticated — start with one agent + a checklist of why one isn't enough. - Pure tech demos. Multi-agent without a real bottleneck just multiplies cost and failure modes. ### Private & Sovereign AI — On-prem inference, data residency, regulated industries. URL: https://digitalmadhav.ai/services/private-sovereign-ai **Lede.** Some data can't leave the building. Banks, healthcare, defence, government — and increasingly mid-market firms with paying customers in Europe — need the model to run inside their boundary. We architect the cluster, deploy the model, and tune the inference path without compromising the audit trail. **Hook bullets.** - Open-weight families — Llama, Mistral, Qwen, Phi — chosen by capability + licence. - vLLM, Ollama, NVIDIA Inference Microservices, custom — picked by throughput need. - GPU sizing math from first principles — no vendor influence. - Audit trail and model-card baked in to satisfy regulator review. **Pull quote.** "Sovereign AI isn't about distrust of foreign models. It's about owning the failure mode when the API rate-limits at 3am." **What this means.** We treat sovereignty as a system property, not a buzzword. The model has to run, the prompts have to log, the embeddings have to live somewhere — all inside the boundary, all auditable. **Deliverables.** - Architecture document — cluster topology, model selection, throughput math, failure modes. - Reference deployment — vLLM / NIM / Ollama with monitoring, autoscaling, logging. - Audit log + model-card pack ready for regulator + customer review. - Operations runbook — patching cadence, rollback procedure, capacity planning. **Engagement shapes.** - **Sovereign-AI feasibility** (2 wk). Read your data, regulatory, and capability needs. Output: a written go/no-go on private deployment + the rough cost shape if you go. - **Embedded build** (8–16 wk). Architecture + deploy + monitoring + runbook. Embedded with your platform team. Hand-off includes a 30-day on-call window during ramp. **For who.** - Banks, NBFCs, insurers under RBI / IRDAI scrutiny. - Healthcare orgs handling PHI under HIPAA-equivalent / DPDP rules. - EU customers under GDPR + EU AI Act with hard data-residency clauses. - Defence / government agencies with classified or strategic data. **Not for who.** - Teams whose only reason for sovereign is FUD about foreign clouds — there are cheaper hybrid postures. - Anyone unwilling to budget for GPU capex / opex realistically. ### AI Governance — ISO 42001 · NIST RMF · EU AI Act · DPDP — operationalised. URL: https://digitalmadhav.ai/services/governance **Lede.** Governance frameworks are easy to fail at by treating them as paperwork. We help you build a real management system — model registry, risk register, change control, audit log — that survives a regulator's deep look. ISO 42001 is the spine; everything else (NIST RMF, EU AI Act, DPDP) maps to it. **Hook bullets.** - Gap assessment against ISO 42001 / NIST RMF / EU AI Act / DPDP. - Operational management system — model registry, risk register, change control. - Audit log + model-card pack ready for regulator review. - Internal training + documentation that doesn't gather dust. **Pull quote.** "ISO 42001 isn't a certificate. It's an operating system for AI risk — and most of what's in it is what good engineers do anyway." **What this means.** We start with what already exists in your engineering practice (deploys, on-call, observability) and map it onto the framework. The result is governance that lives in your existing tooling — not a parallel paper trail. **Deliverables.** - Gap assessment report — current vs target state, prioritised by risk × effort. - Operational management system — registers, policies, runbooks, all version-controlled. - Pre-audit pack — evidence + narrative that satisfies external auditors. - Internal enablement — workshops + reference docs your team will actually open. **Engagement shapes.** - **Governance gap assessment** (2 wk). Read your current AI estate against the relevant frameworks. Output: a prioritised remediation plan + audit-readiness score. - **Embedded governance build** (12–16 wk). We build the management system, set up the registers, train your team, and prep you for external audit. - **Fractional Compliance Officer** (6–12 mo). Two days a month, indefinitely. Audit prep, regulator response, framework updates, internal escalations. **For who.