Totus Consulting
Built to stand.
Technology used to be built by teams you could see, at a pace you could review. Now AI builds at machine speed, everything looks finished, and your people are unsettled. We ran technology the old way for thirty years — and we run it the new way every day, on our own company.
Managing developers
Managing AI
Devs who go their own way
Agents that go their own way
Code review before merge
Approval before anything ships
Scope creep
Confident, wrong answers
Growing your juniors
Repurposing them — towards judgement and design
Same discipline. New team. Your people still on it.
Three things we do
Someone is selling you a system, or building one for you, and you need to know whether it will work. We check it. If it has problems, we help fix them. And once it runs, we can run it with you or for you. Every service below is one of those three.
You are about to buy or build something. We look at it and tell you, in plain words, whether it will work and what it would cost to put right.
Example: a vendor pitches an AI system; before you sign, we spend a few days on it and give you a written answer.
The check found problems. We help you fix them — with a plan, and with our hands if you need them.
Example: the data model cannot carry the growth you plan; we redesign it and build the change.
It works. Now someone has to run it every month. We can do that with your team, or for you.
Example: you subscribe to Totus FIQ and we operate it — month-end, exceptions, master data.
Where most engagements start
Someone is pitching you a system, or asking you to fund a build. We tell you whether it will work — and what it would cost to put right.
We publish the questions we use, so you can ask them yourself — on any vendor, on any internal team, and on us. The review is what happens when you would rather we asked them, on the system in front of you, with the evidence written down.
What you get
A written report structured on the questions. Every finding carries its evidence — what we looked at, and what we found. A verdict you can take to a board, and a repair order where there is one.
How it is priced
Priced by the job, agreed before we start — so the number does not move once we are in it. The first conversation costs nothing.
Why us
Thirty-five years of building the systems we are asked to judge — and we still build one every day. We hand out the questions that would expose a facade, including on ourselves.
Architecture & Design
Building, integration, migration. The blueprint as its own deliverable.
Bespoke Solutions
Architecture, development, integration, training — end to end, per phase.
AI Implementation
We help you run the business with people and AI agents working together: which work moves first, one written company memory, clear rules, client data protected, and your team led through the change. It is the business and the way of running it, not only the tools.
How we run Totus →Because the checker also sells the fixing
Every finding in our report carries its evidence, and the report is written so any competent team can act on it — ours or anyone else’s. Choosing us for the repair is a choice, not a funnel. Saying what is wrong is the easy half; guiding the repair is the hard half, and because we build, we can stay for it.
Two distinct services, both monthly: one operates the platform, the other operates the technology function around it. Take either, or both.
New · three modes, one ladder
You do not have to run it yourself.
You subscribe to Totus FIQ; we operate it with you — exceptions, month-end support, chart of accounts and master-data upkeep, and your team supported while they do the parts only they can do.
Every mode starts the same way: onboarding and training — then you choose how it’s run.
Self-managed
Today’s default
Your team runs FIQ. We are product support behind them.
Co-managed
We operate alongside your team — exceptions, month-end, master data. Your team keeps the judgement.
Fully managed
Complex organisations
We run it. You consume the answers.
The more complex the organisation, the stronger the case for moving up the ladder — and a client can step back down as their own team matures. Priced by the job, agreed before we start; the mode is a conversation, not a tier.
The technology function
The ongoing rental of judgement — quarterly counsel, or days per month. An advisor at your board’s table, not a director on your register.
Distinct from a design project: this one has no end date.
Where Managed Services operates the platform, this operates the function around it — what to build next, what to stop, who steers the architecture, and which of the questions your own systems would fail.
71.8
Programming’s automation-feasibility score — second of 35 skills. Senior architects judging trade-offs are among the least exposed.
The AI Skills Shift · arXiv · 2026
42%
of companies had abandoned most of their AI initiatives — up from 17% a year earlier. The cost of building wrong arrives late, and as a write-off.
S&P Global Market Intelligence · March 2025
12.3%
of changed lines are now copy-paste, up from 8.3% since AI assistants arrived. Refactoring collapsed from 25% to under 10%.
GitClear · 211M lines · 2020–2024
45%
of AI-generated code failed security tests — flat regardless of model size. The constraint is the person judging what the model produces.
Veracode · 100+ LLMs · 2025
We work with ZatoCento, a Johannesburg firm for digital transformation delivery and programme management, on larger programmes. zatocento.co.za
Our own practice runs this way
A two-founder company running a team of AI agents — every change proposed by an agent, approved by a person, audited after. The practice’s proof is the product’s development.
How we run Totus →
A scoped review, priced by the job, agreed before we start.
Talk to technical counsel →Runs on AWS Agents built on Claude Member of NVIDIA Inception
