Joseph Amegashie

Turning ambiguity into systems.

I design systems for the point where complexity becomes a decision.

A way of thinking

Problem
Evidence
Constraint
Judgment
System
Outcome

Problem / Question

What is actually failing?

Name the failure before naming the product. A form, a score or a dashboard is not yet a definition of the problem.

In practice / MarketRise Capital

MarketRise connects borrower information, verification and repayments around the lending decision, rather than treating the application form as the whole operation.

Follow the example

Evidence / Question

What do we know versus assume?

Separate a claim from its supporting record. Keep the source and the limits of the evidence visible enough to challenge.

In practice / MarketRise Capital

Verification records and trust-score events are distinct from the borrower profile. A score is not a substitute for the records behind it.

Follow the example

Constraint / Question

What limits the possible solution?

Some limits are part of the operating reality, not defects to design away. Authority, required evidence and dependencies shape what may happen next.

In practice / Lartiel Ventures

Lartiel’s quality gate requires an assigned inspector, completed required checks and the attachments those checks require before a decision.

Follow the example

Judgment / Question

What decision must be made?

Identify the point where information becomes an authorized action. A system can support that judgment without making accountability disappear.

In practice / MarketRise Capital

MarketRise separates recorded borrower acceptance from the owner-authorized act of final disbursement.

Follow the example

System / Question

What structure makes that decision repeatable?

Connect the records, boundaries and handoffs so the next decision does not depend on reconstructing the same context from scratch.

In practice / Lartiel Ventures

Lartiel tracking records retain event history and event source alongside the current status, connecting movement to its operating record.

Follow the example

Outcome / Question

What changed in reality?

Distinguish a working capability from a demonstrated improvement. The existence of a system is not a measured business result.

In practice / MarketRise Capital

MarketRise has implemented lending, repayment and reporting structures. Claims about repayment improvement still require operational measurement.

Follow the example

Principles behind the decisions

Evidence before confidence.

Confidence should never exceed the quality of the evidence supporting it.

Design around the decision.

Do not begin with features. Begin with the decision someone must be able to make and work backward.

See it in practice / MarketRise Capital See it in practice / Lartiel Ventures

Preserve meaningful human judgment.

Automate repetition. Augment judgment.

A question to consider / AI can increase productivity while quietly reducing independent judgment.

Not every signal deserves influence.

The hardest problems are often not missing information.
They are missing structure.

Where evidence, constraints and judgment become action.

What are you here to understand?

HOW I WORK See the reasoning behind the systems.
WHAT I’VE BUILT Explore the systems and the evidence behind them.
WHAT I’M THINKING ABOUT Questions, observations and ideas I’m still working through.
WHETHER WE SHOULD WORK TOGETHER Start with the problem, not the job title.

Bring a problem.

Use the reasoning model to make one difficult situation more legible.

Selected systems

MarketRise Capital

ORIGINATED SYSTEM

Founder-built / Finance & operations

Can trust become infrastructure?

Small-business finance depends on more than a loan. Borrower information, repayment records and trust need a structure that makes the operation visible.

Founder / systems builder

Building the financial and operational infrastructure around borrower information, repayment and trust.

Follow the reasoning The decision framework Visit MarketRise Capital

Lartiel Ventures

TRANSFERRED SYSTEM

Client-built / Trade & operations

How do you turn international trade uncertainty into a controlled workflow?

Requirements, sourcing and supplier workflows connect to quality-control evidence, approval decisions, movement and tracking. The work is not to digitize the handoffs. It is to make them traceable.

Technical Director

Technology strategy, digital platforms and systems development. Applying the same decision-led thinking beyond finance.

Follow the reasoning The decision framework

THINGS I NOTICE

Observations, open to revision.

Patterns worth questioning before they become assumptions.

AI can increase productivity while quietly reducing independent judgment. Follow the question

Observation
A possibility to examine, not a universal claim.

Why it matters
A faster answer can make the act of checking it feel less necessary. I am interested in the point where useful assistance becomes an unexamined default, especially when the decision still belongs to a person.

Question it creates
What should remain independently reasoned, even when a system can produce a plausible answer?
Related principle / Preserve meaningful human judgment.
Most dashboards fail when nobody defines the decision the dashboard exists to support.
Small businesses may have an information architecture problem before they have a financing problem.
Consistency becomes dangerous when every decision depends on the same intelligence.

One capability becomes another.

Record
What actually happened?

Accounting taught me to begin with the record: what can be captured, checked and explained.
Interpret
What does the information mean?

Banking and analytics moved the question from recording an event to interpreting its significance.
Architect
What should exist instead?

Building MarketRise shifted the work from understanding an existing operation to constructing the structure it needed.
MarketRise Capital
Transfer
Can the same method work in another domain?

Lartiel tested the transfer: from financial decisions to supplier work, quality evidence and commercial handoffs.
Lartiel Ventures
Judge
When should systems support judgment, and when should judgment remain independent?

Current work around AI and decision systems keeps that boundary open to examination. More assistance does not automatically mean better judgment.

What problem are you trying to solve?