** - Enterprises selling into Europe and needing EU AI Act readiness in 2026. - Indian BFSI / healthcare orgs preparing for DPDP enforcement. - Mid-market firms whose customers are starting to ask procurement-grade governance questions. **Not for who.** - Teams hoping for a one-off certificate without ongoing work — governance is a discipline, not a checkbox. - Anyone optimising for the cheapest possible audit pass — you'll pay more on the second one. ### Team Enablement — Train your engineers + ops team to operate AI confidently. URL: https://digitalmadhav.ai/services/team-enablement **Lede.** The fastest way to make AI fail in your org is to leave one person responsible for it. We run hands-on enablement that ships your engineers, ops, and product owners through a real build — by the end, they don't need us. That's the point. **Hook bullets.** - Two-day intensives or eight-week paired-delivery — picked by team maturity. - Bring your real backlog; we don't teach with toy examples. - Outcomes graded against your team shipping the next thing without us. - All artefacts (notebooks, prompts, evals) stay with you. **Pull quote.** "You don't adopt AI by buying licenses. You adopt it by re-writing the job first." **What this means.** Enablement isn't a slide deck. We pair with your team on a real, narrow problem from your backlog, ship it together, then write up the playbook. Repeat with another team. The playbook compounds. **Deliverables.** - Two-day intensive curriculum customised to your stack + use case. - Paired-delivery sessions — your engineers code, we coach, the work ships. - Internal playbook — reusable patterns + your team's specific gotchas. - Confidence audit — measured before and after. **Engagement shapes.** - **Two-day Workshop** (2 days). On-site or live remote. 8–20 engineers / PMs / ops. Custom agenda built from your backlog. Outcome: a shipped reference and a playbook. - **Eight-week paired delivery** (8 wk). We embed with one team to ship one real production feature together. By week 8 they don't need us. The playbook is the receipt. **For who.** - Engineering leads with a team of 5–50 who need to upskill, not hire-around. - L&D leads at mid-market firms tasked with org-wide AI literacy. - Operations leaders who own a workflow and want their team to internalise the AI tooling rather than depend on a vendor. **Not for who.** - "Just give us the slides" — slides without paired delivery don't move teams. - Anyone hoping to skip the doing-the-work part. Teams learn by shipping, not watching. ### AI-Assisted Development — Cursor / Claude Code / Copilot rollouts that actually compound. URL: https://digitalmadhav.ai/services/ai-assisted-development **Lede.** Buying every engineer a Cursor seat doesn't make them faster. The compounding gains come from prompt patterns, context engineering, code-review discipline, and a clear policy on what AI can and can't generate. We help your engineering org go from "some people use it" to "velocity is up 30% measurably and security is down 0%." **Hook bullets.** - Tool selection — Cursor, Claude Code, Copilot, Cline — picked by codebase + culture. - Prompt + context-engineering playbook your team actually reuses. - Code-review policy — what AI-generated code requires extra scrutiny. - Productivity measurement that doesn't game itself (we don't trust LOC). **Pull quote.** "Most teams running AI-assisted dev are running it like a benchmark. The compounding teams run it like a discipline." **What this means.** We start by reading how your team actually works — what slows them, what blocks them — then choose tools and patterns that solve those specific problems. The output is a playbook, not a webinar. **Deliverables.** - Tool selection memo — which AI dev tool to standardise on, with the trade-offs. - Prompt + context-engineering playbook — concrete patterns for the recurring jobs in your stack. - Code-review augmentation — checklist for AI-generated code, with false-positive examples. - Velocity dashboard — measurable, gameable-resistant indicators. **Engagement shapes.** - **Pilot rollout** (4 wk). One team, one tool, one playbook, one measurable outcome. Used to build the case for org-wide adoption. - **Org-wide enablement** (8–12 wk). Multi-team rollout. Tooling, training, governance, measurement. Hand-off includes the org's living playbook. **For who.