Applied Case Study 001

When AI Adoption Still Depends on Human Scaffolding

How the Autonomy Index™ was used to evaluate a real customer adoption workflow inside a VC-backed B2B AI startup, and the post sale operating architecture that was built around it.

Based on firsthand operational experience, 33 account reviews, 30 operational reviews or incident briefs, and a new operating architecture implemented across a portfolio of 36 customer and pilot accounts. The product could automate outbound activity. Handover, readiness, configuration, assurance, recovery and value evidence still depended on people.

  • VC-backed B2B AI startup
  • 36 customer and pilot accounts
  • More than £350k in tracked value
  • 33 account reviews
  • 30 operational reviews or incident briefs
  • February to May 2026

The problem

The AI completed tasks. People carried the workflow.

The product could automate outbound activity, but the operating model around it had not matured at the same speed. Customer Success, customers and Engineering were still holding the workflow together manually.

  1. 01

    Handover

    Human-led

  2. 02

    Readiness

    Human-led

  3. 03

    Configure

    Shared

  4. 04

    Execute

    AI-enabled

  5. 05

    Assure and Recover

    Human-led

  6. 06

    Prove Value

    Human-led

The AI-enabled execution stage was more mature than the handover, readiness, assurance, recovery and value evidence stages surrounding it. That made Workflow Integration the primary constraint.

What was implemented

Diagnose. Structure. Govern. Measure.

A structured post sale operating architecture covering onboarding, activation, health, issue assurance, adoption, value evidence and commercial continuation.

01

Diagnose

  • Mapped the complete customer workflow
  • Reviewed 33 accounts and 30 operational briefs
  • Separated product capability from workflow maturity
02

Structure

  • Six stage post sale lifecycle
  • Five day activation target
  • 25 page onboarding handbook and eight supporting assets
03

Govern

  • RAG customer health scoring
  • Clear issue categorisation, escalation and closure evidence
  • Recurring problems grouped into portfolio level patterns
04

Measure

  • Written success criteria and success plans
  • QBR value evidence
  • Renewal, recovery and pilot to paid pathways

The operational result

Before, and the implemented operating state.

Handover

Before

Context scattered across conversations and individual knowledge

Implemented

Defined lifecycle, inputs, owners and customer readiness conditions

Onboarding

Before

Open ended and account dependent

Implemented

Reusable handbook, readiness checks, account plans and activation target

Customer health

Before

Risk identified through individual judgement

Implemented

RAG view supported by adoption, delivery and commercial indicators

Issue management

Before

Ad hoc reporting and recurring symptoms

Implemented

Categorised issues, reproduction evidence, escalation and validated closure

Value evidence

Before

Reconstructed late in the lifecycle

Implemented

Success criteria, success plans and QBR evidence per account

Commercial risk

Before

Renewal and budget exposure not consistently visible

Implemented

Dates, values, dependencies and recovery actions documented portfolio wide

  • One defect heavy account reached a verified stable QBR baseline.
  • Referral and expansion discussions began.
  • Written success criteria and recovery plans were introduced.
  • Customer validated issue closure was introduced.
  • Material five figure commercial risks were surfaced earlier than existing reporting exposed them.

Assessed baseline

Autonomy Maturity Band: Level 2, Fragmented Pilots

Genuine AI capability, fragmented surrounding workflow.

Implemented state

Operating architecture designed to support movement toward Autonomy Maturity Level 3, Functional Adoption

Defined lifecycle stages, standardised operating assets, structured risk visibility, clearer ownership, stronger issue assurance and customer level value evidence.

Modelled commercial potential

What this operating model was modelled to be worth.

The operational changes were implemented and observed. The longer term commercial effect has been modelled using the assessed portfolio and clearly stated assumptions.

Modelled annualised value
£35k to £70k
Modelled net contribution
£10k to £45k
Modelled net ROI
40% to 180%
Modelled payback
4 to 9 months

Modelling assumptions

  • Assessed portfolio of more than £350k in tracked annual, pilot and contra value
  • Assumed implementation and maintenance envelope of £25k
  • Protected commercial value of £17.5k to £28k
  • Reduced manual rework of £12k to £24k
  • Improved activation and pilot conversion of £7k to £18k

How the conclusion was reached

Evidence, dimensions, constraint, controls, reassessment.

  1. 01

    Evidence collected across 33 account reviews and 30 operational reviews or incident briefs

  2. 02

    Eight formal dimensions evaluated across the four weighted pillars

  3. 03

    Workflow Integration identified as the primary constraint

  4. 04

    Operating controls implemented across the post sale lifecycle

  5. 05

    Reassessment recommended once a full measurement cycle is available

The workflow was scored against the same four weighted pillars and eight formal dimensions used in every Autonomy Index™ assessment.

Leadership & Strategy

20%
  • Leadership Alignment

  • Strategic Integration

Workflow & Operations

30%
  • Workflow Integration

  • Operational Dependency

Behaviour & Trust

30%
  • Behavioural Adoption

  • Operational Trust & Override

Governance & Risk

20%
  • Governance Maturity

  • Risk & Output Assurance

Supporting detail

Methodology and evidence

This applied retrospective case study draws on firsthand operational experience, 33 account reviews, 30 detailed operational reviews or incident briefs, and contemporaneous customer, commercial and product evidence. The assessment reports a maturity band rather than a numerical score because the company ceased trading before a formal post-implementation reassessment could be completed. The company and its customers have been de-identified.

Four weighted pillars: Leadership & Strategy 20%, Workflow & Operations 30%, Behaviour & Trust 30%, Governance & Risk 20%. Autonomy means earning the right to delegate: the system must be embedded enough to rely on and governed enough to rely on safely.

  • Leadership AlignmentDoes the stated AI narrative match operational reality?
  • Strategic IntegrationIs AI connected to objectives, budget and decision making?
  • Workflow IntegrationIs AI embedded in the workflow or bolted alongside it?
  • Operational DependencyWould the workflow degrade if the AI were removed?
  • Behavioural AdoptionDo people use it by default or only when prompted?
  • Operational Trust & OverrideHow often is output overridden, and on what basis?
  • Governance MaturityWho owns the output, and how is it controlled?
  • Risk & Output AssuranceHow are errors detected, escalated and corrected?
Commercial modelling assumptions

The modelled range applies a £25k implementation and maintenance envelope to the assessed portfolio. Components: protected commercial value £17.5k to £28k, reduced manual rework £12k to £24k, improved activation and pilot conversion £7k to £18k. Together these give a modelled gross range of approximately £35k to £70k, a net contribution of £10k to £45k and a net ROI of 40% to 180%.

AI adoption is not the same as operational autonomy.

The Autonomy Index™ identifies where AI is genuinely carrying a workflow, where people are still holding it together, and what must change before greater autonomy can create dependable business value.