Next chapterRead the announcement

Event

What we heard at our banking technology forum in Riyadh

Rows of empty dark blue seats in a dimly lit auditorium

A day with banks and technology leaders from across the region, under one condition: say what is actually happening. Our notes on AI, reach, branches, security and what we are changing as a result.

Tayseer Innovations hosted a one-day banking technology forum in Riyadh this month, bringing together people who run technology, risk, operations and digital teams at banks and financial institutions across the region.

We set one condition for the day. Nothing said in the room would be attributed to a person or an institution, so that people could say what was actually happening rather than what their press office would prefer. It worked better than we had dared to hope, and the conversations in the corridors were even more candid than the sessions.

What follows are our own notes. They are not a transcript, and they are not a claim that the room agreed on everything. It very much did not, which was the most useful part of the day.

Why we held it

We spend most of our time inside individual projects, and that gives us a narrow view. Every bank tends to believe its problems are unique, and many of them are not. We wanted to put people in one room who rarely get to compare notes with their peers, and to listen more than we talked.

For that reason there were no product presentations. Each session opened with a short framing from someone who had lived through the problem, and the rest of the time belonged to the room.

An empty theatre with curved rows of wooden seats facing a lit stage
The sessions were kept short, so that most of the time belonged to the room.

Agentic AI: the questions have changed

A year ago, the question in a room like this was whether AI could do the work. This time, almost nobody asked. The questions had moved on to who is accountable when it does the work wrongly, how to tell a supervisor what it did, and how to stop it.

Several institutions described the same pattern. A pilot had gone well, internal enthusiasm was high, and the project had then stalled for months in the gap between a system that could act and an organisation ready to let it. The gap was rarely technical. It was about who would sign the risk acceptance.

We have stopped asking whether it works. We are asking who gets the phone call when it doesn't.

A head of digital banking

Where people are actually using it

The room was noticeably more conservative in practice than the industry's marketing would suggest. Most live deployments were doing narrow, well-defined jobs: explaining charges, answering policy questions for staff, preparing summaries for a person to review. Very few agents were moving money without a person involved, and nobody present seemed embarrassed by that.

Arabic remains the sharpest test

Dialect came up in every session on this topic, and usually with some feeling. Several people described systems that performed well in evaluation and poorly with real customers, particularly on voice. One participant suggested that any vendor quoting an accuracy figure for Arabic should be asked which Arabic, and spoken by whom.

Digital banking: reach is still the hard part

The session on digital banking quickly left apps behind. Nobody in the room thought their main problem was the interface. The problem was the last stretch: the customer in a rural district with a phone and an intermittent connection, who still needs somewhere to turn a balance into cash.

Agent networks dominated the discussion, and the mood was realistic. Recruiting agents is the easy part. Keeping them supplied with float, keeping them honest, keeping them motivated when volumes are low and supporting them when something goes wrong is where these programmes succeed or fail.

  • Several participants said agent support deserved as much investment as customer support, and rarely received it.
  • Offline and poor-connection behaviour was named as the most underestimated requirement in wallet projects.
  • Women's access came up repeatedly, as a question of who feels able to walk into which shop, and not only as a question of product design.

The branch is not going away

We had expected this session to be about closing branches. It turned out to be about what branches are for. Nobody present was planning to abandon them, and several institutions were investing, though in a different kind of branch from the one they had inherited.

The consensus, as far as there was one, was that self-service hardware works best when it frees people rather than replaces them. An ATM or teller machine that handles routine cash lets staff spend time on the conversations that need a human. Banks described mixed results where the machines arrived and the staff roles did not change.

Uptime came up more often than features. A self-service terminal that is out of order does more damage than no terminal at all, and in the words of one participant, customers do not remember the nine times it worked.

Security: the questionnaires are getting longer

The security session was the most animated of the day, partly because it brought together people from both sides of the vendor assessment. Banks explained that their questionnaires keep growing because each new incident or regulatory change adds questions and nothing ever removes them. Vendors explained that they spend weeks answering them, and suspect nobody reads the replies.

Both were probably right. The more useful conversation was about what actually reassures a risk team, and the answer came back consistently.

  1. A precise statement of where the data sits, and who decides if that changes.
  2. A clear description of what a certificate does and does not cover.
  3. A record of how a real incident was handled, with the embarrassing parts left in.
  4. Being allowed to speak to the engineers rather than only the account manager.

Several risk leaders said a vendor admitting a limitation raised their confidence, because it suggested the rest of the answers were honest as well.

Open banking: fewer connections, better consent

We had expected the open banking session to be about standards and timelines. It spent most of its time on customers. Several institutions said that connecting to a third party had turned out to be straightforward, and that the difficulty lay in everything around the connection: explaining it, keeping a record of it and undoing it.

One participant described testing their own consent journey with members of their family, and watching none of them manage to find where a permission could be withdrawn. Nobody in the room looked surprised. There was broad agreement that a bank carries the reputational risk for connections it did not build, because that is who the customer will call.

The more interesting disagreement was commercial. Some saw open banking mainly as a compliance cost, others as the first real chance in years to offer something new, and a few said quietly that they were waiting to see what their competitors did. Nobody claimed to have found the business model yet, and that honesty was a relief after a year of confident slides.

Running what you launch

The final session was the least glamorous on the agenda and the one people talked about most on the way out. It asked what happens in the third year of a platform, when the launch team has moved on, the vendor's attention has moved on and the people left running it were not there when the decisions were made.

Stories were remarkably consistent. Documentation that described the system as it was designed and not as it had become. Support contracts measured on response time while nobody measured whether the problem went away. Monitoring dashboards that everyone had learned to ignore because they were always a little bit red.

What helped was unexciting. Keeping one or two people from the build team on the operations side for at least a year. Reviewing incidents without looking for someone to blame. Treating a platform's running costs as part of the original decision, and not as a surprise that arrives later. Several people said they wished they had asked more about year three when they were choosing a vendor, and we suspect the same is true of those vendors' other clients.

Five things we took away

Sitting down afterwards, our own team agreed on five points that had come up repeatedly, across sessions that were supposedly about different subjects.

  • The hard problems are organisational. Technology is rarely the thing holding a project back.
  • People trust a narrow system that admits its limits more than a broad one that does not.
  • Local language and local context are not a feature to add later. They decide whether the thing works at all.
  • Operations are undervalued. Launches get the attention, and year three decides the outcome.
  • Institutions want to talk to each other far more than they get the chance to.

What we are changing

A forum is only worth holding if it changes something, so we have made ourselves write this part down.

We will share our incident review format with clients, so that they can tell us whether it gives them what they need when they have to explain an event internally. We will put more of our engineering time into dialect testing on voice, using recordings from real conditions rather than clean ones. And we will stop opening first meetings with a capabilities presentation, which several people pointed out, kindly but firmly, that nobody had ever asked for.

What comes next

Many of those who attended asked for a smaller, more technical session, and for one focused on risk and compliance teams alone. We intend to do both. Our thanks to everyone who gave up a day and spoke so openly.

Institutions interested in future sessions can contact the team. You can also read more about our work and the people behind it.

About Tayseer Innovations

Established in 2016 and built on the Kuraimi Group, Tayseer Innovations is a fintech company in Saudi Arabia and the UAE: over 5,500 professionals from more than 20 nationalities, serving over 4 million customers regionwide with agentic AI, digital banking, and core banking technology.

Media contact

info@tayseer.me
  • Sample post
  • Events
  • Agentic AI
  • Digital banking
All news

More news

More from the newsroom