nDigital Intelligence
AI-assisted engineering, governed by senior judgement
Governance, review and delivery — three ways to bring AI into your development team safely, each backed by an experienced engineer.
nDigital Intelligence
AI-assisted engineering, governed by senior judgement
Specialist AI agents now handle much of the delivery — but these days anyone can get an AI to write and review code. What sets us apart is judgement earned over 20+ years in the financial and logistics industries: we have learned the hard way what holds up and what hurts down the line, and that simplicity is an art. Our agents are trained over months against the pains real organisations hit, to get it right from the start. They do the work, and a senior engineer personally reviews and signs off every completed item — is the approach simple, is it correct, and what else in your system could the change affect? The agents move fast; the senior review is what makes the result safe to own.
AI-assisted tooling and architect oversight, applied at the points that matter most.
Plan
Goals turned into requirements, user stories and scope.
Design
Architecture decisions for the Microsoft stack.
Build
Features delivered for your review on a regular cadence.
Review
Feedback shapes the next cycle.
Deploy
Working software goes live, and you own the code.
Test
Every change tested, plus a managed quality pass.
Plan
Goals turned into requirements, user stories and scope.
Design
Architecture decisions for the Microsoft stack.
Build
Features delivered for your review on a regular cadence.
Test
Every change tested, plus a managed quality pass.
Deploy
Working software goes live, and you own the code.
Review
Feedback shapes the next cycle.
Requirements analysis
We turn your goals and conversations into clear requirements, user stories and scope — the groundwork done before any code is written.
Solution design
We translate requirements into solution designs and architecture decisions that fit the Microsoft stack and are built to scale.
Development
We implement features against the design and deliver working changes into your project, ready for review on a regular cadence.
Testing & quality checks
Every change is tested and put through a managed quality pass before it reaches you. We stand behind everything we ship.
Service Tiers
From a governance check to fully managed delivery
Five tiers, one ladder. Start where your team is today and move up as your AI adoption matures.
| Tier | Price | Description |
|---|---|---|
| Advisory | R 2,000 / month | AI governance & guardrail advisory |
| Review Intelligence | R 12,000 / month | Nova-powered PR review insights + Review Profile |
| Light | R 45,000 / month | 40 changes/month — full agent delivery + senior sign-off |
| Medium | R 65,000 / month | 50 changes/month — full agent delivery + senior sign-off |
| Heavy | R 85,000 / month | 60 changes/month — full agent delivery + senior sign-off |
All prices exclude VAT. Monthly engagements, cancellable anytime. No fixed-term contracts.
Governance · Advisory
AI Governance & Guardrail Advisory
R2,000
per month · excl. VAT
For organisations beginning their AI adoption journey. Each month:
- Monthly review of 5–10 AI-generated commits or PRs with a written findings report
- POPIA prompt hygiene check — flags any personal information that entered AI prompts
- Developer prompt standards card — specific to your stack, what to include and what never to include
- Monthly one-page governance report for the IT manager or CIO
The entry point into nDigital. Most Advisory clients progress to Review Intelligence or a delivery tier within 6–12 months.
Review Intelligence
Review Intelligence
R12,000
per month · excl. VAT
For teams where AI-generated code volume is growing faster than review bandwidth.
nDigital works with your team to establish a Review Profile — a structured, enforceable representation of your senior reviewer's standards. Every PR your team submits is then met by Nova, nDigital's architect agent, before it reaches the human reviewer. Nova delivers a structured insight report covering:
- What was flagged and why
- Which items match patterns the reviewer has corrected before
- What is genuinely novel and needs human judgment
The reviewer's attention goes to judgment calls, not pattern violations they have already corrected a hundred times.
Setup is handled during the initial engagement — there is no additional onboarding fee.
“Nova meets every PR before you do — so your review is a judgment call, not a pattern hunt.”
Managed Delivery
Architecture & Delivery Assurance
nDigital's specialist AI agents — analyst, developer, tester and architecture — work through the agreed backlog. Every completed work item is reviewed and signed off by a senior engineer before handover.
Light
R 45,000
per month · excl. VAT
- 40 changes per month
- Full agent delivery
- Senior engineer sign-off on every work item
- Architecture assessment included
Medium
R 65,000
per month · excl. VAT
- 50 changes per month
- Full agent delivery
- Senior engineer sign-off on every work item
- Architecture assessment included
Heavy
R 85,000
per month · excl. VAT
- 60 changes per month
- Full agent delivery
- Senior engineer sign-off on every work item
- Architecture assessment included
Monthly engagements, no fixed term, cancellable anytime
nDigital takes responsibility for the output — you own the code
One project per retainer — a second project is a second retainer at no additional onboarding cost
Senior engineer sign-off on every completed work item (delivery tiers)
Standards codified and enforced — the same rulebook for permanent staff, contractors and new hires
Your tech lead stops being the throughput constraint and goes back to design, architecture and mentoring

tether is nDigital's internal orchestration system for managed AI-assisted software development. It is not a product clients buy directly — it is how nDigital delivers. Clients contract with nDigital for development outcomes; nDigital's agents do the work and nDigital's senior engineers review every output before it leaves.
tether agents are purpose-built, not general-purpose. They have been developed and refined over months of continuous improvement through the pains organisations experience adopting AI in delivery — so clients do not have to learn those lessons themselves.