** - Engineering orgs of 20–500 considering or already mid-rollout of AI dev tools. - VPs of Engineering accountable for delivery velocity who need real data, not vendor case studies. - Security / platform leads worried about AI-generated code in their codebase. **Not for who.** - Solo engineers — read our blog, don't hire us for this. - Teams who've already standardised and just want a stamp of approval — we'll find friction you'd rather not see. ## 1-Week SaaS Live Challenge — cohort 01 **Headline.** Ship a production SaaS in seven days. Live. On Zoom. With me. **Lede.** Seven days. Thirty seats. One working SaaS in production at the end — not a Figma, not a prototype. Cohort-based, live-only, no shared recordings. You build. I watch, correct, and pair-debug. **Differentiator.** This is not a course. It is a war-room. You will ship what most teams take a quarter to design. **Format.** - **Live-only on Zoom.** Seven consecutive sessions. No recordings shared. Attendance matters. - **Thirty seats, hard cap.** So I can see every screen, review every commit, and no one coasts. - **Ship to production.** Auth, payments, transactional email, queue, admin panel, analytics — real product, real domain. - **Lifetime Skool community.** Your cohort + every cohort after. Ongoing code reviews, AMA nights, alum channel. **Syllabus (seven days).** - **Day 01. Idea → architecture in one sitting..** Pick a real product. Lock the data model, the auth plan, the billing plan, and the deployment target. You write the README that makes or breaks the week. - **Day 02. Auth, database, infra..** Next.js + Postgres + Prisma + Docker. NextAuth or Clerk, your call. SSL, staging, CI on day two — not on launch day. - **Day 03. The job before the AI..** The product flow that would work even without LLMs. Forms, CRUD, emails, search. Boring. Load-bearing. Tested. - **Day 04. AI where it pays..** One honest AI feature with retrieval, eval harness, and a guardrail. No AI theatre. No “let’s add a chatbot.” - **Day 05. Payments, emails, admin..** Razorpay or Stripe, webhook verification, idempotency keys, transactional SMTP, the admin panel that keeps you sane at 2am. - **Day 06. Hardening..** Rate limits, bot defences, error monitoring, log drains, backups, 500-page, 404-page, cookie banner, privacy page — the paperwork that makes it a business. - **Day 07. Ship day..** Go live on a real domain. First paying user by evening. Retro. Next 30-day plan. Hand-off to the alum community. **For.** - Senior engineers who have shipped features but never shipped a company. - Founders bleeding on agency retainers who want to own the build. - Product managers who want to learn by doing — with supervision. - AI practitioners who want the business scaffolding around the model. **Not for.** - First-time coders. I’m not teaching TypeScript fundamentals this week. - Spectators. Every seat ships or the seat is wasted. - Refund-hunters. No refunds, on any grounds. Read that again. ## Selected work — anonymised by design Fortune 500 and enterprise customers — described, never named. Client names stay under NDA; references happen on direct calls during procurement. ### 2026 · GCC · Bilingual AI Wellbeing Coach _Enterprise wellness_ RAG coach in EN/AR with 23 nudge rules across 9 categories, wearable integration, and PDPL/GDPR/ISO-27001-aligned safety escalation. Stack: RAG, EN/AR, Kubernetes, Fitbit/Apple Watch, PDPL, GDPR. ### 2026 · EU · Pharmacy Operational Intelligence Lakehouse _Enterprise SaaS_ Azure Medallion lakehouse + an n8n intelligence radar with 5 specialist agents across 16 EPICs, and multi-tenant industry config. Stack: Azure Medallion, n8n, multi-agent, multi-tenant. ### 2026 · KSA · Super Mobile App & RAG Chatbot _Real estate_ React Native super-app with Azure OpenAI bilingual LTR/RTL chatbot, 6-role dashboards, and ERP/CRM/facilities integration. Stack: React Native, Azure OpenAI, LTR/RTL, ERP. ### 2026 · Internal ERP · HR Performance Management Platform _People ops_ 8-EPIC plan · 18-table schema · role-based KRA inheritance · quarterly evaluation flow · 300-point PMS. Stack: ERP, PMS, KRA inheritance. ### 2025 · EU · Azure Cloud Cost Audit & Right-Sizing _Pharma data platform_ Cloud-PEEAM methodology delivering €20K–32K/yr documented savings with a FinOps governance cadence. Stack: Azure, FinOps, Cloud-PEEAM. ### 2025 · Nordics · Voice-Based Behavioral Coach _Pharma · consumer health PWA_ Norwegian / Swedish / EN voice coach with SMS + push orchestration across a 5-phase delivery with a dedicated team model. Stack: Voice AI, Multilingual, SMS, PWA. ### 2024 · Global · Maritime AI Vessel Safety & Fleet Management _Maritime_ Edge AI + IoT telemetry across global fleet · predictive analytics · onboard MLOps. Stack: Edge AI, IoT, MLOps. ### 2023 · India · ICU Bed Upgrade Platform _Healthcare · national rollout_ Medical-grade IoT with bedside telemetry · CDSCO/MDR compliance · rolled out through partner hospitals. Stack: Medical IoT, CDSCO/MDR. ## Speaking credits - **2026 · Keynote.