Cloud · data · agentic AI consulting

Cloud, data and AI architectures that hold up in production.

Polykroma helps large organisations with their data and AI projects, from scoping all the way to production, on the cloud of their choice. We design architectures that follow your roadmaps, then deploy them alongside your teams.

Polykroma, from the Greek for “many colours”

One colour per expertise, one painting.

A data project rarely sits in a single field. A cloud migration raises security questions. A data platform without governance quickly loses its users' trust. An AI agent is only as good as the data it reads.

Each colour in our palette stands for one area of expertise. We combine them according to what your project needs.

Data is only worth something when your teams can trust it and use it at scale. That is what we build, from scoping to production.

Expertise

Five areas of expertise, combined as you need.

Each engagement draws on the colours it needs. Here is what each one covers. We also design and build websites and web applications, like the one you are browsing.

01 · Cloud

Solid cloud foundations

Landing zones, networking, identity: we design foundations your teams can evolve without starting over. Public, sovereign or hybrid cloud, connected to the environments you already run.

  • Reference architecture and landing zones
  • Sovereign and trusted cloud
  • Hybrid and multicloud interconnection
  • Virtualisation and private cloud
  • Infrastructure audits

PlatformsOCI · AWS · VMware

02 · Data

An open data infrastructure, ready for agents

Your data in open formats, readable by the engine of your choice and by your AI agents. We build tested, monitored pipelines from source ingestion to business use, without locking you into a single vendor.

  • Open lakehouse on Apache Iceberg
  • Open formats and interchangeable engines
  • ELT ingestion and source integration
  • Transformation, modelling and semantic layer
  • Data exposed to agents through MCP

PlatformsApache Iceberg · Snowflake · Databricks · DuckDB · Parquet

03 · Artificial intelligence

Agentic AI, in production

Many AI projects stop at the prototype. Agentic engineering takes them all the way to real use: agents that act inside your systems, built like production software, with a human who stays in charge.

  • Agents with clear roles, multi-agent systems
  • Token-efficient MCP servers
  • Skills shared across the team
  • Document search (RAG)
  • Evaluations and human approval
  1. $agent run "sales pipeline"
  2. plan3 steps · 2 agents
  3. mcpcatalog.sources()14 sources · 210 tokens
  4. buildmodels/sales.sqlwritten
  5. test6 checks6/6
  6. cost1,842 tokens€0.02
  7. reviewawaiting human approval

PlatformsClaude · Codex · LangChain · MCP · Python

04 · Migration

Modernise without stopping the business

Leaving an ageing system takes a precise plan. We map what exists, break the migration into reversible steps and organise each cutover with your business teams, whether it involves databases, applications or virtualisation.

  • Assessment of the existing estate and roadmap
  • Database migration
  • Moving VMware environments to the cloud
  • Application modernisation
  • Gradual cutovers and rollback

PlatformsVMware · Oracle · OCI · AWS

05 · Governance

Governed, protected data

Knowing where data comes from, who can access it and what it may be used for. We put in place the rules and tools that answer these questions, including for your auditors.

  • Catalogue and lineage
  • Access control and encryption
  • GDPR and data sovereignty
  • Data organisation and roles
  • Audit readiness

Approach

Bridging strategy and delivery.

A large project brings many people to the table: business leaders, IT, security, procurement. We connect them, make technical topics clear for those who decide, then turn those decisions into measurable results.

  1. 01

    Scope

    Qualify the need with the business: goals, budget, existing systems, regulation. We start by listening.

  2. 02

    Design

    Define the target architecture, with its options and what they cost, so you can decide with full knowledge.

  3. 03

    Deploy

    We work with your teams, in your code, all the way to production. First results come quickly and can be measured.

  4. 04

    Hand over

    Documentation, upskilling, handover. Your teams should be able to carry on without us.

Work

Experience with large organisations.

Financial services, insurance, luxury, retail, industry, pharmaceuticals, media, energy, telecoms, public sector: cloud, data and AI architecture engagements in environments where every mistake is costly.

  • Société Générale
  • BNP Paribas
  • AXA
  • LVMH
  • Sanofi
  • Thales
  • Canal+
  • M6
  • Lacoste
  • Orange
  • Oracle
  • Monaco Cloud
  • Ministère des Armées et des Anciens combattants
  • Ministère de l'Agriculture, de l'Agro-alimentaire et de la Souveraineté alimentaire

Selected engagements

  • Agentic AI · Data engineering

    Pipelines designed by agents

    A multi-agent system that generates, tests and deploys data pipelines, built on token-efficient MCP servers.

  • Agentic AI · Our own product

    An agentic prospecting engine

    Agents that watch a market's signals, qualify each company and draft the first message, leaving the sales team with a single step: approval.

  • Data · Luxury and retail

    Pipelines at scale

    A data pipeline architecture aligned with the analytics and AI roadmap of large international groups.

  • Cloud · Public sector

    A sovereign cloud

    A sovereign cloud infrastructure built to the security and hosting requirements of a public body.

  • Cloud · Banking

    A hybrid cloud

    Interconnecting a major bank's data centres with a public cloud, to extend existing systems without rewriting them.

  • Web · Websites

    Websites people enjoy

    Design and development of fast, carefully crafted websites, like this one and the seentia site: animation, contact form, hosting and search visibility.

  • Applications · Pharmaceuticals

    A management application

    A low-code application with dashboards to follow a pharmaceutical laboratory's activity day to day.

  • Migration · Energy

    Switching over an HR system

    Data takeover and migration of an HR system, prepared so that teams would not notice the difference on the day.

Our own product

The signal first, then the outreach.

is the prospecting engine we designed and run. Its agents follow public signals in a market, such as a new hire or a public tender, check each company against official registers, then draft a message. Nothing goes out without human approval.

It is our approach to agentic AI, applied to our own business.

  • Prospects spotted before your competitors
  • An intent score for every company
  • Messages ready to send
  • Follow-ups and replies tracked
Visit seentia.io

A project in mind?

A few lines are enough to start. We read every message and reply personally.

Write to us