Twenty or Odoo: where to start a custom system?
Not a feature list, but real selection criteria: code, extensibility, licences and maintenance.
01 / PLATFORM
twentyCRM / open source
VS
odooERP / business modules
Business systems · custom apps · AI · data engineering
Not everything needs
building from scratch.
I build business systems fitted to how you work, using open-source platforms, AI-assisted development and a solid data architecture. From a CRM or internal tool to AI agents and analytics.
You work directly with the specialist. I pick the technology to fit your problem, not the other way round.
DATA123 / solution sketchEXAMPLE
FoundationTwenty CRM · customers and sales
→
Tailored + AIEnquiry analysis · AI assistant
→
OutcomePower BI · sales analytics
One business process. Connected tools.
Example: from a customer enquiry to a clear view of sales.
Twenty CRM / OdooAI agentsFabric / DatabricksOpen-source LakehousePower BI / Plotly
Notes and experiments
Not just opinions.
Engineering decisions.
I write about what can be built, how to choose an architecture and what's worth knowing before you start. Below are the next topics I'm publishing.
02 / LAKEHOUSE
Bronze
Silver
Gold
Fabric, Databricks or an open-source Lakehouse?
When is PostgreSQL enough, and when is a more complex platform worth it?
03 / AGENTS
Which supplier's quote is best?
I checked 3 quotes and their terms. ↗
An AI agent that works with your business system, not just chats
RAG, access rights, APIs and safe execution of actions in a practical application.
Already published – in the journal →
From idea to working solution
I write about what
I can build.
If a topic matters to your organisation, I can help put it into practice: from adapting an existing CRM or ERP to AI features, a data platform and reporting.
Let's discuss your situation ↗DATA123 / engineering principleModular architecture
01
A working software foundationTwenty CRM, Odoo or the system you already have
02
Custom modules + AIAI-assisted development, agents, API integrations
03
A reliable data layerFabric, Databricks, Spark / Iceberg or PostgreSQL
04
Solutions people actually usePower BI, Plotly and interactive applications
The diagram shows possible building blocks. Not every project needs all four layers.
What I build
One solution.
Several disciplines.
I don't sell a catalogue of separate technologies. I connect the systems you already have and the new pieces you need into one working whole.
01
Custom business applications
CRM, ERP modules, purchasing, budgeting and internal tools. Where it makes sense, I start from a mature open-source system rather than an empty project.
Twenty CRMOdooCustom modules
02
AI built into your processes
AI agents, internal knowledge assistants, document analysis and automated actions that work inside your own systems.
LLMRAGAI agentsAPI
03
Data platforms and integrations
I connect ERP, CRM, SQL, Excel and API data. The architecture can be Microsoft, Databricks or open source, depending on scale.
Microsoft FabricDatabricksOpen-source
04
Analytics and interactive tools
Reports, metric models, custom dashboards and AI interfaces that put data to use in everyday work.
Power BIPlotlySQLAI
How I work
More attention on your business logic. Less on rebuilding standard features.
Open source gives you working features and architecture out of the box. AI coding tools can read that code, its documentation and APIs, so custom changes can often be delivered faster.
AI helps build, but it doesn't replace security, testing or architectural judgement. Not every task needs a new platform.
Where does the build effort go?ILLUSTRATION, NOT A TIME ESTIMATE
From an empty project
A lot of basic components have to be built and tested first.
On an open-source foundation
More of the time goes into what actually creates value for the business.
Choosing the technology
One tool doesn't fit every problem.
Where PostgreSQL is enough, you don't need a full Lakehouse. Where you need more data and AI scale, I pick platforms built for it.
Business systems
A ready foundation, custom modules and integrations.
AI features
Knowledge assistants, automated actions and custom agents.
Data layer
A Lakehouse or a simpler database, depending on real needs.
User interface
Trustworthy reports and custom interactive interfaces.
These are possible directions to choose from, not a fixed stack every solution must use.
A product I'm building
Kilmo. So it's clear where the numbers come from.
A data lineage and dependency tool I build. It shows the links from the original source to the report, and what a practical solution fitted to one specific problem looks like. Try it: click any object.
Kilmoby DATA123Ddemo · editor
Interactive Lineage Click any object to see where it comes from and what it feeds.
Kilmo's lineage view with demo data, not client data
About DATA123
A technical partner, not an anonymous agency.
DATA123 is an independent business systems and data engineering practice. I work directly with your team: from understanding the problem and the architecture to the actual coding and integration.
HOW I WORKFewer intermediaries.
Fewer intermediaries.
More technical clarity.
I'm not trying to install as many tools as possible. First I look at what you already have, what's worth keeping and where custom development will make a tangible difference.
Architecture · Development · Integration · Analytics
Start with your problem
Maybe you don't even need a new system.
Sometimes an integration, one custom module or an AI feature is enough. Describe your situation briefly and let's start with the right question.
Or write directly: labas@data123.lt · I reply within 1–2 working days
The button opens your email app with this text.