AI inventory: where a company already using AI unknowingly should start
The AI Act's first step is not legal — it is finding what is already running inside the company.
Article to be written AI Transformation · Optimization · Governance
Where does your operation lose time, money and decisions?
We find those processes and build the intelligent systems that automate, optimise and govern them. We sell measurable operational improvement — not “an AI agent”.
Discover → Automate → Optimize → Govern → Measure
Logistics & Transport
Routing, fleets, ETA, maintenance, documents and operational control.
SMEs & Back office
Finance, documents, email, reporting, purchasing, support and workflows.
Data-intensive operations
Plenty of data, little automation, little decision capacity.
01 Positioning
AI only counts when it shows up in cost, time and decision quality.
01
Administrative work that burns whole teams on repeatable tasks — documents, email, data entry, reconciliation.
02
Late decisions made blind: routing, resource allocation, priorities, capacity — decided by habit instead of by model.
03
AI entering the company uncontrolled: tools adopted team by team, with no inventory, no owner, no risk classification and no AI Act evidence.
02 Methodology
One five-stage path, every time.
The AI Transformation Framework is our intellectual property: reusable with any client, auditable at any stage.
01
Discover
- Processes
- Resources
- Data & systems
- Cost & cycle time
- Bottlenecks
02
Automate
- Agents
- Workflows
- Integrations
- Document processing
- Automated actions
03
Optimize
- Machine learning
- Mathematical optimisation
- Digital twin
- Simulation
- Resource allocation
04
Govern
- AI Act & GDPR
- Security
- Human oversight
- Auditability
- AI literacy
05
Measure
- ROI & cost
- Productivity
- Quality & SLA
- Risk
- Adoption
No stage is optional: without Discover there is no business case, without Govern no system survives an audit, and without Measure there is no way to prove it was worth it.
03 Services
Four services, one relationship that grows.
It starts with a closed, low-risk diagnostic. It ends with a managed, governed AI operation.
S1
AI Transformation Assessment
Entry pointA diagnostic of processes, people, systems, data, cost, cycle time and bottlenecks, run with the AI Process Scanner.
Deliverable — AI Opportunity Map: per process, the current problem, the solution, the technology, expected impact, complexity, risk, estimated ROI and human-in-the-loop needs.
S2
Implementation
ProjectBuilding whatever the map prioritised: agents, automations, integrations into existing systems, optimisation models, dashboards and workflows.
Deliverable — systems in production, wired to real data and able to take action, not just suggest it.
S3
AI Governance Setup
ComplianceInventory of AI systems, risk classification, policies, documentation, human oversight and logging — structured preparation for the AI Act and GDPR.
Deliverable — AI Registry, risk register, policies and a remediation plan with dates and owners.
S4
Managed AI & Governance
RecurringContinuous operation: monitoring, model cost, performance, risk, agent improvement and compliance follow-up as regulation evolves.
Deliverable — a periodic report of value created versus total cost of the solution.
Assessment→ Business Case→ Implementation→ Governance→ Managed Service
04 Products
Every project funds a reusable asset.
Consulting produces methodology; methodology produces software.
AI Process Scanner
A methodology — and next, software — to find where AI creates the most impact. It evolves into a digital twin of how the company actually runs.
In use in assessmentsDocument & Email AI Agent
Ingestion, understanding, classification, extraction, filing and action-taking over documents and email — with human-in-the-loop whenever confidence is insufficient.
MVP / PoCAI Transformation & Governance Platform
Scanner, Opportunity Map, AI Registry, agents, integrations, risk, cost, performance, ROI and compliance evidence in one place.
Roadmap 2027–2028 05 AI Act
Governance is a layer, not an audit bolted on at the end.
For every AI system we keep the owner, business purpose, models, data sources, personal data, decision authority, human oversight, affected people, risk classification, logging, testing, known limitations and the next review date.
Precise wording
i9.ai performs a private maturity assessment — AI Act Readiness Assessed — and prepares the evidence. We issue no EU certification, no CE marking, and we do not act as a notified body. Where an external conformity assessment is required, we work with legal partners and competent bodies, keeping consulting and certification independent.
Open resources on the AI Act in Portugal: ai-act-portugal.com →
1 · AI Discovery2 · AI Inventory3 · Risk Classification 4 · Gap Analysis5 · AI Risk Register6 · Governance Policies 7 · AI Literacy8 · Remediation Plan9 · Continuous Monitoring
06 Outcomes
What we measure on every project.
01
Less manual work
Hours returned to the team.
02
Fewer errors
Rework and correction rate.
03
Faster processes
End-to-end cycle time.
04
Better resource use
Utilisation, routing, capacity.
05
Data-backed decisions
From habit to model.
06
Compliance and control
Evidence ready for audit.
07
Measurable ROI
Value created vs. total cost.
08
Real adoption
Actual use, not licences.
07 Insights
We write about what we apply.
Route optimisation is not an AI problem — it is a data problem
Where mathematical optimisation still beats an LLM, and where the two combine.
Article to be writtenWhat processing a document by hand actually costs
A simple method to work out cost per document before deciding to automate.
Article to be written 08 Contact
Start with one process. Just one.
Tell us where it hurts most. We come back with an initial read of the process, its automation potential and what it would take to measure it.