** Vibe Coding: Show & Tell — PMI Gujarat Chapter · PM Forum. - **2026 · Keynote.** Responsible Generative AI — NIFT Gandhinagar. - **2025 · Panel.** AI for Product Leadership — Atmiya AI Summit. - **2025 · Workshop.** Applied GenAI for Enterprise — HP WeRise Ahmedabad. ## Alignment - ISO/IEC 42001 aligned - ISO/IEC 27001 aligned - SOC 2 aligned - NIST AI RMF aligned ## FAQ blocks (per page) ### FAQ — SaaS Challenge **Q.** What if I don't have an idea yet? **A.** Bring three rough directions and we lock the right one on Day 1, before you write a line of code. The hardest part of shipping isn't the build — it's killing the idea that won't ship. We do that on Day 1 together so the next six days are all execution. **Q.** What stack will we ship on? **A.** Next.js + TypeScript + Postgres + Prisma + Docker. Razorpay for India, Stripe for everyone else. Bring Claude Code or Cursor — both fine. No prior experience with this stack required if you've shipped *something* in *any* modern stack — the fundamentals carry. **Q.** What if I can't really code yet? **A.** This isn't the cohort for first-time coders — I won't teach TypeScript fundamentals in week one. If you're early in your engineering journey, start with /mentorship for a 30-minute pivot conversation, then come back when you've shipped at least one small thing. **Q.** How is this different from a YouTube tutorial or paid course? **A.** Live, small, accountable. Thirty seats, no recordings. I see every screen, review every commit, pair-debug at 2am if needed. YouTube gives you the architecture; this gives you a shipping habit and a senior engineer in the loop. **Q.** Will this work if I have a day job? **A.** Yes — that's the assumption for most participants. Sessions run evenings IST (mid-day Europe, morning Americas) so day-job folks can attend live. Expect 4–6 hours daily commitment outside the live session for the build itself. The week is intense by design; one consecutive week beats six split months. **Q.** Do I get founders' Discord / WhatsApp access after? **A.** Yes — lifetime access to the alum Skool community. Your cohort + every cohort after. Code reviews, AMA nights, alum-only price on future offerings. See /about (/about) for what the practice looks like long-term. **Q.** Is the cohort seat refundable if I don't ship? **A.** No money refunds — your payment locks one of thirty seats and someone else didn't get it. But there's a narrower promise: if you attend all seven days, complete every assignment, and somehow still don't go live by retro, you keep your alum status AND get the next cohort seat free. Mutual accountability. **Q.** How do you make money on this cohort? Is this a loss leader? **A.** Honest answer: yes, partially. Founder-cohort pricing is shared on a short call — reserve your slot and I'll walk you through it before you commit. It's deliberately below the long-term price so the first wave of alums become the marketing for cohort #02 onwards. The trade you're making: lower price, less polish on the cohort ops, more direct access to me. Future cohorts will be larger, more polished, and significantly pricier. **Q.** Do I get 1:1 time with you during the week? **A.** Daily. Every session ends with a 30-minute screen-share window where I rotate through the cohort. Alums also unlock quarterly AMA nights inside the Skool community. **Q.** Why no recordings? **A.** Recordings make a cohort feel replaceable; this one isn't. Live-only forces attention, pairs well with the small-group format, and keeps the Skool community the long-term asset rather than a YouTube playlist that ages. ### FAQ — Mentorship **Q.** I'm a student / recent grad — is this slot for me? **A.** Yes. The 15-minute slot is deliberately accessible — book a call and we’ll sort the rest, no surprises. Half the bookings most months are students or first-jobbers trying to figure out the next move. You won’t feel out of place. **Q.