Services
AI & Systems Opportunity Review
Know what to improve first, before you commit to a build.
The entry point for most engagements
A structured diagnostic that identifies the next useful operating improvement, from better spreadsheets and data habits through to apps, dashboards, automations, and AI workflows. You leave with a Diagnostic Findings Pack, a Staged Value Recommendation, and a First-90-Day Plan: a prioritised view of what to improve, in what order, and why.
Duration: 1–3 weeks. Format: fixed-scope diagnostic. Pricing on enquiry.
Who it is for
If you know work is slow, manual, risky, or hard to see, but not whether the answer is a better spreadsheet, cleaner data, an app, a dashboard, an automation, or an AI workflow, that is exactly the question the diagnostic answers.
Why start here
The right first step is not always an app. Sometimes the next useful improvement is a structured workbook, a confirmed data model, or an operating rhythm that makes ownership and decisions clearer.
What you receive
Three deliverables, in plain language, owned by you whether or not Salix AI does the next stage of work:
- Diagnostic Findings Pack
- Staged Value Recommendation
- First-90-Day Plan
Deeper reviews
Where the review surfaces deeper questions, such as the maintainability of an estate of apps and workflows built on Microsoft’s Power Platform or whether an AI workflow can be trusted in production, that deeper review is scoped as part of the recommendation rather than sold as a separate product.
AI-Native App Proof of Concept (POC)
Test a real workflow with real users, then decide.
A working prototype and a clear decision
A focused working prototype that tests whether a sufficiently clear business workflow should become a production system. Within 2–4 weeks you have a working prototype in front of a small group of real users, and a clear, evidenced decision: stop, iterate, build for production, or take a different route entirely.
Duration: 2–4 weeks. Format: fixed-scope proof of concept. Pricing on enquiry.
Why a POC
This is not a throwaway demo. The POC is a learning asset: it tests workflow fit, data assumptions, user behaviour, and architecture choices, and it documents exactly what would be required before production use. AI-assisted delivery means the prototype takes weeks rather than months, but the value is still in the problem framing, the architecture, the user feedback, and the decision at the end.
What you get
Business Intelligence helps organisations make informed decisions by turning raw data into useful information. Here are some real-world examples of how BI can benefit your business:
- A working prototype or proof of concept
- POC scope note and user flow
- Architecture notes and stack rationale
- Test notes and known limitations
- A production-readiness gap list
- A recommendation: stop, iterate, production build, or alternative route
- Where AI is part of the workflow: first-pass eval examples, trace considerations, and review criteria
When a POC is not the first step
A POC is also not the default first step. Where the process, data, or adoption readiness is still uncertain, we will say so and recommend an earlier rung of the ladder first: a structured workbook, a dashboard, an operating rhythm, or an Operational DNA phase. A prototype built on an unconfirmed process just proves the wrong thing faster.
If a workflow looks ready to test, the AI & Systems Opportunity Review confirms the POC is the right next step and defines the slice worth building.
Before production
A POC is not production software unless it is explicitly scoped and assured as such. Before operational use, a system typically needs a proper repository and deployment process, data model and access-control review, structured testing, senior code review, and, where AI acts in the workflow, a private eval standard and human review rules. The POC names these gaps; it does not silently skip them.
Where AI carries part of the workflow, the POC also defines what makes it testable: the examples, edge cases, expert standards, traces, and private evals that let quality be measured against your business rather than a public benchmark.
Operational DNA
Your organisation’s way of working, made usable in everyday work.
Operational DNA Assessment
Operational DNA helps organisations turn scattered knowledge, templates, processes, and decision habits into reusable AI-assisted workflows. The Operational DNA Assessment maps where your organisation’s knowledge actually lives and which repeated workflows would benefit most from it.
Duration: 1–2 weeks. Format: fixed-scope diagnostic, scoped around one first use case. Pricing on enquiry.
Why it matters
Most organisations already have operational DNA. It just lives in key people’s heads, old proposals, chat threads, and shared-drive folders. It shows up in the work that repeats: proposals, onboarding, handovers, status reporting, decisions, quality review. When that knowledge is structured and governed, AI-assisted work becomes reliable rather than generic; when it is not, every repeated task starts from a blank page and depends on whoever remembers.
The four layers
The assessment covers all four layers of Operational DNA:
- Knowledge layer – the structured source of truth: services, processes, templates, project context, decisions, lessons learned
- Behaviour layer – the prompts, commands, and skills that encode how the organisation should think and act
- Tool layer – the actions an AI workbench can take: create documents, update trackers, prepare packs, query systems
- Governance layer – the guardrails: source citations, approval steps, permissions, quality checks, audit trails
Built out over time, those layers become an operating system unique to your business: a working layer of knowledge, workflows, and governance that no competitor can buy off the shelf.
What you receive
You leave with an operational knowledge map, a shortlist of high-value repeated workflows, knowledge-risk notes, and a recommended first use case. The first buildout focuses on the Knowledge layer, with project management as the default proof case: it brings together context, decisions, actions, reporting, and handover, so the value is visible quickly.
Operational DNA is not a chatbot, a generic knowledge base, or a once-off prompt pack. It is your organisation’s way of working made usable in everyday work.
After the assessment
The Operational DNA Buildout is the practical buildout that follows the assessment: the knowledge base, project-management workflows, prompts, commands, and governance rules that make repeated work easier to start, check, hand over, and improve. The first buildout focuses on the Knowledge layer.
Business Systems
AI makes software easier to build. It does not make business systems easier to operate.
Built for real operations
A system people actually use: built for real operations, with documentation, QA, deployment discipline, and maintainability from day one, not a prototype quietly promoted into production. Every build starts with the problem confirmed.
Production business apps
A custom, Microsoft-platform, or hybrid application built for real operations. The result is a reliable operational system with support arrangements, assurance evidence, and documentation, including apps and workflows built on Microsoft’s Power Platform where they fit.
Automation, dashboards, and reporting
Delivered as stages in their own right where the main gap is repeatable manual work or visibility, not as afterthoughts bolted onto an app.
We operate what we build
Systems Salix AI delivers are looked after once they are live: monitored on a regular rhythm, patched and improved through disciplined releases, and supported by the people who built them. For AI workflows, that includes private evals built from your own real cases, trace review, and testing of every prompt or model change before it reaches production, so the system keeps behaving the way it did when it was accepted. How that ongoing care works is agreed as part of any build we recommend.
After go-live
Launch is not the finish line. Business systems need monitoring, small improvements, user support, release discipline, and governance. AI-enabled workflows need more: eval regression review, trace review, prompt and model change testing, and clear human escalation rules. Knowledge assets decay if nobody owns them.
If you are running a system nobody looks after, an AI & Systems Opportunity Review establishes its current state and the path to supported operation.
