Skip to content
MLLabz

Solutions

Have a problem? Let's understand it.

Every piece of work starts with the problem: who has it, why it exists, and what better would look like. The technology comes after.
  1. Problem
  2. Data
  3. Intelligence
  4. Engineering
  5. Impact

Where we work today, and where we're heading.

Each area shows how far along we honestly are. Core today means we've built real, working systems of this kind. Developing means we're actively building the capability, with early work to show. Direction means it's where we're heading, and we don't offer it yet.

Operations and business systems

Core today

“Our operations run on spreadsheets, paper, and memory, and things fall through the cracks.”

Who has it
Landlords and property managers, small businesses, schools, and any organisation that has outgrown manual processes.
What we might build
Management platforms, internal tools, payment matching, automated reminders, and reports people actually read.
Where intelligence comes in
Once the system holds real data, things like spotting who is likely to fall behind on payments, flagging unusual readings, and forecasting.
Evidence
LandlordPMS, in development.

Education and access to opportunity

Core today

“People can't find, understand, or reach the opportunities that fit them.”

Who has it
Students, parents, schools, and counsellors, and anyone who has to make a big decision from scattered information.
What we might build
Calculators and matchers built on official rules, directories, guidance tools, and dashboards for counsellors.
Where intelligence comes in
Assistants grounded in a person's own data, recommendations, and estimating chances from past patterns.
Evidence
CareerNext, live.

Data and decisions

Developing

“We have data, but it doesn't help us decide anything.”

Who has it
Organisations sitting on records, spreadsheets, and exports that nobody has time to analyse.
What we might build
Pipelines that collect and clean data, dashboards, reports, and analyses that answer one specific question.
Where intelligence comes in
Forecasting and models, once the data is trustworthy.
Evidence
CareerNext's data work. We collected and cleaned course and cutoff data from official sources (about 1,000 TVET, KMTC, and TTC courses across more than 14,000 offerings, plus several years of degree cutoffs) and turned it into charts and matching. Lab entries will follow.

Prediction and intelligent systems

Direction

“We want to see what's coming, not only what happened.”

Who has it
Anyone whose decisions depend on what happens next: admissions, rent collection, demand, maintenance.
What we might build
Prediction models, early-warning systems, and decision support, each measured against a simple baseline.
Where we are
We haven't built a production prediction model yet. Our first is planned for CareerNext: estimating admission chances from several years of cutoffs. We'll publish the experiment in the Lab, including if it doesn't beat the simple approach.
Evidence
Lab experiments, as they appear.

What happens when you bring us a problem.

  1. 1Understand the problem

    Talk to the people who have it. Learn the constraints, the evidence, and what success would mean.

    What you getA short written problem brief, with a measure of success we agree on

  2. 2Map the system

    Look at the people, processes, data, and tools already involved.

    What you getA clear picture of where the problem really sits

  3. 3Find the highest-leverage fix

    Choose the simplest thing that would make the biggest difference: software, analytics, automation, AI, or a change in process.

    What you getA recommendation, with the reasoning behind it

  4. 4Build the smallest useful solution

    Build something real that people can use early.

    What you getA working first version

  5. 5Measure the impact

    Compare the results with the measure from step 1.

    What you getAn honest report of what changed

  6. 6Improve

    Use what we learned to make the next version better.

    What you getA system that gets better over time

Sometimes step 3 shows that the best answer isn't new software. If so, we'll tell you.

What we won't do.

  • We won't add AI where it doesn't help.

    If a simple rule or a clear report solves the problem, that's what we'll build.

  • We won't build complexity to look advanced.

    Simple systems are easier to trust, use, and maintain.

  • We won't promise results we can't measure.

    We agree on the measure first, then report against it honestly.

  • We won't use your data carelessly.

    Data is handled securely, used only for the problem, and treated with respect for the people behind it.

Not sure where your problem fits?

That's normal. Most real problems don't fit neatly into a category. Describe what's going wrong, and we'll work out the rest together.

Bring a problem →