“AI solutions” is a category so broad it means nothing. Here is the useful map: the six families of business AI systems, the honest decision between off-the-shelf and custom, and how to tell which one your problem actually needs.
Every vendor selling anything with a model in it now calls it an AI solution. That makes the phrase useless for deciding — so let’s replace it with a map. Business AI systems cluster into six families, defined by the work they do rather than the technology inside them.
For businesses whose pricing rules and commercial workflows have outgrown spreadsheets: estimating and quoting from your own rate libraries, tender management with controlled revisions, job costing and variation tracking. AI helps compare supplier, material, labour and delivery scenarios against scope and commercial rules — not merely the lowest headline price — while estimators keep judgement and approval.
Obligation registers, inspection checklists, evidence and sign-offs in one place. AI checks records against defined requirements, flags missing evidence and routes exceptions to a qualified reviewer — useful anywhere from mining obligations to environmental reporting. The accountable person stays in the loop by design.
Work orders, field teams, maintenance, contractor records and shift handovers, connected so managers can see what needs doing and who owns the next step. AI earns its place in scheduling and re-planning when priorities change, and in surfacing exceptions before they become expensive.
Lead, pipeline and customer management shaped like the way you actually sell, plus the marketing automation your team genuinely needs. AI brings CRM history, open work and customer signals together to prioritise follow-ups — with people approving customer-facing actions.
Internal assistants answering from your approved business knowledge with source references and permissions — inside Oppermind, or inside your own software through the Developer API; document extraction and classification that moves information between emails, documents and systems; workflow automation with human checks where the decisions matter.
Sometimes the highest-value “AI solution” is not AI at all: connecting your CRM, ERP, finance tools and spreadsheets so people stop being the integration layer, replacing the parts that hold you back, and keeping what still works.
| Approach | When it makes sense |
|---|---|
| Configure what you have | Your existing platform can do the job with a better setup, cleaner data or a simpler process. |
| Connect your tools | Your tools work individually, but handovers and information between them are breaking down. |
| Build custom | Your workflow is specific enough that workarounds and generic software keep limiting the business. |
Custom software carries real obligations: agree ownership, source-code access, data export, licences and handover in the project agreement before work starts. If a vendor won’t discuss those early, that tells you something.
The most common confusion in AI purchasing is treating a workspace platform and a custom project as competitors. They solve different problems:
Many organisations use both: the platform for everyone’s daily work, a project for the one workflow that generic software keeps getting wrong — and some build their own app for the workflow in between.
For the full implementation path — the five steps from first conversation to everyday use — see How to implement AI in your business.
They cluster into six families: estimating and tenders, compliance management, operations and delivery, CRMs built around your sales process, custom AI assistants and document processing, and system integration. The right one depends on where your friction is, not on the technology.
Configure what you have when your existing platform can do the job with a better setup. Connect your tools when handovers break down between them. Build custom only when your workflow is specific enough that generic software keeps limiting the business.
A platform is a ready-to-use AI workspace for everyday work, on a subscription. A project builds software around one organisation’s specific workflow. Many organisations use both.
Ownership, source-code access, data export, third-party licences and handover should be addressed in the project agreement before work starts, so procurement and legal can assess the arrangement up front.
Tell us where work is slow, inconsistent or expensive — we’ll identify the most practical place to start, even when the answer isn’t ours to sell.