Artificial intelligence does not have to begin with a new ERP, a large transformation programme or a system that employees must learn from scratch. In many companies, the fastest route to useful AI is to improve the business software that already contains customers, orders, products, documents and operational knowledge.
At Lolli Group, we integrate AI into existing ERP and management systems so that it becomes part of everyday work. The objective is practical: reduce repetitive activity, make information easier to find and help people make better decisions, while keeping the data, processes and software that already work.
AI for ERP systems should start with the work, not the technology
An ERP is more than a database. It is where a company records commitments, responsibilities, prices, stock, deadlines and exceptions. Adding a separate AI chat tool may be interesting, but it rarely changes operations by itself.
Useful AI is connected to the right business context. It knows which records a user is allowed to see, works with current information and returns its result inside a controlled process. A sales assistant may need help preparing a quotation. A purchasing manager may need a summary of supplier history. An administrator may need documents classified and matched to the correct record.
This is why our approach starts with a focused ERP and AI assessment: we identify the tasks that consume time, the data already available and the smallest integration capable of producing a measurable result.
How AI can improve an existing ERP
There is no need to transform every process at once. A company can introduce AI one useful workflow at a time and extend it only after the first result has been verified.
1. Search and understand business information
People often know that information exists but do not know where to find it. AI can help users search orders, activities, notes and documents using natural language. It can also summarise a customer history or a project before a meeting, always within the permissions already defined by the company.
2. Read documents, emails and requests
AI can extract useful information from incoming documents and messages, classify it and suggest where it belongs. An employee remains in control, but no longer has to copy every value manually or read every page just to understand what action is required.
3. Assist quotations and customer service
An AI assistant connected to the ERP can prepare a first draft from approved products, prices, previous work and company rules. It can also help answer common customer questions using real order and service information, rather than inventing a generic response.
4. Improve data quality
Duplicate records, incomplete descriptions and inconsistent classifications reduce the value of every report. AI can identify suspicious similarities, propose consistent descriptions and bring anomalies to the attention of the person responsible. It supports control; it does not silently change critical data.
5. Support purchasing and stock decisions
By combining ERP history with current demand, AI can highlight unusual consumption, late suppliers, products that deserve attention and possible stock risks. The final decision remains human, but the system makes the relevant signals easier to see.
6. Turn reports into explanations
Traditional dashboards show numbers. AI can help explain what changed, identify records behind a variation and prepare a concise management summary. This makes business intelligence accessible to more people without removing the underlying figures or the ability to verify them.
Making AI available to everyone without creating another disconnected tool
“Everyone working with AI” should not mean asking every employee to open a separate chatbot and paste company information into it. AI should appear where the work already takes place: next to the customer, order, task, document or report concerned.
Different roles need different assistance. Sales, administration, operations and management should not receive the same prompts, data or permissions. A well-designed integration provides small, specific functions inside the existing workflow and clearly shows when a result was generated by AI and when it requires review.
This approach improves adoption because people do not have to abandon familiar software. It also improves governance because the company can decide which data is used, which provider processes it and which actions always require confirmation.
Why the entry cost can remain low
AI projects become expensive when they begin with a complete replacement, an undefined objective or a platform built before anyone has chosen the first useful process. A more sustainable approach is progressive.
- Keep the existing ERP: connect to the data and functions already available instead of rebuilding everything.
- Choose one high-value task: start where repetitive work, delays or errors have a visible cost.
- Use APIs and usage-based AI services: pay for the functions actually used and monitor consumption.
- Add human approval: automate preparation and analysis before automating consequential decisions.
- Measure the outcome: compare time saved, quality and adoption before extending the project.
This creates a low entry cost and a controlled monthly cost. It also prevents the company from becoming dependent on a large project whose value will only be known at the end.
Claris FileMaker as a bridge between ERP, people and AI
Claris FileMaker is particularly effective when a company needs to connect existing systems, design role-specific interfaces and improve processes that a standard ERP does not cover well. It can act as the operational layer between databases, APIs, documents and AI services.
Lolli Group is a certified Claris Partner with 25 years of FileMaker experience. We use that experience to preserve the logic that makes a system reliable while adding modern AI capabilities where they deliver a real advantage. Our work includes FileMaker AI integration, connections to external ERP platforms and the modernisation of existing business databases.
Security and control are part of the design
Company data should not be sent to an AI service simply because a technical connection is possible. Before activating a workflow, we define what information is required, who can use the function, how results are recorded and where human approval is mandatory.
We also consider data minimisation, provider settings, access control, logging and retention. This allows the business to benefit from AI while maintaining a clear relationship between the original data, the generated result and the person who approved the next action.
A practical path from assessment to everyday use
- Assess: map the ERP, data sources, users and costly manual activities.
- Select: choose one use case with a clear business result and limited risk.
- Connect: integrate the required records, documents or APIs without replacing the whole system.
- Test: run the workflow with a small group and verify quality, permissions and cost.
- Extend: add new roles and processes only when the first integration is useful in real work.
Our AI integration service for ERP and databases follows this path. It turns AI from a demonstration into a controlled business function that employees can use every day.
Frequently asked questions about AI and ERP systems
Do we need to replace our ERP to use AI?
No. In many cases, AI can be connected to an existing ERP through APIs, database access or a dedicated integration layer. Replacement is considered only when the existing system cannot safely provide the required data or workflow.
Can a small company afford an ERP AI integration?
Yes, if the project begins with one focused process. Reusing the existing system and paying for AI according to actual usage can keep both the initial investment and ongoing cost under control.
Will employees need to learn a completely new application?
Not necessarily. The most effective integrations place AI functions inside the software and screens employees already use, with actions designed for each role.
Can AI change ERP data automatically?
It can, but critical changes should normally require validation. We prefer to begin with suggestions, drafts and anomaly alerts, then automate further only when the rules and results have been verified.
Can FileMaker connect an ERP to modern AI services?
Yes. FileMaker can connect databases, REST APIs, documents and role-specific interfaces, making it a practical integration layer for adding AI to existing business processes.
Start with one process that deserves to work better
The best AI project is not the one with the largest promise. It is the one that removes a real obstacle, uses trustworthy business data and becomes part of the way people work.
Lolli Group can assess your current ERP or management software, identify the most useful first integration and build a pilot without forcing a complete replacement. Request an ERP and AI assessment and tell us which process takes too much time today.