AI readiness checklist
A scored worksheet for strategy, data, permissions, security, governance and skills before you invest in AI.
AI Solutions
AI consulting and implementation, built into the systems you already run
Not more experiments: a few AI tools that do real work inside your systems, with your data protected and people in control. We help you choose them, build them properly and keep them secure.
Problems we help solve
AI with no rules
Staff paste work into free chatbots; nobody knows what data left.
Copilot, little adoption
Generic answers, or files people should never see.
Answers buried in documents
Policies, procedures, contracts and project files take hours to search.
Routine work by hand
Skilled people keying and routing invoices, forms, emails, requests.
Pilots stuck at demo
Never connected to real systems, secured or supported.
Our AI services
AI strategy and readiness
Start here: readiness assessment, use cases scored on value and risk, a roadmap to approve.
Microsoft 365 Copilot
Permissions, sensitivity labels and data governance first; measured rollout, training, adoption support.
Generative AI development
Assistants and apps answering from your approved data, citing sources, respecting permissions.
Private AI and self-hosted models
Open-weight models (Llama, Qwen, DeepSeek, GLM, Mistral) on GPU servers you control.
AI agents
Routine process steps across systems, with approval points and full logging.
Document processing
Invoices, forms, contracts and requests read, checked and routed into ERP, CRM or workflow; people review exceptions.
AI governance and security
Use policies, risk assessments, shadow AI discovery, security testing, controls for auditors and regulators.
Where AI helps first
Start with a readiness assessment or one contained use case: frequent, well-understood work a person can easily check, such as staff questions from an approved policy library or supplier invoices into your ERP. Early, measurable wins build confidence and governance.
Examples by team
| Team | AI helps with | People keep |
|---|---|---|
| Finance | Reading invoices and receipts, matching transactions, drafting variance commentary | Approving payments, posting adjustments |
| Operations | Demand and stock forecasts in Prometheus ERP, exception summaries, draft purchase suggestions | Committing orders, changing plans |
| Customer service | Reply drafts from your knowledge base, case history summaries | Sending replies, making exceptions |
| HR and administration | Answers from approved policies, draft job descriptions and letters | Decisions about people |
| IT | Ticket triage and routing, incident summaries, how-to answers from your documentation | Changes to systems and access |
| Leadership | Summaries of reports, meetings and long documents | Judgement and decisions |
Not every problem needs AI. If a fixed rule, a report or workflow automation does it better for less, we say so.
How we deliver AI projects
- AssessGoals, systems, data, AI use; use cases scored on value, readiness, risk, cost.
- ProveOn your data, against accuracy and cost thresholds agreed first.
- BuildIntegrated, with permissions, logging and human review points.
- LaunchTraining, staged rollout; accuracy, adoption and running costs monitored.
- Govern, improvePolicies, risk reviews, monitoring as models, vendors and needs change.
Every engagement has a written scope: deliverables, acceptance criteria, costs in naira (or US dollars where a licence or service is priced in dollars).
AI with people in control
Governance is designed in from day one, because AI handles personal and confidential information. Your data stays yours:
| Option | What it means |
|---|---|
| Enterprise AI services | Terms exclude your content from training; zero data retention where offered |
| Your cloud account | Models deployed in your own account |
| Private models | Open-weight models on GPU servers you control (AWS, Azure, Google Cloud, DigitalOcean) or on your own hardware, connected to your documents and applications; prompts, files and outputs never go to an outside AI provider (compared with AI APIs) |
We match the option to the data's sensitivity, confirm vendor data terms with you before real data is used (does AI train on your data?), and document where data is processed and stored. Nigeria has no major public cloud region and most hosted AI services process data abroad. Where location matters, we keep AI processing in Nigeria (self-hosted models on your premises or in a Nigerian data centre) or use regions such as South Africa or Europe, with a lawful NDPA transfer basis and any sector rules in view.
People decide
Outputs affecting individuals, money or compliance are reviewed; a manual fallback keeps work going if AI is unavailable.
Checkable answers
Sources shown, drawn only from documents the asker may see.
Measured
Accuracy tested on your examples before launch, monitored after.
Data protection law applies to AI. The Nigeria Data Protection Act 2023 (NDPA) covers personal data in AI systems, including people's right not to be subject to decisions based solely on automated processing that significantly affect them. We design for it with your legal and data protection advisers; see AI and the NDPA. General information, not legal advice.
Why Promatics for AI
AI projects rarely fail on the model, but on permissions, messy data, integrations, security and ownership after launch: our daily work in managed IT, cybersecurity and business systems.
One accountable partner
The team running your Microsoft 365, identity, security and ERP builds and supports your AI.
Built into your systems
Microsoft 365, Prometheus and ERPNext, CRMs, business apps, via supported interfaces.
Vendor-neutral
Microsoft, AWS, Google Cloud, OpenAI, Anthropic or open-weight: what fits your data, security, budget.
Honest scoping
We say when AI is not the answer, and stop projects that miss agreed thresholds.
Naming a product does not mean we partner with its vendor.
Common questions
Copilot, or something custom? Often both: Copilot for everyday Microsoft 365 work; a custom assistant or agent for specific systems, tighter control or processes Copilot does not cover. Compare AI assistants.
What does it cost? It depends on the use case, data and systems; our cost guide explains the drivers. Proposals set out costs before you commit, with dollar-priced AI and cloud subscriptions shown separately, since exchange rates affect what they cost to run.
Guides for planning AI
What is agentic AI?
What AI agents are, how they differ from chatbots, where they help and the controls needed.
What an AI project costs
The one-time and running costs of AI projects, their drivers, and budgeting pilot versus production.
Where could AI save your team the most time?
Tell us the work, systems and data involved. We will arrange a conversation and suggest a first step, often a readiness assessment or proof of concept.