ITF Digital

Questions before you commit

Ownership, data, AI, Arabic, cost and support, and when the honest answer is to build nothing.

Code ownership
Yours, in your repository, from the first commit
Data export
Tested in the first release
Support
Named owner, response window, cancellable
Quote
Fixed, after the systems audit

The questions below are the ones that decide whether a software engagement goes well. Most of them are about what happens when it ends: who owns the code, who holds the data, how you leave, and whether the thing should have been built in the first place.

A decision tree with three questions: does somebody own the step, does a product already cover it, and does the work actually repeat. Three of the four outcomes are not a build, and the do-not-build outcome is drawn live.A PROCESS THAT HURTSDOES SOMEBODY OWN THE STEPNOFIX THE OWNERSHIP FIRSTSOFTWARE ONLY HIDES ITYESDOES A PRODUCT ALREADY COVER ITYESBUY THE PRODUCTWE WILL NAME ITNODOES THE WORK ACTUALLY REPEATNODO NOT BUILDTHE SAVING WAS IMAGINARYYESNOW IT IS WORTH SPECIFYINGONE RECORD, ONE OWNER, ONE OUTPUTTHE MOST COMMON HONEST ANSWERMOST AGENT AND AUTOMATION PILOTS DIE HEREAND IF IT SURVIVES ALL THREEONE WEEK TO MAP ITA WRITTEN SPECIFICATIONA FIXED NUMBERTHEN YOU DECIDE AGAIN
Should this be built at allThree of the four outcomes are not a buildSwipe to pan
01

The three questions behind all the others

Most of what a buyer wants to know reduces to three worries, and they are all reasonable.

Will I be locked in. Will I be sold something I do not need. And will it still work when the person who built it moves on. Every answer below is shaped by those three, and the structural answers are the same each time: the code is yours, the export is tested early, the specification says what is out of scope, and the operations manual is written for the person who inherits it rather than for us.

If an answer here is uncomfortable, that is deliberate. The ones we could soften are the ones worth reading.

02

The short version

TopicThe short answer
CodeYou own it, in your repository, from the first commit.
DataYours. The export exists and is tested in the first release.
PriceA fixed quote after the systems audit.
TimelineOne week to map, days to specify, two to five weeks to something usable, then about two weeks of pilot.
AIReads, checks, chases and drafts. A named person approves.
ArabicDesigned in, not retrofitted. Say so at specification.
SecurityReadiness and governance work sits with ITF Cyber.
SupportA named owner and a window per severity. Cancellable.
03

Answered properly

Q01

Who owns the code and the data?

You do, both. The code sits in your repository from the first commit rather than being transferred at the end, and the data export is built and tested as part of the first release rather than written the week you ask to leave. Support is a separate, cancellable agreement. If you want to move to another supplier, what you need is the repository, the data model document and the runbook, and you have all three before we finish.

Q02

What if software is the wrong answer?

Then the recommendation says so. If nobody owns a step, automating it hides the problem, and the fix is deciding who owns it. If a product you can buy already covers the need, we name the product. Either way you keep the process map from the systems audit.

Q03

Who makes the decisions when AI is involved?

A named person. Agents read, check, chase, assemble and draft. Approvals, money and anything regulated go to a person, and who that person is gets written into the specification before the build, so it is part of the design rather than a setting.

Q04

Will our documents be used to train someone's AI model?

The specification names the model provider and what its data terms say, and lists each class of data a feature sends to it. You approve that list before the build. Where a class of data has to stay inside your environment, the feature is designed around that.

Q05

Who writes the code?

The people who scoped the work. A person reviews every change before it reaches your repository. The tests behind each change are in the test record you receive.

Q06

Who handles security work?

ITF Cyber. Readiness against NCA ECC and PDPL, security policies and governance, and secure development advice for teams that build or buy software all sit in that division.

Q07

How long does a project take?

One week for the systems audit, days for the specification, two to five weeks to the first increment you can use on real work, then about two weeks of pilot with one team. A large build is several of those cycles rather than one long one, and you get something usable at the end of each.

Q08

What does it cost?

We publish no prices, because the honest number depends on your process and any published figure would have to be padded to cover the range. What we publish instead is what moves the number: record types, roles, approvals, integrations, output formats, AI steps and data migration. The quote comes after the systems audit, as a fixed figure.

Q09

Can you work with the systems we already have?

Usually, and it is the first thing we check. Many products have workflow and reporting features nobody switched on, and using what you have already paid for beats a new tool. Where a connection is needed we build it against a written interface specification, and the identifier mapping comes before the code.

Q10

Does everything work in Arabic?

Interfaces and documents, yes, and right-to-left is designed in rather than retrofitted. Tell us at specification, because bilingual changes some layout decisions and almost no data decisions. Retrofitting Arabic into a finished English interface is the expensive order to do it in.

Q11

What happens after launch?

Support is optional. It gives you a named owner, a response window per severity written into the agreement, a change log and a monthly note, and it is cancellable. Some clients run the system themselves from the operations manual.

Q12

What happens if you stop trading?

You keep working. The code is in your repository, the data is exportable, the data model and the operations manual are written, and the systems are built on ordinary tooling that another developer can pick up. That is not a promise about our longevity, it is a structure that makes the question less important.

Q13

Can we start with something small?

That is the preferred way in. One painful workflow, scoped in a week, with the success measure being the hours it takes today. It is small enough that stopping after the pilot costs you very little, which is exactly why it is the right first project.

Q14

Do you publish case studies?

We publish example projects: typical builds written down to the record, the rules and who signs off, so somebody who has done this work can judge whether we have. A client's operational detail belongs to the client.

Ask the one we missed

If the question that matters to you is not here, it is probably the useful one. Bring it to the systems audit.

Book a systems audit