Data cleaning · Data engineering · AI
The data team your business is missing.
We clean up your messy data and engineer it into one reliable foundation, on your own servers, so everyone works from numbers they can trust. Then we build the software that captures it and the AI agents that use it.
Example client
Your data foundation
- Systems connectedERP · CRM · Billing
- Duplicate customers merged3,412
- Sales matches the ledgerTo the cent
- Quality checks passing126 of 126
- Numbers refreshed15 min ago
Now it can carry dashboards, forecasts and AI agents.
Illustrative figures.
01The problem
You don’t have a data shortage. You have a data mess.
- Month-end means rebuilding the numbers by hand.
- Sales and finance quote different figures.
- Nobody trusts the dashboard.
A new dashboard won’t fix that. The data underneath will.
Not sure how messy your data is?
Take the Business Clarity Check. Five questions, about two minutes, and you see your observations straight away. No email needed.
Take the 2-minute checkYour management team is discussing performance. What usually happens?
- Everyone agrees on the numbers
- Different departments bring different numbers
- We spend more time debating data than deciding
02How we work
Clean before you engineer
Capture, clean, transform, verify. Then BI and AI go on top.
Your data today
- ERP
- CRM
- Billing
- Excel
Clean
- ACME PTY LTD
- Acme Hldgs
- Acme Holdings
- 04/03/24 → 2024-03-04
- Deduplicate
- Standardise
- Reconcile
Engineer
- Pipelines
- Data model
- Warehouse
BI and AI
- Dashboards
- Forecasts
- AI agents
Data first. Software and AI on either side.
We build it brick by brick. If your systems can’t capture the data, we build them. Then we clean and engineer the data into a foundation. If your foundation is solid, we build AI agents that work with it.
Skip a layer and everything above it wobbles. That’s why we start with the data.
Softwarethat captures the data, if your systems can’t
Clean datathe core of our work
Foundationpipelines and one data model
BI and AIonly once the rest can carry it
Every layer rests on the one below it.
Your systems can’t capture the data
We build the software
Custom systems and internal tools that record your work properly, so good data exists in the first place.
HKAL: a purpose-built lab system
You have data, but it’s a mess
We clean and engineer it
Our core work. Messy data becomes one reliable foundation for reporting and forecasting.
DNS Africa and TMC
Your foundation is solid
We build AI agents on it
Agents and models that work with your data: answering questions, spotting risks and doing the follow-up.
UbuntuIQ matching, DNS renewal forecasts
03The value
What a solid foundation lets you answer
Quickly, and with numbers you can believe.
Who should we target?
Clean customer records show who your best clients are, and who looks like the next one.
What is happening right now?
Pipelines refresh on their own, so everyone works from today’s numbers.
- Order in0.4d
- Credit check0.6d
- Approval3.8d
- Picking0.5d
- Delivery0.9d
Where is work getting stuck?
Consistent process data shows exactly where work waits.
Where are we heading financially?
Forecasts built on reconciled history hold up when finance checks them.
Charts are illustrative.
04Case study · DNS Africa
30 years of registry data. One number everyone trusts.
Legacy systems, duplicated sources and manual reports. Nobody could point to a number and say “that’s correct.” We cleaned three decades of domain history into one warehouse the registry now runs on every day.
- Legacy registry systems
- EPP and DAMS commands
- DNS traffic
- Billing
- Manual reports
- Capture
- Clean
- Transform
- Verify
Star-schema data warehouse
1994 to today · updated every 15 minutes · on-premises
- Regulatory reporting
- Registrar dashboards and APIs
- Abuse monitoring
- Renewal forecasts
- Brand Watch
Presented by Neil Dundas of DNS Africa at the ICANN84 meeting in Dublin.
