HardpointAI Advisory · Professional Services

Engagements

Fixed scope, written artifacts, dated plans.

Hardpoint does not sell open-ended retainers or discovery that bills by the hour. Each engagement has a defined artifact list, an explicit boundary, and a completion condition — and each one has to earn the next.

Assessment-first · fixed fee

Primary offers

Six engagements you can actually buy.

Every card states best fit, what you provide, what Hardpoint does, what is written down, where the boundary sits, when it is finished, and what typically follows.

All layers

Four-Layer Architecture Assessment

Environments where the overall architecture, ownership, or sequence is unclear.

Client inputs

  • · System and provider inventory
  • · Access to architecture owners
  • · Existing diagrams and decision history

Hardpoint activities

  • · Structured interviews across all four layers
  • · Current-state reconstruction
  • · Risk, dependency and ownership analysis
  • · Target-state design and sequencing

Written deliverables

  • · Current-state map
  • · Risk and dependency register
  • · Four-layer findings
  • · Target-state architecture
  • · Prioritized implementation roadmap

Engagement boundary

Assessment and design only. No production changes are made during the engagement.

Completion condition

Findings, target-state architecture and roadmap are delivered and reviewed.

Likely next phase

A bounded pilot or implementation phase against the roadmap.

Start this engagement Read the practiceWalk the guided tour— publishing soon

Knowledge

OKF Readiness Assessment

Fragmented knowledge that enterprise agents cannot reliably discover, trust, or maintain.

Client inputs

  • · List of candidate source systems
  • · Knowledge owners for each domain
  • · Sensitivity and retention policy

Hardpoint activities

  • · Source and ownership inventory
  • · Domain prioritization workshops
  • · Sensitivity and exclusion analysis
  • · Draft taxonomy and lifecycle modeling

Written deliverables

  • · Source and ownership inventory
  • · Knowledge-domain prioritization
  • · Sensitivity and exclusion rules
  • · Draft concept taxonomy
  • · Lifecycle and provenance requirements
  • · Bounded pilot recommendation

Engagement boundary

No bulk extraction and no custody of production content. Selected domains only.

Completion condition

A recommended pilot boundary is agreed with the knowledge owners.

Likely next phase

Enterprise OKF Transformation Pilot.

Start this engagement Read the practiceExplore a synthetic OKF transformation— publishing soon

Knowledge

Enterprise OKF Transformation Pilot

One business domain ready to be modeled, governed and consumed by agents or retrieval.

Client inputs

  • · One business domain
  • · Two or three approved source systems
  • · One or two agent or retrieval use cases
  • · A defined corpus boundary and evaluation question set

Hardpoint activities

  • · Extraction and normalization
  • · Concept and relationship modeling
  • · Provenance and lifecycle workflow design
  • · Git review process and consumer integration
  • · Evaluation harness construction

Written deliverables

  • · Extraction procedure
  • · Normalized staging model
  • · Illustrative or validated OKF bundle
  • · Provenance and lifecycle workflow
  • · Git review process
  • · Consumer integration
  • · Evaluation harness
  • · Operating handoff

Engagement boundary

Bounded to the agreed domain, sources and corpus. Not a universal enterprise conversion.

Completion condition

The pilot bundle passes the agreed evaluation set and ownership is transferred.

Likely next phase

Domain-by-domain knowledge transformation program.

Start this engagement Read the practiceInspect sample artifacts— publishing soon

Traffic

AI Model Portfolio & Routing Assessment

Multiple providers, subscriptions, APIs, local models, tools or human-review paths with no consistent routing model.

Client inputs

  • · Provider and entitlement inventory
  • · A representative sample of real workloads
  • · Policy, privacy and locality constraints

Hardpoint activities

  • · Workload classification
  • · Acceptance-contract definition
  • · Eligibility and entitlement mapping
  • · Route ranking, fallback and continuity design

Written deliverables

  • · Workload classes
  • · Acceptance contracts
  • · Model and provider inventory
  • · Entitlement inventory
  • · Eligible-route matrix
  • · Ranked routing table
  • · Fallback design
  • · Continuity design
  • · Routing Receipt schema
  • · Implementation roadmap

Engagement boundary

Design of the routing model. Production routing is not changed during the assessment.

