Every business we run stands on one superintelligence we own.

We created our own AI system. It builds our software, talks to our customers, and lets a compact team ship like a company five to ten times its size.

8
Businesses, one platform
1
Owner per feature
2
Week sprints
~200
Production pods
24/7
Nikol answers first

One platform. Several diverse businesses.

SESSIA is our technology company. It maintains a library of ready-made components: loyalty, payments, CRM, analytics, and more. Every FIG business is assembled from them.

From day one

A new business starts with working parts, never from zero.

Its own engine

Separate infrastructure, so it can be scaled, funded or sold on its own.

Any provider

New payment or POS systems plug in without changing the rules.

A customer is never charged twice.

Points, payments and wallets are money. If the same request arrives twice by mistake, our system records it once and stops the copy.

Records can't be edited

Every transaction can be audited.

Balances live on our servers

An app can't change them by itself.

Watched 24/7

On-call engineers are alerted immediately.

Built to be changed.

Loyalty, payments, content, analytics, notifications and moderation are independent services. Each one is deployed, scaled and changed on its own, so a problem in one part never stops the others.

CLOUDAWSevery environment defined as codeORCHESTRATIONKubernetes~200 production pods for FIRSTLINESERVICESGo · Pythonfast backend · our own MLAPPSFlutterone codebase for iOS and Android

Event-driven

Services talk through durable events. A retry never doubles a business operation.

Never by hand

Servers and networks are code, changed only through reviewed updates.

Every operation traceable

One action can be followed through every service it touched.

Safe to change

New behaviour runs next to the old until everyone has moved over.

Seven hand-offs. Or one owner.

Usually a request passes through seven people and loses meaning at every step. Since April 2026, one owner takes it from request to release.

1person
Meaning kept100%

Client, analysts, designers, architect, developers, QA. Everyone waits for someone.

Inside a two-week sprint

Day 1

Architecture first

Rules and failure cases are agreed before coding.

Every day

Decisions written down

Next to the code, so anyone can pick it up.

Every change

Tested, not assumed

Automated tests and real experiments.

Day 14

Release

A short request in, a release out.

After

Measured by results

Value for users, not features shipped.

Five rules every feature follows.

Cadence

Two-week sprints

A short request goes in, a release comes out. Questions get answered along the way, not in a documentation phase.

Memory

Decisions live next to the code

Every feature records what changed, why, and what we ruled out. Anyone can pick it up mid-flight.

Evidence

Validated, not assumed

Product changes are tested with experiments. Code is checked by automated tests, monitoring and careful rollouts.

Design

Architecture first

Business rules, failure cases and integrations are agreed before coding, and challenged by peers.

Measure

Outcome over output

We count value delivered to users, not features shipped. Good tech is reliable and cheap to change.

Halcón builds. People decide.

Halcón is our own AI development pipeline. It writes, tests and reviews code. People decide what to build and approve every result.

01
Business

A short request

We didn't add superintelligence to our process. We rebuilt the process around it. Halcón does the repetitive work, the owner makes the decisions, and nothing is lost in between. We are not just developers: every owner thinks like a founder. That is how a team our size runs many projects in parallel, and why we are hard to compete with.
Oscar ArrietaCTO, Freedom International Group
One source of truth

An engineering knowledge base, versioned with the code. Change a standard once, and every engineer and every tool follows it.

The right model per task

Locally hosted and international AI models, chosen for speed, cost and where data is stored.

The recipe stays ours

What Halcón does is public. How it is built is proprietary and stays inside SESSIA.

Nikol: a personal assistant for every customer.

Nikol is our AI for the customers of every FIG business. She answers first, day and night, and learns each customer over time.

Writes first

A delay, a reorder, a free table: before you ask.

Signed Nikol

A message from your assistant, never a mass campaign.

day 1 · new
Can I book a table for Friday?
Done: Friday, 20:00, table for two. — Nikol

Remembers you

Orders, tastes, context. Each next step gets better.

Calls a person

When it matters, a human joins with the full history.

Proof, not promises.

Vision’s turnover had been falling for eight years. In 2017 it merged with FIG and moved onto our technology.

2008201020122014201620182017 · moves onto our stack$119M $34.9M $35M 2018 target · 2.1×
Sources: Cyprus Stock Exchange filings (2008), public reporting on the 2017 Freedom–Vision merger. The technology worked together with a new operating team.

Why customers come back.

Seven things built into every SESSIA app.

Effortless store
Community inside
Support that knows you
Delivery you can watch
Privilege programs
Analytics that advise
Wallet built in

Next: every engineer becomes a founder.

We're creating a new role, the SIPO, Superintelligent Product Owner (working title): one person who owns a product and its revenue, with Halcón as the team.

A founder's mind

Decides what to build and answers for its revenue.

An engineer's hands

Shapes the architecture, owns reliability and cost.

A superintelligence's reach

Directs Halcón. Thinks like a founder, ships like a team.

Few people have this profile, so we train it ourselves. The programme is run by the group's president.

SIPO one person

How it will work

  1. 01

    Split by audience

    Customer app, partner CRM, internal tool: one owner each.

  2. 02

    No managers between

    The person who builds is the owner.

  3. 03

    Revenue goal first

    Every feature has a KPI before release.

  4. 04

    Prove it or stop

    One owner, one checkpoint date.

  5. 05

    Halcón tests ideas

    Prototypes go to real users automatically.