# fares suliman

From Knowledge to Impact

- Site: https://faressuliman.com/

## Build for the Context, Not the Demo

- Link: https://faressuliman.com/build-for-the-context-not-the-demo/
- Published: 2026-11-20T18:18:20+00:00
- Author: admin

A polished demo can make a service look universal: one prompt, one connection, one
clean answer. Real use may involve shared devices, intermittent connectivity, several
working languages, or a paper step the demo never showed. Across African markets
and communities, conditions differ widely; the useful question is which conditions
shape this particular problem.

## Start with a local job

Describe the job in the words of the people doing it, not in the language of a product
category. A farmer, a clinic administrator, and a small-business owner may each 
need help finding or organizing information, but their decisions, constraints, and
tolerance for error are not interchangeable. A local partner can help reveal what
an outside design team has missed.

Spend time with the existing process before proposing an AI feature. Notice where
people switch between voice notes, messaging, forms, and face-to-face clarification,
and ask what makes each step useful. The objective is not to replace familiar work
simply because a new tool can.

## Treat constraints as design inputs

Language support involves more than translating a button: it can affect how people
phrase questions, understand a response, and correct a misunderstanding. Connectivity
and device access can determine whether a feature should work offline, use less 
data, or offer a non-digital fallback. These choices are product decisions, not 
last-minute adaptations.

Consider who maintains the service, where information is stored, and how a person
gets help when something goes wrong. The answers depend on the specific country,
sector, and community, so broad regional assumptions are a poor substitute for local
research. Build the questions into discovery before committing to the technical 
design.

## Test value beyond launch

A strong early test asks whether the tool fits the existing task, whether people
can verify important outputs, and whether the intended users actually choose to 
return to it. Include the people who will support and maintain the service, not 
only its first users. Their experience often surfaces costs and dependencies a demo
leaves out.

African opportunities are not a single market waiting for a ready-made solution.
They are specific problems, expertise, relationships, and possibilities that deserve
to shape what gets built from the start. Begin with a local job, test in the conditions
where it happens, and let the evidence—not the demo—decide what comes next.

## Governance Is Part of the Product

- Link: https://faressuliman.com/governance-is-part-of-the-product/
- Published: 2026-11-20T18:18:20+00:00
- Author: admin

Imagine an automated system that routes applications to the right team—until a missing
field sends a person’s request into a queue nobody checks. The technical feature
may be functioning as intended. The service is not. Governance matters because it
shapes what happens around the system, especially when real people meet its edge
cases.

## Name the decision

Before approving a tool, write down what decision it informs and who remains accountable
for that decision. “We use AI for operations” is too broad to guide a review; “staff
use it to draft a response, then verify the facts before sending” gives a team something
concrete to examine. The distinction sets the boundary between assistance and authority.

That boundary should reflect the consequence of getting the answer wrong. A rough
internal summary and a decision affecting someone’s access to a service do not call
for the same level of review. Context determines the safeguards, not the novelty
of the software.

## Build checks into the flow

Put the check where the work happens: a reviewer confirms key facts before a response
goes out, a team can flag a misrouted case, or a manager can see when an automated
recommendation was overridden. A rule that staff cannot follow during a busy workday
is unlikely to protect anyone consistently. Good governance is legible at the point
of use.

Ask what information enters the system, who can access its output, and how long 
records need to be kept. The answers will vary with the organization, the task, 
and the applicable rules, so they should be checked with the right internal or local
expertise. Recording those decisions makes later review possible.

## Make learning part of oversight

A launch review is a beginning, not a guarantee that a system will remain suitable.
Assign an owner to revisit complaints, unexpected outputs, process changes, and 
whether the original use still makes sense. This does not need to become a grand
committee; it does need a named person and a clear route for raising concerns.

The point of governance is not to slow every experiment. It is to make the useful
experiments easier to trust and the risky ones easier to question. If people can
tell who is responsible, what the system may do, and how to challenge a result, 
governance is already part of the product.

## The Best AI Pilot Begins Before the Model

- Link: https://faressuliman.com/the-best-ai-pilot-begins-before-the-model/
- Published: 2026-11-20T18:18:20+00:00
- Author: admin

A team asks for an AI pilot, and the first question is usually which tool to try.
I’d start somewhere less exciting: trace one task through the people, documents,
approvals, and exceptions that already shape it. The best pilot often begins with
a workflow that is tedious, bounded, and important enough for someone to notice 
if it improves.

## Follow the work first

Take a real process, such as preparing a meeting brief or sorting incoming service
requests, and map what happens from arrival to handoff. Note where information is
missing, who checks it, and which decisions depend on context that never appears
in the form. Those details help distinguish a language task from a coordination 
problem.

Then choose a narrow step where assistance could save attention without taking away
judgment. Drafting a first summary might be a reasonable test; deciding which request
deserves urgent treatment may not be. A clear boundary gives people a way to challenge
the output instead of treating fluency as proof.

## Design the human check

A useful pilot names who reviews the result, what they should verify, and what happens
when the system is uncertain. If staff must quietly redo the whole task, the experiment
is measuring a demo rather than a change in work. Give reviewers a simple way to
record errors and recurring edge cases.

The right measure depends on the task: fewer minutes spent assembling a draft, fewer
incomplete submissions, or clearer handoffs between teams. Pick a baseline before
introducing the tool, and include the review effort in the comparison. A faster 
first step is not a gain if it creates a slower last step.

## Keep the decision reversible

Use information the team is allowed to share with the chosen system, and make the
trial small enough to pause if the output is unreliable or the workflow changes.
Record what the tool can and cannot do, who owns the process, and what evidence 
would justify another round. That note becomes more valuable than a polished pilot
presentation.

AI in practice is less about finding the most impressive model than making a careful
change to how work happens. Start with a task people recognize, keep a person responsible
for the consequential call, and measure the whole path. If the pilot teaches the
team something useful even when it fails, it was still worth building.
