Outcome
State the decision, task, throughput, quality, or service change sought and the current baseline.
Sense AI · Portfolio and readiness
Turn an AI wish list into an evidence-gated investment portfolio.
We compare candidate workflows through operating value, technical feasibility, consequence, foundation readiness, and learning value. Sponsors receive decisions and dependencies, not a catalogue of model demonstrations.
- Scope
- Candidate workflows, shared foundation, data and integration dependencies, operating risk
- Method
- Interviews, workflow evidence, feasibility probes, consequence framing, portfolio gates
- Outcome
- Fund, discover, hold, sequence, or stop with reasons recorded
02
01 · Workflow demand
A candidate starts with a service measure, affected role, and present constraint.
We replace broad goals such as ‘use generative AI’ with an operating claim that can be observed and challenged.
Population
Name the users, case types, languages, channels, and exceptional conditions the claim covers.
Consequence
Identify affected people, money, access, operations, and regulatory or contractual duties.
03
02 · Portfolio scoring
Value and feasibility are incomplete without consequence and shared dependency cost.
The portfolio distinguishes an attractive workflow from one that is ready to consume delivery investment.
Value
Volume, delay, quality loss, avoidable effort, service effect, and strategic relevance.
Feasibility
Source authority, representative cases, integration access, deterministic rules, and model capability.
Consequence
Material error, affected parties, reversibility, oversight duty, and action authority.
Foundation
Identity, network, data, platform, security, evaluation, operations, and ownership readiness.
04
03 · Investment control
Discovery, pilot, operational trial, and production each require different proof.
A gate records the claim, evidence, unresolved dependency, owner, spend envelope, and decision. Stopping is a valid outcome.
Discovery
Prove source access, task framing, representative cases, and plausible technical route.
Pilot, further discovery, or stopBounded pilot
Measure task behavior and controls with constrained users, data, integrations, and no implicit production claim.
Operational trial, repair, or stopOperational trial
Exercise identity, workflow integration, operator intervention, telemetry, support, and change under controlled load.
Production review or repairProduction
Approve acceptance thresholds, controls, known limits, owners, runbooks, rollback, and review triggers.
Release dossier and accountable approval05
04 · Decision view
The review surface shows why each initiative can proceed, must wait, or should stop.
Sponsors see evidence maturity, shared foundation dependencies, risk decisions, investment consumed, and the next decision rather than percentage-complete reporting.
Illustrative reference patternEvidence-gated portfolio
Service knowledgeDiscoverySource authorityProceed
Claims triageBounded pilotException recallMeasure
Access changesHoldAction authorityResolve
Document intakeOperational trialField acceptanceProceed
Comparable state
Every candidate uses the same stage, evidence, dependency, owner, and decision vocabulary.
Shared dependencies
Platform, data, identity, integration, and operating investments are exposed across the portfolio.
Revisable decisions
A held candidate can proceed when its named evidence or dependency changes.
06
05 · Engagement contract
The roadmap names what must be learned, built, approved, and operated.
Mobilisation separates shared foundation work from workflow delivery and establishes the sponsor, product, technology, data, security, risk, and service roles.
Portfolio register
Candidate claims, stage, decision, evidence, dependencies, owner, and next review.
Foundation roadmap
Shared Cloud, identity, data, integration, evaluation, security, and operations capabilities.
Decision cadence
Gate templates, review calendar, decision rights, and stop or escalation routes.
Engagement contract
What must be true, who owns what, and what leaves the engagement.
A good fit when
- Leaders with many competing AI requests
- Teams planning a shared AI platform
- Sponsors deciding whether an existing pilot merits production investment
Required before delivery
- Executive sponsor and portfolio owner
- Process owners available for evidence-based interviews
- Visibility of current platforms, data, integrations, and constraints
Customer owns
- Provide baseline operating and investment information
- Name policy, data, security, and risk authorities
- Make and record funding and risk decisions
Delivery outputs
- Prioritised workflow portfolio
- Readiness and shared dependency assessment
- Sequenced roadmap, investment gates, and governance cadence
Not included by default
- Azure tenant, network, identity, Foundry, Search, monitoring, or platform foundation unless Sense Cloud is included in scope
- A production guarantee based on a prototype or vendor benchmark
- Unrestricted autonomous action or implicit approval authority
- Customer policy, source ownership, or risk acceptance decisions
- A business case based only on generic productivity assumptions
Questions to answer first
- What decision must this portfolio review enable?
- Which shared dependency could invalidate several candidates?
- Who has authority to stop an attractive but unsupported initiative?
The next step is a bounded working session: one workflow, its evidence sources, the decision owner, and the conditions under which the system must stop or hand over.
Assess your AI portfolio