SCUTUM.ADVISORY
Productized advisory · MCP · Agentic AI · Cloud

Most agent rollouts stall at governance. Most migrations stall at identity. We fix the parts your team is doing for the first time.

Fixed-milestone engagements for enterprises deploying agentic AI or modernizing cloud workloads. You stop after any milestone. We ship working systems, not slideware.

4–6 weeks typical engagement ₹2.5L–₹50L scoped per outcome 10% kickoff · net-15 milestones

Two things are happening simultaneously at every enterprise we talk to.

Teams are wiring up MCP servers, A2A protocols, and tool-using agents in production without a clear story for access control, policy enforcement, audit trails, or cross-protocol interop. The result: stalled rollouts, security review escalations, and rework.

At the same time, cloud workloads are being re-platformed under cost and AI-readiness pressure. Azure→GCP, AWS optimization, and GKE/Kafka/observability redesigns happen in parallel with the agentic push. Most internal teams are doing both for the first time.

Generalist consultancies bill ₹40–80L for 12-week assessments that produce slideware. We deliver working systems in milestones you can stop at any time.

Three sizes. Milestone-priced. You pay only for milestones you authorize.

Each milestone has fixed scope, fixed price, and a concrete deliverable. Independently valuable. You can stop after any one.

MCP / A2A governance layer Agent eval + observability harness Cloud migration Workload redesign FinOps review LLM security review
Size S01
Diagnostic
₹2.5L – ₹4L · 2–3 weeks

A focused diagnostic when you need an expert second opinion fast, not a multi-month engagement.

For: Director-level decision, discretionary budget, needs an outside view in under 30 days.

Milestone 1 · 1 week · ₹1.5L
Discovery & current-state assessment
Milestone 2 · 1–2 weeks · ₹1L–₹2.5L
Recommendations, roadmap, one proof-point
Size M02
Productionization
₹7L – ₹12L · 4–6 weeks

When you have a real use case and need to go from prototype to production-grade.

For: VP/Head-level decision, dedicated project budget, use case already approved internally.

Milestone 1 · week 1 · ₹2L
Architecture & threat model
Milestone 2 · weeks 2–3 · ₹3L
Policy & governance layer build
Milestone 3 · week 4 · ₹2L
Observability & eval harness
Milestone 4 · weeks 5–6 · ₹1L–₹3L
Handoff, runbook, 30-day support
Size L03
Programmatic
₹20L – ₹50L+ · 3–6 months

When you're building an agent platform, running a multi-workload migration, or setting up a CoE.

For: CTO/CIO sponsor, multi-quarter program budget, internal team to absorb the work. Typically requires MSA.

6–10 monthly milestones · ₹3L–₹6L each
Scoped jointly at kickoff.
Typical shapes
Platform ADRs · reference build · governance layer · CoE enablement · FinOps · security review
Termination
Either party can pause after any completed milestone · 2 weeks notice

First five engagements get founding rates. Three slots remain.

3 / 5Slots remaining
  • 10% kickoff payment (vs. 30% standard)
  • Direct access — no account manager layer
  • Anonymized case study or reference call at engagement close
  • Same milestone scope and quality as full-rate engagements

Built by an architect currently shipping the problem you have.

Published research
Paper on MCP↔A2A protocol translation
90.5% accuracy across heterogeneous agent stacks. One of the first formal treatments of cross-protocol agent interop.
Live production work
Lead architect, 300-microservice Azure→GCP migration
Fortune-500-tier consumer platform. GKE Gateway API, MirrorMaker 2 Kafka topology, Dataproc dual-mode, Datadog + Chronicle.
Working product
Scutum — LLM control plane
Self-hosted MCP/A2A governance, Cedar policy enforcement, agent behavioral attestation. DPIIT-recognized.
Multi-cloud depth
13+ years · AWS SAP · GCP PCA
Senior architecture roles at Tiger Analytics, Oracle (OCI pre-sales), Paytm (incident command), Xebia, Nagarro, Ericsson.
Author
Concrete Generative AI · Amazon
Plus secscan-cli (1,400+ PyPI downloads), AISentry (OWASP LLM Top 10 scanner), By Induction technical channel.
Education
MS, Data Science · IIT/IIM Indore
PhD applications 2026 — formal foundations of multi-agent coordination, session types, protocol verification.

How we work, in plain terms.

Contracting Standard SOW per engagement. Can sign your MSA. Mutual NDA available.
Working mode Remote-first. On-site for kickoff and major reviews if needed (travel at cost).
Cadence Weekly written status. One live working session per week. Async-first otherwise.
IP You own all deliverables and code. Scutum stays ours; deployments licensed for your use indefinitely, no recurring fee.
Payment 10% kickoff (founding) or 30% kickoff (standard). Milestone payments net-15 from signoff.
Confidentiality Anonymized case studies only with written consent. Default: full confidentiality.

If any of this sounds like the problem on your desk, let's spend 30 minutes on it.

If there's a fit, you'll have a one-page SOW within 48 hours. If not, you'll have a sharper read on what to do next either way.

Book 30 minutes →