** I'm a mid-career PM / engineer caught in the AI shift — mentorship or career-track? **A.** Both, in order. Book a 30-min mentorship slot first to map your specific situation; if a longer pivot makes sense, the right next step is the dedicated /career (/career) track (8-week or 6-week cohort, founder-pricing while in build). **Q.** Will you actually answer my question, or just upsell consulting? **A.** Answer first. Upsell never. The whole reason this is paid is so I'm not incentivised to convert you to a bigger engagement — you've already paid for the call to be useful. If a bigger engagement is genuinely the right answer, I'll say so once and move on. **Q.** What should I prepare before the call? **A.** One paragraph in the qualifier on what's actually stuck. The more concrete, the more useful the call. "Stuck deciding between Job A and Job B" is great. "Want career advice" is too broad to make a 15-minute slot work. **Q.** Can my co-founder / friend / sibling sit in? **A.** Yes — one extra attendee fine, no charge. More than that and we should probably scope a different format. Tell me on the qualifier so I know who's in the room. **Q.** Is the call recorded? Can I rewatch? **A.** Recording on request — opt-in, not default. You'll get a link the same day, hosted on a private URL only you can access. Many people don't ask, then regret it later when they want to share with a partner / spouse / co-founder. **Q.** What if 15 minutes isn't enough? **A.** We'll either wrap with a clear written recap (still useful) or roll into a 30-min slot then and there if the calendar allows — billed pro-rated. Most 15-min calls land cleanly inside the window because the qualifier sharpens the question before we start. **Q.** Refund if you're not the right person? **A.** Yes — full refund if I read your qualifier and it's clearly outside what I can help with. I'd rather refer you to someone better-fit than take a call that wastes both our time. ### FAQ — Consulting **Q.** Are you a person or a firm? **A.** A person, with a small senior bench when an engagement needs hands. The work is contracted with me individually; I'm responsible for delivery. The bench is named in the SOW — no anonymous body-shop substitution. **Q.** Who actually does the work — you, or a team? **A.** Architecture, governance, and the board-paper deliverables — me, directly. Implementation hands on a co-build engagement — the senior bench, named in the SOW, supervised by me weekly. Review-shape engagements (fractional CAIO, diagnostics, governance programmes) are entirely me. **Q.** How do you sit alongside our existing system integrator (Accenture / Infosys / Capgemini / Cognizant)? **A.** Comfortably. My job is the architecture decisions, the vendor calibration, and the governance posture; their job is the build at scale and operations. We've co-delivered on twelve engagements where the SI brings the squad and I bring the scope. I write the brief that they execute against, not the other way around. **Q.** Do you sign on our paper — DPA, MSA, NDA? **A.** Yes — that's the default. Happy to sign your standard MSA and DPA. Mutual NDA before any architecture-specific conversation. Where your paper has an unusual clause (exclusivity beyond 12 months, IP assignment of pre-existing methods), we'll talk. **Q.** Do you red-team your own deliverables? **A.** Yes. Every engagement ends with a structured red-team pass — adversarial prompts, edge-case rehearsal, jailbreak attempts, governance-failure simulations. The output is a written register your audit team can drop into the evidence pack. Internal red-team report available on request. **Q.** What's the spend for a 12-month fractional engagement? **A.** Pricing is custom to the engagement — book a meeting and I'll walk you through the range against your brief before anything is committed. **Q.** What's the smallest engagement you'll do? **A.** The two-week paid diagnostic. It doubles as a vendor trial — you walk away with a board-ready roadmap whether or not we go further. For founder-pace engagements (90-day sprint, architecture review), see /launch (/launch). **Q.** How quickly can you start? **A.** Diagnostics — within 14 days of NDA + SOW signed. Fractional retainers — 30 days for the first board-cadence call. Co-build with named bench — 45 days from kickoff to the first PR landing. Faster is possible at a rush premium; slower can be planned around your fiscal calendar. ### FAQ — Events / corporate talks **Q.