- Regulatory reporting time
- Hours → minutes
- regulatory reporting time
- Rows of domain history
- 600–800m
- rows since 1994, queried in milliseconds
- Refresh interval
- 15 min
- between updates, inside the registry’s own environment
“Once data is cleaned and verified, the numbers never change again.”Our golden rule for the warehouse
- Real-time DNS abuse monitoring
- Renewal forecasting and anomaly detection
- A standard layer across all DNS Africa deployments
- Spun out Brand Watch, a look-alike domain monitoring product
TMC · The Music Company
Dailystats on every label and artist · being finalised
From Excel and a locked-down data provider to their own platform: daily and weekly performance, social stats and forecasting in one place
Hearshaw & Kinnes Analytical Laboratory
Livesince September 2026
Spreadsheets and workarounds replaced by a purpose-built lab system
Read the projectThe Lekker Network · UbuntuIQ
147AI matches for 74 attendees at one event
Network data structured first, so AI matching could work
Read the project05Ownership
Your data. Your servers. Your platform.
We don’t take your data away to our cloud or rent you a tool. We work as an extension of your team, on your servers and infrastructure, and everything we build belongs to you.
DNS Africa’s warehouse runs inside the registry’s own environment, so data sovereignty is never a question.
- ERP
- CRM
- Billing
Your warehouse
on your servers
- Dashboards
- Forecasts
- AI
Nothing is copied out to us. It all stays with you.
06How it starts
It starts with 5–10 days inside your data
A data landscape analysis. Once we understand what you have, you get a roadmap: what to fix, how we’d fix it, and the value each step brings.
Days 1 to 10
Data landscape analysis
- Every system, spreadsheet and export your business runs on
- How data moves between them, and where it breaks
- The quality of the data itself, tested, not assumed
- Conversations with the people who use the reports
Then build it with us, build it internally, or take it elsewhere.
Your roadmap
Example
- 1One customer record across CRM and billingHigh valueWeeks 1–3
- 2Automated month-end reportingHigh valueWeeks 3–6
- 3Process tracking to expose bottlenecksMedium valueWeeks 6–8
- 4Revenue and renewal forecastingHigh valueWeeks 8–12
For each item: what’s wrong, how we’d fix it, and what it’s worth to you.
Build it with your team
Our engineers work alongside yours, on your servers.
Put BI and AI on top
Only once the foundation can carry it.
Hand it over
Documented, so it never depends on one person, including us.
07Who we are
Tyto started as a student project on 30 years of messy data.
A University of Pretoria final-year project: a few CS students trying to bring structure to DNS Africa’s registry data. It became a long-term partnership, then a company. We still spend our time inside messy data, not slide decks.
We start by understanding how your people work and where problems arise. Rather than promoting a particular product or technology, we choose the tools that fit your business, your existing systems and your budget.
Services
Data diagnostics and consulting
We find the process pain hidden in your data before anyone writes a line of code — so you fix the right problem first.
Data engineering
We build the trusted data foundation underneath everything else — clean pipelines, clear definitions, one version of the truth.
Reporting and analytics
We turn a trusted foundation into reporting people actually use — portals, dashboards and analytics that answer real questions.
Workflow automation and internal software
We remove repetitive admin and build the internal tools your team needs to focus on the work that matters.
AI and modern data systems
We make data ready for AI and modern tooling — because AI on messy data is just faster confusion.
Writing on data foundations and AI
What a data diagnostic workshop actually finds
A diagnostic workshop is not a sales meeting. It is structured work that maps process pain, data mismatches and the manual reporting nobody talks about.
Read insightWhy AI fails when your data foundation is not trusted
AI does not fix a broken data foundation — it accelerates it. AI on messy data is just faster confusion. Trust the foundation first.
Read insightExcel is not the enemy. Uncontrolled Excel is.
Excel did not break your reporting. Uncontrolled Excel did — spreadsheets that quietly became production systems nobody designed, owns or trusts.
Read insightHow to start
Find out what state your data is really in
We spend 5–10 days inside your data and systems, then hand you a roadmap: what to fix, how, and what it’s worth.