Completion condition

The routing table, fallback design and receipt schema are accepted.

Likely next phase

Shadow Routing Pilot, then implementation.

Start this engagement Read the practiceTest a class-of-service decision— publishing soon

Traffic

Shadow Routing Pilot

Evaluating candidate routes alongside the current production path before changing consequential behavior.

Client inputs

  • · A bounded, representative workload set
  • · The current production route as baseline
  • · Agreed evaluation criteria

Hardpoint activities

  • · Baseline capture
  • · Candidate route configuration
  • · Parallel evaluation and correction measurement
  • · Continuity-defect tracking

Written deliverables

  • · Bounded representative workload set
  • · Baseline route
  • · Candidate routes
  • · Evaluation criteria
  • · Correction and retry measurement
  • · Continuity-defect tracking
  • · Routing receipts
  • · Promotion or rejection recommendation

Engagement boundary

Candidate routes run alongside production; user-facing behavior is unchanged until promotion is approved.

Completion condition

A promotion or rejection recommendation is delivered with the receipts behind it.

Likely next phase

Implementation of the promoted routes, or a revised candidate set.

Start this engagement Read the practiceTrigger a failover scenario— publishing soon

Operations

Production Assurance Review

Systems missing clear evaluation, observability, failure handling, ownership, rollback or operating documentation.

Client inputs

  • · Access to monitoring and incident history
  • · Existing runbooks, if any
  • · Named operational owners

Hardpoint activities

  • · Readiness review
  • · Observability and failure-mode analysis
  • · Rollback and incident-process review
  • · Ownership and handoff definition

Written deliverables

  • · Readiness findings
  • · Observability review
  • · Failure-mode analysis
  • · Rollback analysis
  • · Incident and runbook review
  • · Ownership and handoff requirements

Engagement boundary

Review and documentation. Hardpoint does not operate the system on your behalf.

Completion condition

Findings and the ownership/handoff requirements are accepted by the operating team.

Likely next phase

Remediation work, or standing evaluation and observability implementation.

Start this engagement Read the practiceOpen the NOC view— publishing soon

Lifecycle

How an engagement progresses.

  1. 01

    Architecture triage

    Thirty minutes, no fee. Which layer is blocking you, and whether Hardpoint is the right practice for it.

  2. 02

    Fixed-scope assessment

    A defined artifact list against a defined environment, quoted as a fixed fee.

  3. 03

    Dated design and acceptance

    Dates, owners, and the acceptance criteria each item has to clear.

  4. 04

    Bounded pilot or implementation

    One domain or one workload set, evaluated before anything is promoted.

  5. 05

    Assurance

    Evaluation, observability, failure handling, rollback and incident process.

  6. 06

    Handoff

    Operating documentation and ownership transfer. This is the exit condition.

  7. 07

    Optional requalification

    Later, if you ask for it. Not a standing retainer.

Not every engagement becomes a ninety-day implementation. Many stop at the assessment, which is a legitimate outcome: you own the findings either way.

When implementation follows

Measure, route, operate — in that order.

Day 0–30

Measure

Instrument what exists. Traffic inventory, knowledge inventory, per-request telemetry, baseline quality scores. No architecture changes yet — you cannot route what you have not counted.

Day 31–60

Route

Introduce classification and routing policy for the highest-volume classes, with failover wired and quality budgets enforced. Changes land behind measurement.

Day 61–90

Operate

Productionize: environments, access, deployment, observability, runbooks, standing evaluation, and the handoff that makes the practice yours.

Terms

What to expect commercially.

How we quote

Assessment and design are fixed fee against a written artifact list. Execution is scoped from the design, so estimates reference a documented environment. Fees depend on environment complexity, number of systems in scope and access constraints — we quote after triage, not before, and fees are not published.

What we do not promise

No guaranteed savings percentage, no vendor scorecards we are paid to produce, no standing engineering bench, no 24×7 managed operations, and no claim that a specific model is best in general. Cost outcomes depend on your workload mix and utilization, and any figure we produce arrives with its assumptions attached.

Start with triage.

Thirty minutes, no fee, no deck. If Hardpoint is not the right practice for the problem, we will say so on the call.

Discuss an Architecture See engagement modelsOr book an architecture triage call