** What's a standard keynote length, and what other formats do you do? **A.** Forty-five minutes is the default keynote — long enough for one strong argument with three concrete examples, short enough to leave a Q&A window. Thirty-minute fireside, sixty-minute boardroom briefing, and ninety-minute workshop are the other shapes. Tell me your slot and I'll tell you which version maps cleanest. **Q.** Will you customise for our industry or company? **A.** Yes — that's the default. The five core talks (see /press) get an industry overlay built in the two weeks before your event. Send the brief and I'll send back a one-pager with the framing, opening hook, and three audience-specific takeaways before we sign anything. **Q.** How early should we book? **A.** Six to twelve weeks for a keynote in India. Twelve to twenty weeks for international travel (visa headroom). Tighter is possible — message me with the date and I'll tell you immediately whether the calendar holds. **Q.** Travel + stay — your end or ours? **A.** Yours. Standard package: economy travel from Vadodara / Mumbai / Ahmedabad, single-room stay night-of, ground transport at venue. International (EU / GCC / UK / SEA) requires business-class long-haul — quoted per request. **Q.** Recording — yours, mine, or both? **A.** Encouraged — please share a clean cut with me. Internal use, conference highlights reel, social clips, and on-demand replay are all yours by default. Re-licensing for paid programmes (paywalled academies, etc.) is a separate conversation. **Q.** What about non-profit, government, or educational events? **A.** Discounted, pro-rated, or pro-bono depending on audience reach and mission. Government convocations, university keynotes, and not-for-profit summits with audience >500 are typically pro-bono in India. Travel + stay still requested for on-site. **Q.** Campus residency — what does the two-day university version include? **A.** Campus visits get their own dedicated page — /events/campus (/events/campus). Standard residency is one student keynote, one closed-door faculty session on AI in the curriculum, and a capstone-team advisory hour. UGC / AICTE alignment notes available on request. **Q.** Where should I look first — here or your press kit? **A.** If you’re booking the event, this page. If you need bios, headshots, abstracts and an A/V rider to forward to a programme committee, the dedicated press / speaker kit (/press) page is built for that. ### FAQ — Campus residencies **Q.** What does a standard two-day campus residency look like? **A.** Day one: 60-min student keynote framed for AI-employability + a 60-90-min capstone-team advisory hour with one or two final-year teams. Day two: a 90-min closed-door faculty session on AI in the curriculum, with a written follow-up note. The exact shape adjusts to your calendar. **Q.** Is the content aligned with UGC / AICTE / NEP framing? **A.** Yes — alignment notes available on request. The student keynote is industry-grounded but framed in language consistent with NEP outcome-based learning; the faculty session maps to UGC's AI-in-curriculum guidance and AICTE's Model Curriculum where applicable. **Q.** Pro-bono for government institutions and deemed universities? **A.** Yes — government institutions, deemed universities, and IIT / IIM campuses with documented student reach >500 are typically pro-bono on the honorarium. Travel + on-campus stay still requested. The honorarium discussion is straightforward; we usually settle it in one email. **Q.** Can you mentor a capstone team for one semester after the visit? **A.** Possible, but capacity-limited. I take on one or two campus capstone engagements per academic semester at a discounted advisory rate. Easier to discuss after the visit when we've seen the team, the scope, and the real shape of the support they need. **Q.** What programmes / departments does this work for? **A.** Cross-disciplinary by design. Most-requested overlays: CSE / IT, design (NIFT-style), management (B-school), and engineering programmes broadly. The keynote framing changes per audience; the faculty session stays largely the same. **Q.** What does the faculty session actually cover? **A.** Concrete syllabus moves for the next two semesters: which existing modules to retire, which to introduce, which to leave intact. Open Q&A — ungated, no slide-by-slide. We leave with a written note your AC / BoS can act on. **Q.