AWS cloud-economics work shaped around the decision at stake.

Bedrock Cost Attribution

Bedrock inference cost can be attributed to the calling IAM principal from April 2026 onward. Attribution does not backfill, so the cost history begins when it is enabled.

Method. The engagement enables principal-level attribution, validates it against Cost and Usage Report data, and models provisioned throughput, guardrail, and data transfer contributions to the Bedrock bill.

Deliverables

• Principal-level Bedrock attribution enabled and validated against CUR
• Cost driver breakdown across inference, provisioned throughput, guardrails, and data transfer
• Chargeback structure for teams, products, or features
• Forecast model with the assumptions stated
• 60-minute engineering and finance readout

EDP & Private Pricing Advisory

An Enterprise Discount Program or Private Pricing Agreement commits spend over a multi-year term in exchange for a discount. The sizing and the structural terms determine whether it saves money or carries a shortfall.

Method. The engagement builds the usage and forecast model behind the commitment, tests growth and downside scenarios against demand, migration, and AI adoption assumptions, and identifies which structural terms to raise before signature.

Deliverables

• Amortized cost and usage baseline by service
• Commitment sizing model with growth and downside scenarios
• Shortfall exposure analysis
• Structural terms to negotiate, including ramp, step-down conditions, service scope, and AI spend treatment
• Position paper for procurement, finance, and executive stakeholders

Per-Tenant Cost Engineering

Multi-tenant SaaS architectures share infrastructure across customers, so cost allocation tags describe the resource rather than how its capacity was consumed. Per-tenant cost requires application-level instrumentation.

Method. The engagement instruments services to emit per-tenant usage, lands that telemetry alongside Cost and Usage Report data, and correlates the two to produce cost per tenant, per tier, and per feature.

Deliverables

• Per-tenant usage instrumentation across the services that carry shared cost
• Telemetry and CUR correlation pipeline running in the customer's own account
• Cost per tenant, per tier, and per feature
• Gross margin view by customer cohort
• Handover documentation so the customer's engineers own it afterward

AI Cost Management Training

FinOps and platform teams are being handed responsibility for AI spend faster than the practices exist. Bedrock, inference, and GPU workloads bill on mechanics that traditional cloud cost training does not cover.

Method. A two-day remote workshop built on AWS-native cost tooling and current Bedrock billing behaviour, delivered against the team's own architecture patterns and questions.

Deliverables

• Two-day remote workshop for up to 15 participants
• Bedrock cost mechanics: inference, provisioned throughput, guardrails, and data transfer
• Principal-level attribution and chargeback design for AI spend
• GPU and accelerated compute utilization economics
• Forecasting AI spend, and where AI consumption meets commitment structure
• Workshop materials and reference queries the team keeps

Delivered remotely against the team's own architecture patterns, using materials prepared in advance.

Migration Cost Modeling

A migration business case is built on assumptions about volume, architecture, rate, and commitment structure. The costs that break the case are usually the ones outside the technical scope: parallel running while both environments are live, data egress during transfer, licensing that changes shape in the cloud, and observability that arrives with the new architecture.

Method. The engagement models the target-state run rate against the workload's actual usage profile, tests the business case assumptions against AWS pricing mechanics, and states which assumptions carry the most financial risk.

Deliverables

• Target-state run-rate model by service, with assumptions stated
• Parallel-running cost across the planned cutover window
• Data transfer and egress modeling for the migration itself and steady state
• Licensing, managed-service premium, and observability cost the on-premises baseline does not contain
• Commitment strategy for the post-migration estate, with timing
• Sensitivity analysis naming which assumptions move the case most

scope --other

Other engagements.

Cost investigation, commitment decision review, allocation and evidence foundation work, team enablement, and ongoing investigation partnership are available on request.

scope --exclusions

What I do not offer.

• No 24/7 managed service.
• No staff augmentation.
• No resale or reseller quota.
• No large implementation team.
• No generic transformation program.
• No autonomous remediation or AWS write access, from any service or the Mission FinOps Agent.

engage --init

How all engagements begin.

Thirty minutes, free. A live look at your bill, forecast, or the AWS decision behind it. No slides. Before work begins, you and I agree on the question, evidence requirements, deliverable, timeline, commercial terms, and definition of done.

A disputed bill, a migration that broke the business case, or a commitment nobody can defend are all valid inputs. No commitment, no sales process. I understand the question and decide whether I can genuinely help.

If the question is not mine to answer, I say so. If it requires something I do not offer, I say that too.

access --policy

Access and evidence principles.

Review the data flow, access boundary, and engagement controls at Trust.

capacity --status

Capacity and continuity.

I work on a limited number of investigations at a time and remain directly responsible for every engagement.

If I become unavailable during active work, the timeline is extended at no additional cost, and you retain every artifact and piece of evidence produced to that point.

Because the evidence and method are documented through Kulshan, the investigation does not exist only in private notes or inside a proprietary platform. See the full product platform and the Mission FinOps Agent Early Access page for where this is heading.