AI strategy and readiness
A readiness assessment, a short list of use cases scored on value and risk, and a practical roadmap your leadership can approve.
AI Solutions
AI consulting and implementation, built into the systems you already run
Most organisations do not need more AI experiments. They need a few AI tools that do real work inside the systems they already run, with their data protected and people in control. We help you choose those use cases, build them properly and keep them secure.
Problems we help solve
- Staff already use AI, but nobody has set the rules. People paste work into free chatbots, and leadership cannot say what data has left the organisation.
- Copilot licences are bought, but adoption is slow. Answers are generic, or Copilot surfaces files people should never have seen.
- Answers are buried in documents. Policies, procedures, contracts and past project files hold the answers, but finding them takes hours.
- Routine work eats skilled people's time. Invoices, forms, emails and requests are read, keyed in and routed by hand.
- Pilots that never reach production. A demo impressed everyone, but it was never connected to real systems, secured or supported.
Our AI services
Microsoft 365 Copilot
Permissions, sensitivity labels and data governance put in order first, then a measured rollout with training and adoption support.
Generative AI development
Custom assistants and applications that answer from your approved documents and data, cite their sources and respect permissions.
Private AI and self-hosted models
Open-weight models such as Llama, Qwen, DeepSeek, GLM and Mistral on GPU servers you control, so prompts and documents are never sent to an outside AI provider.
AI agents
Agents that carry out routine steps in a business process across your systems, with approval points and full logging.
Document processing and automation
Invoices, forms, contracts and requests read, checked and routed into your ERP, CRM or workflow, with people reviewing exceptions.
AI governance and security
AI use policies, risk assessments, shadow AI discovery, security testing of AI applications and controls that satisfy auditors and regulators.
Where AI helps first
The best first projects are frequent, well understood tasks where a person can easily check the result. Typical starting points by team:
| Team | Where AI helps | What stays with people |
|---|---|---|
| Finance | Reading supplier invoices and receipts, matching transactions, drafting variance commentary | Approving payments and posting adjustments |
| Operations | Demand and stock forecasts in Prometheus ERP, summarising exceptions, drafting purchase suggestions | Committing orders and changing plans |
| Customer service | Drafting email replies from your knowledge base, summarising case history for the next person | Sending replies and making exceptions |
| HR and administration | Answering staff questions from approved policies, drafting job descriptions and letters | Decisions about people |
| IT | Triaging and routing tickets, summarising incidents, answering how-to questions from your documentation | Changes to systems and access |
| Leadership | Summaries of reports, meetings and long documents | Judgement and decisions |
We are just as clear about where AI is the wrong tool. If a fixed rule, a report or a workflow automation solves the problem more reliably and for less, we will say so.
How we deliver AI projects
- Assess. We review your goals, systems, data and current AI use, and score candidate use cases on value, data readiness, risk and cost.
- Prove. We test the most promising use case on your own data, against accuracy and cost thresholds agreed in advance, before you commit to a full build.
- Build. We integrate the solution with your systems, apply permissions and logging, and design the points where people review or approve.
- Launch. We train the people who use it, roll it out in stages and monitor accuracy, adoption and running costs.
- Govern and improve. Policies, risk reviews and monitoring keep the solution safe and useful as models, vendors and your needs change.
Every engagement runs to a written scope with deliverables, acceptance criteria and costs in naira (or US dollars where a licence or service is priced in dollars).
AI with people in control
AI systems handle information that is often personal or confidential, so governance is part of the design from the first day:
- Your data stays yours. Depending on how sensitive it is, we use enterprise AI services whose terms exclude your content from model training (with zero data retention where the provider offers it), models deployed in your own cloud account, or open-weight models on servers you control, where nothing is sent to an outside AI provider. We document where data is processed and stored. There is no major public cloud region in Nigeria and most hosted AI services process data abroad, so where data location matters we either keep AI processing in Nigeria, with self-hosted models on your premises or in a Nigerian data centre, or use regions abroad, such as South Africa or Europe, with a lawful transfer basis under the NDPA and any sector rules in view.
- People decide what matters. Outputs that affect individuals, money or compliance are reviewed by a person, and every solution has a manual fallback, so work continues if the AI service or the connection to it is unavailable.
- Answers you can check. Assistants show the documents they used, and only use documents the person asking is allowed to see.
- Measured, not promised. Accuracy is tested against your own examples before launch and monitored after it.
- Data protection law applies to AI. The Nigeria Data Protection Act 2023 (NDPA) applies to personal data used in AI systems, including people's right not to be subject to decisions based solely on automated processing that significantly affect them. We design with it in mind and work alongside your legal and data protection advisers; our guide to AI and the NDPA explains the main points. This is general information, not legal advice.
Why work with Promatics on AI
AI projects rarely fail because of the model. They fail on permissions, messy data, missing integrations, weak security and nobody owning the system after launch. Those are the things we already do every day as a managed IT, cybersecurity and business systems partner.
- One accountable partner. The same team that manages your Microsoft 365, identity, security and ERP builds and supports your AI solutions.
- Built into your systems. We connect AI to Microsoft 365, Prometheus and ERPNext, CRMs and line-of-business applications through supported interfaces.
- Vendor-neutral advice. We work with Microsoft, Amazon Web Services, Google Cloud, OpenAI, Anthropic and open-weight models such as Llama, Qwen, DeepSeek, GLM and Mistral, and recommend what fits your data, security and budget. Naming a product does not mean we are a partner of its vendor.
- Honest scoping. We tell you when AI is not the answer, and we stop a project that does not meet its agreed thresholds.
Common questions
Where should we start? With a short readiness assessment or a single, contained use case, such as answering staff questions from an approved policy library or reading supplier invoices into your ERP. Early, measurable success builds the confidence and governance for harder cases.
Is our data used to train public AI models? It should not be, and we design to prevent it. We confirm each vendor's data handling terms with you before any real data is used. Our guide to whether AI tools train on your data explains what business terms say.
Can AI run on our own servers, so our data never leaves? Yes. We deploy open-weight models on GPU servers in your own cloud account, such as AWS, Microsoft Azure, Google Cloud or DigitalOcean, or on hardware in your data centre, and connect them to your documents and applications. Prompts, files and outputs stay in your environment. See private AI and self-hosted models and our comparison of self-hosted models and AI APIs.
Do we need Microsoft 365 Copilot, or something custom? Often both. Copilot works well for everyday productivity inside Microsoft 365. A custom assistant or agent makes sense when you need answers from specific systems, tighter control or a process that Copilot does not cover. Our comparison of business AI assistants explains the trade-offs.
What does an AI project cost? It depends on the use case, your data and the systems involved. Our guide to what an AI project costs explains the cost drivers. Every proposal sets out costs in naira (or US dollars where a licence or service is priced in dollars) before you commit, with dollar-priced AI and cloud subscriptions shown separately, since exchange-rate movements affect what they cost to run.
Guides for planning AI
AI readiness checklist
A practical worksheet to check strategy, data, permissions, security, governance and skills before you invest in AI, and to score readiness honestly.
What is agentic AI?
What AI agents are, how they differ from chatbots and automation, where they help in the back office and which controls they need.
What an AI project costs
The one-time and ongoing costs of an AI project, what makes each bigger or smaller, and how to budget a pilot and production.
Where could AI save your team the most time?
Tell us about the work, the systems and the data involved. We will reply to arrange a conversation and suggest a sensible first step, often a short readiness assessment or a proof of concept on your own data.