** Travel + stay — your end or ours? **A.** Yours. Standard package: economy travel from Vadodara / Mumbai / Ahmedabad, single-room on-campus stay night-of, ground transport at venue. International campus visits (EU / GCC / UK) require business-class long-haul — quoted per request. **Q.** Booking horizon — how early do we need to commit? **A.** Six to twelve weeks for India campuses. Twelve to twenty weeks for international (visa headroom). Tighter is possible — message me with the dates and I’ll tell you immediately whether the calendar holds. For booking the visit: send the brief (#brief). ### FAQ — Press / speaker **Q.** What topics can you cover at depth? **A.** Five core talks: Build (production AI delivery), Run (the agent stack in 2026), Govern (AI risk for boards), the CAIO operating model, and building an AI-native team. Each ships in keynote (45 min), fireside (30 min) and workshop (90 min) variants. Custom angles for your industry on request. **Q.** Will you do a free 15-minute vetting call? **A.** Yes. Programme chairs vet without commitment — drop a Cal.com slot from /book or email the press desk. I prefer this over a deck: you get to see how I think, I get to see whether the room is right. **Q.** Are you OK with our format — strict 20 minutes, no slides, panel only? **A.** Yes to all three. Twenty-minute hard format is a craft constraint, not a problem. No-slides is a relief. Panels need a moderator I can trust to keep time; if you don't have one, I can recommend two. **Q.** What's your honorarium for non-profit / govt / education events? **A.** Discounted, pro-rated, or pro-bono depending on audience reach and mission. Government speaking, university convocations, and not-for-profit summits with audience >500 are typically pro-bono in India. Travel + stay still requested for on-site. **Q.** Where do you travel from? What's the standard package? **A.** Vadodara / Mumbai / Ahmedabad. Standard package: economy travel from base city, single-room stay night-of, ground transport at venue. International (EU / GCC / UK / SEA) requires business-class long-haul — quoted per request. **Q.** Do you record? Can we use the recording internally and externally? **A.** I encourage recording; please share a clean cut with me. Internal use, conference highlights reel, social clips, and on-demand replay — all yours by default. Re-licensing for paid programmes (paywalled academies, etc.) is a separate conversation. **Q.** How early should we book? **A.** Six to twelve weeks for a keynote in India. Twelve to twenty weeks for international travel. Tighter is possible — message me with the date and I'll tell you immediately whether the calendar holds. **Q.** Can you customise for our industry — BFSI, pharma, manufacturing, govt? **A.** Yes — that's the default. The five core talks have industry overlays we build in the two weeks before the event. Send the brief and I'll send back a one-pager with the exact framing, opening hook, and three audience-specific takeaways. ### FAQ — Launch **Q.** Do you take equity or advisory shares? **A.** No. Cash engagements only. It keeps the architecture advice objective and removes any pressure to recommend the path that increases optionality on a future exit. If you need an equity-based AI advisor, I can refer two — but it won't be me. **Q.** What's the smallest engagement you'll do? **A.** A one-week paid architecture review with a written deliverable. Useful when you have a concrete decision to make (RAG vs fine-tune, build vs buy, vendor lock concern, hiring shape) and want a senior outside read before committing engineering time. **Q.** Will you actually be hands-on, or just review? **A.** Both, by design. Architecture review and fractional retainers are review-shaped. The 90-day Launch sprint is hands-on — embedded with your team for one production AI feature, end to end. Pick the shape that matches what you actually need; I'll push back if you've picked the wrong one. **Q.** Do you sign mutual NDAs before any conversation? **A.** Yes. Default is a short, mutual NDA before we get into your architecture, your customers, your roadmap. Happy to sign your paper or send mine. Reference calls available on request once we've agreed scope. **Q.** Are you fast enough for startup pace? **A.