How it works
A thin AI layer next to the systems you already use
We don’t replace your ERP or make you change how you work before you see results. We connect to what is already there and put the information to work.
What stays the same
- The ERP remains the system of record
- Nothing goes into the ERP or the accounts without a person’s approval
- Each role sees only what it needs
- Procedures change little, and step by step
What changes
- Repetitive tasks done in minutes
- Past information found by anyone with access
- Decisions made with the full context in view
- Reports and documents in each client’s language
Architecture
From the information you already have to work done
Your systems
- ERP (for example, PHC)
- Spreadsheets and PDFs
- Email and documents
- Photos, voice and video
AI layer
Can run in your environment
- Connectors to each system
- Voice, text and images in any language
- Search index over your data
- Local or approved cloud models
- Actions approved by a person
- Permissions and change history
Work done
- Quotes
- Planning
- Archive search
- Service reports
- Translated documents
Read-only first
No changes to core systems unless agreed.
Your data stays yours
Never used to train models.
A person decides
Every entry in the ERP is approved, except what we agree to automate.
Ready for system changes
If a system is replaced, only its connector changes.
Where the AI runs
Inside your company, or in the cloud if you prefer
On your company’s servers
- Local models: the data never leaves the company
- FairAI has no access to the data
- Fixed cost and stable versions that can be audited
- Works where cloud AI is not allowed
It is the option most of our clients choose.
In the cloud
- Models such as those from OpenAI or Anthropic
- We tell you what leaves, where it goes and on what terms
- Cloud models still lead on the hardest reasoning
Integrations
We connect to what your company already uses
Every connection we build becomes a reusable component: each new project starts from work already done, not from scratch.
- ERP · SAP, Sage, Primavera, PHC…
- We read your ERP and, where agreed, write to it, so quotes and documents stop being retyped.
- Spreadsheets and databases
- Years of orders, quotes and records scattered across spreadsheets, searchable in one place.
- Documents and PDFs
- Specifications, manuals, standards and reports, found by what they say, with the source of every answer.
- Photos and images
- The AI reads the text on photographed plates and documents and describes each image, so a photo can be found by what it shows, even if nobody filed it.
- Voice and video
- Voice notes, site videos and recorded meetings become reports and summaries you can search.
- Custom applications
- When the right system is missing, we build it: from a quoting CRM to a field service app, on mobile and desktop.
For IT teams
Questions we often hear
Where does our data go?
In the setup we recommend, the AI runs on your servers with local models, and the data never leaves the company. If you choose the cloud, we agree in writing what leaves, to which provider and on what terms.
Do you train models on our data?
No. The data is yours and is only used to make your solution work.
What hardware do we need?
It depends on usage and on the models. We size the hardware with you in the proposal, as an estimate, and handle the installation.
What if we change ERP?
Only the connector to that system changes. The rest of the solution stays the same.
Do we need a dedicated IT team?
No. We need one point of contact per area (during the pilot, about 1 hour a week) and access to the agreed systems. If we need to talk to the company that supports your ERP, we handle it. Installation, maintenance and updates are on us.
How long until we see results?
In the pilot, about a month, a first version is already working with your data. After that, each module goes live within a few weeks.
Have a question about architecture, security, models or your systems? We answer in writing or in a call with your team. Ask a technical question