Financial services · Programmes · AI

Critical programmes rarely fail on technology. They fail when nobody knows where they really stand.

Coperis Partners leads transformation programmes and AI projects for banks, asset managers, payment firms and fintechs. We establish the real status, we say it out loud, then we take the programme to completion.

  • 4 → 1portfolio management systems consolidated across Brussels, Paris and Luxembourg
  • 10,000points of sale integrated for a global payments company, serving 26 million customers
  • TARGET2brought into service for a public financial institution: the euro area’s large-value payment system
  • Certifiedby Pennylane: the connector of our own AI product

Programmes led by our partners, including in roles held before Coperis.

Situations

The five situations we get called in for

  1. 01

    “Everything has been green for six months. Go-live is in eight weeks.”

    The reporting reassures, the teams worry. We start by establishing credible dates, milestone by milestone, and what each one depends on.

  2. 02

    “The date is regulatory. It will not move.”

    We plan backwards from the deadline, isolate the critical path and get early decisions on what must leave the scope.

  3. 03

    “The AI prototype convinced the board. Nobody knows how to put it into production.”

    Data, GDPR, costs, monitoring, accountability when it gets things wrong: we deal with what the demo did not show, or we recommend stopping.

  4. 04

    “We need to be listed on a partner’s marketplace.”

    APIs, OAuth, synchronisation, certification file: we have done it for our own product, all the way to certification.

  5. 05

    “Our programme director leaves in a month.”

    A senior profile takes over the programme under your governance, with a handover organised with the team in place.

Method

Start from the real status

Every organisation and every programme is different: pace, format and scale adapt to your context. The principle does not change: establish the real status before proposing anything.

  1. 01

    Listen

    Interviews with the people running the programme: sponsor, business, IT, contractors, vendors. A read of the plan, the budget, the risk register and the latest steering committees.

  2. 02

    Establish the real status

    For each milestone: the reported date, the credible date and what it depends on. Gaps are quantified, not commented on.

  3. 03

    Get decisions made

    A short note for the sponsor: the real status, the decisions to take, and what happens if nobody takes them.

Fictional example

Situation note · Payment platform migration

To: programme sponsor · Restricted circulation

MilestoneReported dateCredible dateWhat it depends on
Business acceptance testing complete15 March12 April · lateTest data sets delivered by the vendor before 1 March. They have not been ordered.
Partner flows cut over30 April28 May · lateTwo of seven partners have not started end-to-end testing.
Legacy system shut down30 June30 JuneAchievable if regulatory reporting moves to phase two.

Decisions requested from the 4 March steering committee

  1. Move regulatory reporting to phase two, or add two people to acceptance testing for six weeks.
  2. Escalate this week to the two late partners, at management level.
  3. Freeze scope changes from 1 April.

Without a decision on 4 March, the legacy shutdown date becomes uncertain as well.

Example of a situation note delivered at the start of a programme leadership engagement. The format adapts to each client. Names, dates and figures are fictional.

Services

Four ways to work together

01

Programme leadership and recovery

For the life of the programme

A programme with several teams and contractors, a deadline that will not move, or a programme that is drifting.

You get A situation note at the start, then a plan that holds, a risk and decision log, and a dashboard the board can read.

See details
02

AI scoping and pilot

A few weeks

Decide on measured results, not on a demo: one use case, a limited scope, metrics and a stop threshold set in advance.

You get A pilot tested on your data, a results report and a recommendation: continue, adjust or stop.

See details
03

Integrations and production rollout

Scoping, then delivery leadership

Connect your product to a partner platform, get it certified, or take an AI pilot into production with its safeguards.

You get A documented integration architecture, a complete certification file, a production rollout and monitoring plan.

See details
04

Senior reinforcement

As long as you need

Replace a departure, absorb a peak, bring in a rare skill: a senior profile, from our team or our network of independents, under your governance.

You get A counterpart who sits in your committees and reports at the pace that suits you.

See details

References

What we have led

Asset management

Four systems, three countries, one platform

Context
A European asset manager with a middle office spread across Brussels, Paris and Luxembourg: equities, fixed income, FX, settlement, SWIFT flows.
Our role
Led the middle-office platform transformation programme, with a team of fifteen across three sites.
Outcome
Four legacy portfolio management systems consolidated into one. Project portfolio costs cut by 30%.
Payments

Ten thousand points of sale

Context
A global payments company opening its services through retail chains and banking partners in France.
Our role
Led delivery of the API integrations, with teams of more than thirty people across five countries.
Outcome
Services available to 26 million customers, in 10,000 points of sale.
Public banking

TARGET2

Context
A public financial institution joining the new European system for settling large-value payments.
Our role
IT project management: SWIFT flows, architecture, a team of engineers and consultants.
Outcome
TARGET2, the euro area’s large-value payment system, brought into service.
Product

Our own AI product, certified

Context
Relanzy, invoice follow-up software for French small businesses, built and run by Coperis.
Our role
Design, Pennylane integration, sending from the customer’s own mailbox, AI classification of replies, GDPR by design.
Outcome
Connector certified by Pennylane; app verified by Google and Microsoft.

References without client names reflect programmes led by our partners, including before Coperis was founded. We never publish client names.

Coperis AI Labs

We run what we recommend

Relanzy chases unpaid invoices from our users’ own mailboxes and lets an AI classify the replies. What production teaches us, we write down.

See the AI Labs and our field notes

invoicesremindersreplyread byyour customerpausesor resumesYouraccountingRelanzytracks due datesYourmailboxYour customerthe debtorReplyclassified by AIPennylane
  1. Your accountingPennylane supplies the invoices and their due dates
  2. RelanzyIt watches due dates and prepares the reminders
  3. Your mailboxReminders go out from your own address: Gmail, Microsoft or SMTP
  4. Your customerThey read the reminder and reply
  5. Reply classified by AIPromise, dispute, request for time: the sequence is paused or resumed
How Relanzy works, simplified.

Questions

What we get asked before signing

You are a small firm. Why trust you with a critical programme?

Because the person leading the work is the person who made the commitment to you. No pyramid, no junior trained on your budget. When an engagement needs more people, we bring in senior independent consultants, under a single contract and under our responsibility. We have worked this way before: from 2014 to 2018, more than ten consultants, seven key accounts in Belgium, France and the UK, more than 2,200 consulting days.

Who actually does the work?

It depends on the engagement. Coperis remains your counterpart and coordinates the whole; the team is put together for the subject, with senior profiles we introduce to you.

Do you do software development?

It is not our core business. We scope, lead and accept delivery. We can read an architecture, an API and a test plan, because we build our own product.

How do you handle confidentiality?

A non-disclosure agreement before any detailed exchange, if you wish. No data from your organisation in an AI tool you have not approved. No client names published, here or anywhere else.

Why does a consulting firm build software?

To live with the constraints we ask our clients to accept: certification, data protection, production incidents, the running cost of an AI model. It is the best way not to recommend what we could not do ourselves.

Where do you work?

From the Paris region, across France, Belgium and Luxembourg, on site or remotely depending on the engagement. In French or English.