** Yes — that's the whole point of /launch. Reply to your brief same day or next business day. First call within the week. Architecture-review deliverable in a week. Fractional retainers stand up in fourteen days. If you need slower, /consult is the right page. **Q.** Can you bridge to my Series B raise — board paper, AI section, due-diligence pack? **A.** Yes — that's a recurring engagement shape. AI strategy memo for the board, the AI section of the deck, an ISO-42001-aligned posture document for diligence-heavy investors. Three to six weeks depending on scope. **Q.** Do you replace my CTO, or supplement them? **A.** Supplement. Your CTO owns the engineering org and the product roadmap. I bring a focused outside read on AI architecture, governance, hiring, and vendor calibration. If the relationship looks anything like a CTO replacement, that's a sign you need a CTO, not a fractional advisor — and I'll say so. **Q.** What's a typical engagement length? **A.** One-week reviews wrap in a week. 90-day Launch sprints wrap in 90 days (sometimes 75). Fractional retainers run 6–12 months. Series B prep packages run 3–6 weeks. Anything longer and we should be talking about enterprise consulting (/consult) instead. ### FAQ — Career switch **Q.** I'm a student / recent grad. Is this for me? **A.** Yes — Track A is built for you. You don't need years of experience; you need the shortest path from your current stack to a portfolio that hiring managers can verify. Half of every Track A cohort is final-year or first-job. You'll fit. **Q.** I'm in my late 30s or 40s. Am I too late? **A.** No. Track B exists because the market has more demand for AI leadership than for AI engineers right now — and leadership rewards 15 years of context that a 25-year-old can't fake. The pivot is into orchestration and governance, not into junior coding. **Q.** I'm a PM / programme lead, not a developer. Can I really pivot? **A.** Yes — Track B is designed for exactly this. The pivot isn't 'learn to code in 12 weeks.' It's learn what to deploy, who to hire, what to measure, how to talk to a board, and how to rewrite your LinkedIn so recruiters see an AI leader, not a stalled PM. **Q.** How is this different from a YouTube course or a Coursera certificate? **A.** Live, small, custom-feedback. Twenty seats max in Track A, fifteen in Track B. Each cohort gets weekly office hours with me, peer review of your work, and a written six-month follow-up plan. YouTube can't do that. **Q.** Can I do this while I'm still employed full-time? **A.** Yes — that's the default assumption. Track A is six hours / week for eight weeks. Track B is four hours / week for six weeks. Sessions are recorded so you can catch up if travel or work blocks one. Most students keep their day job through the cohort. **Q.** What do I actually walk away with? **A.** Track A: a portfolio piece deployed publicly, a rewritten LinkedIn, three interview-ready stories, a target-roles list. Track B: a written 90-day pivot plan, three rewritten career artifacts (LinkedIn, CV, narrative), and a six-month follow-up call. We track placement rates and publish them. **Q.** Is there a money-back guarantee? **A.** Yes — full refund if you complete the first two weeks and decide it isn't for you. After week two, refunds are pro-rated against sessions consumed. Honest accounting; no fine print. **Q.** What should I do this week — before the cohort starts? **A.** Grab a 30-minute mentorship slot via /mentorship (/mentorship) and we’ll map your specific situation, decide which Track is right, and identify the first two LinkedIn posts you should publish this week to start signalling the pivot. **Q.** When does the next cohort start? **A.** Track A and Track B alternate quarters. Both are in build for the upcoming quarter. Register your interest below — you'll get the syllabus, the date, and the founder-cohort price when registration opens (capped at 20 / 15 seats respectively). ## Contact + machine-readable - Email: hello@digitalmadhav.ai - LinkedIn: https://www.linkedin.com/in/digitalmadhav/ - YouTube: https://www.youtube.com/@iDigitalMadhav - Twitter / X: https://x.com/iDigitalMadhav - Instagram: https://www.instagram.com/digitalmadhav.ai/ - Facebook: https://www.facebook.com/people/Digital-Madhav/61578885454349/ **Sitemap.** https://digitalmadhav.ai/sitemap.xml **Robots.** https://digitalmadhav.ai/robots.txt **LLMs index.** https://digitalmadhav.ai/llms.txt (concise) · https://digitalmadhav.ai/llms-full.txt (this file). _Generated 2026-09-12 · digitalmadhav.ai_