A closer look
What the engagement includes
Managed Delivery tiers (Light, Medium and Heavy) run as Architecture & Delivery Assurance. nDigital's specialist AI agents — analyst, developer, tester and architecture — work through the agreed backlog. Every completed work item is reviewed and signed off by a senior engineer before handover.
The service has two phases.
Phase 1 — Architecture Assessment
nDigital's architecture agent analyses the system and produces a structured document covering technical risks, areas of complexity, cost-saving opportunities and matters for further investigation. Reviewed and contextualised by a senior engineer.
Architectural & technical risks
Areas that may present risk to delivery, stability or maintainability.
Areas of complexity
Components or interactions that require additional attention or careful management.
Cost-saving opportunities
Where efficiency, simplification or rationalisation may reduce cost.
Matters for further investigation
Items that fall outside the available information and may warrant further review.
The output: a clear, structured architecture document that gives stakeholders a shared understanding of the system and a solid foundation for decisions about its future direction.
Phase 2 — Delivery and Review (ongoing)
Agents deliver completed work items. A senior engineer personally reviews each item for simplicity, correctness and wider system impact before handover:
01
Is the approach appropriately simple?
Could the requirement have been met in a clearer or less complex way? Simplicity in delivered software reduces future cost and risk.
02
Has the work been implemented correctly?
Does the completed work address the intended requirement, based on the information and evidence available for review?
03
What could this change affect?
How does the change fit into the broader system, and what could it mean for other modules, systems or business processes over time?
Why this matters
This goes beyond a standard code review. It provides personal technical oversight of every completed work item, with direct attention to the wider architectural picture — and ensures the reviewing engineer builds an informed understanding of the changes being made as your system evolves.
One project, one retainer. A client running two projects carries two retainers — there is no bundling cost and no additional onboarding for a second engagement.
Risk mitigation
Human in the Loop
AI accelerates the work — but no change ships on the say-so of a machine alone. A senior engineer stays accountable for every completed work item, with the authority to question, correct or reject it before it is considered done.
For financial services and other regulated environments, that human checkpoint is not a nicety — it is how automation stays safe, governed and auditable. Every change is reviewed, signed off and owned by a named person, so you always know who stands behind the code running your business.
Who delivers this service
nDigital's specialist AI agents are not general-purpose tools — they are purpose-built for specific stages of the software delivery lifecycle and trained over months against the real pains organisations run into, so the work is right from the start rather than fixed later. Senior review is carried out by engineers with more than 20 years in the financial and logistics industries — domains where mistakes are expensive and simplicity is hard-won. We have learned what works and what hurts in the long run, and every piece of output is reviewed, assessed and taken responsibility for by nDigital.
Why nDigital
The real risks of AI in development teams
Most development teams are already using AI tools informally. The risks are not theoretical — they are accumulating in codebases right now:
Unreviewed AI-generated code entering production
Developers accept suggestions without fully understanding them. The code compiles, tests pass, but the logic is subtly wrong or conflicts with the existing architecture. This is the fastest-growing source of technical debt in teams that have adopted Copilot or ChatGPT informally.
Sensitive data entering public AI models
Developers paste code into ChatGPT to get help. That code often contains connection strings, API keys, internal business logic or customer data structures. Under POPIA, this is a live exposure.
No audit trail or accountability
When a bug surfaces six months later, there is no way to know whether it was AI-generated, who accepted it, or whether it was reviewed.
Inconsistent quality across the team
Different developers, different tools, different prompts, different levels of scepticism. Quality varies by developer temperament rather than by standard.
Architecture drift
AI tools generate code that works in isolation but does not respect wider architectural decisions. Without explicit standards fed into the AI context, generated code creates divergence through hundreds of small, individually harmless deviations.
Compliance and licensing exposure
AI models are trained on open-source code. Generated output may carry licensing obligations the organisation is unaware of.
nDigital exists to solve these risks — from a R2,000/month governance check to a fully managed, senior-reviewed delivery service.
Get In Touch
Tell us what you're building
Need software built, work automated, or the right people hired? Send us a short message about your project and we'll get back to you.
Location
Johannesburg, South Africa