Why now?
We look for a real trigger: growth, cost, capacity, customer pressure, compliance or recurring process friction.
Fresh Kiwi · AI consultancy
Start small. Prove value fast. Scale what works.
A complete Efficiency AI Audit at an 80-person organisation, from first conversation to working pilot. Modelled figures, not client results.
Method
From bottleneck to demonstrable result
Understand and prioritise first. Build only after that.
01DiscoverDiscover · Qualify · Map
02PrioritiseScore · Prioritise
03PilotPilot · Measure
04OperateDeploy · Monitor
05ScaleExpand · Reassess
- 01Discover
- 02Qualify
- 03Map
- 04Score
- 05Prioritise
- 06Pilot
- 07Measure
- 08Deploy
- 09Monitor
- 10Expand
- 11Reassess
Overview
The organisation
| Attribute | At the start |
|---|---|
| Sector | Business services |
| Employees | 80 |
| Departments | Sales, Operations, Finance, HR, Customer Service, IT |
| Current AI | Copilot + disconnected public AI tools |
| Goal | Grow without proportional headcount growth |
Overview
The ladder: from cost to revenue
What follows works through the Efficiency AI Audit in detail. That is the first rung on the ladder. Capability AI and Opportunity AI follow the same three lines, but with their own kind of evidence; the scoring method below is specific to the audit.
| Cost line | Capability line | Revenue line |
|---|---|---|
| Efficiency AI Audit: the unit is the task (existing, recurring work) | Capability AI: the unit is the capability gap (demand you're already turning away or rationing) | Opportunity AI: the unit is the opportunity hypothesis (an offer that doesn't exist yet) |
Capabilities
0. Qualification: before the audit starts
Is there a sponsor?
Someone must be able to organise access, set priorities and make decisions.
Is there scale?
A task performed by one person for ten minutes a month is rarely interesting. We look for repetition, volume and impact.
Is there budget willingness?
No purchase commitment is required, but there should be willingness to invest if the business case is strong.
Is the environment workable?
We test systems, data, access, security and feasibility at a high level without designing solutions yet.
What does success mean?
We agree what should be measurably better within six to twelve months.
Capabilities
1. Discovery: understand before automating
Business goals
Which three goals matter this year and where does operational pressure get in the way?
Headcount pressure
Which teams are asking for more people and what would those new employees actually spend time doing?
Process friction
Where do we see waiting, duplicate work, correction loops, copy-paste and recurring reporting?
System landscape
We inventory Microsoft 365, CRM, ERP, HR, support, storage and existing integrations.
AI maturity
Which tools and pilots already exist and where could Shadow AI be developing?
Security & governance
Which data is sensitive, who may access what and where must approvals remain mandatory?
Method
2. From 80 employees to an opportunity portfolio
Not everybody needs a long interview. We combine surveys, workshops and targeted validation.
01Collect80 employees receive task survey · ~400 task observations collected
02ClusterDuplicate observations clustered · ~63 unique tasks validated
03AssessValue and time quantified · AI suitability tested · Technical feasibility assessed · Security/privacy/compliance scored
04RankTop opportunities ranked
05Substantiate5 business cases developed
- 0180 employees receive task survey
- 02~400 task observations collected
- 03Duplicate observations clustered
- 04~63 unique tasks validated
- 05Value and time quantified
- 06AI suitability tested
- 07Technical feasibility assessed
- 08Security/privacy/compliance scored
- 09Top opportunities ranked
- 105 business cases developed
Efficiency AI Audit
3. From inventory to decision
This is how a broad inventory narrows to a manageable roadmap.
63unique tasks
8Do Now
17Next
21Explore
17Don’t Automate
AI Opportunity Portfolio63 opportunities
VALUEFEASIBILITY
8DO NOW
17NEXT
21EXPLORE
17DON'T AUTOMATE
Overview
4. Select the first pilot
Values are calculated at a fully loaded hourly rate. Time freed is capacity first, not money by itself. Before the pilot we record a baseline and compare it afterwards against time tracking, plus a sample check on quality.
| Opportunity | Before | After | Estimated value |
|---|---|---|---|
| Weekly reporting | 8 hrs/week | <30 min/week | ~€32k/year |
| Lead research | 5 hrs/week | ~1 hr/week | ~€28k/year |
| Invoice processing | Heavy manual work | Automated extraction + review | ~€41k/year |
Perspective
5. Prove value fast
We implement one low-complexity, highly measurable opportunity first. Baseline, KPIs, errors, time and user experience are agreed before implementation so the value review is evidence-based.
Perspective
6. Scale what works
After value is proven, we move to the next priority in the same portfolio. Existing Fresh Kiwi capabilities are configured where possible instead of being rebuilt.
Perspective
7. Managed AI & reassessment
After go-live we monitor usage, cost, quality and risk. We periodically reassess opportunities as processes, data and model capabilities change.
Next step