Conversational AI for IFS Cloud
Talk to IFS Cloud.
It answers — and acts.
NgageChat is a conversational AI interface for IFS Cloud. Ask about work orders, stock, purchase orders or invoices in plain language and get the answer from live IFS data — then create, update and release the actual records, signed in with your own IFS account and inside your own IFS permissions.
- 27
- native IFS tools for discovery, CRUD, actions and documentation
- 10
- IFS domain modules — work orders to finance — with built-in expertise
- Read + write
- not a read-only chatbot: it creates, updates and releases records
- Your login
- each user signs in through IFS; IFS permissions are the boundary
Which work orders at site 100 are overdue?
Seven overdue work orders at site 100. Four are on the pump house.
| WO | Object | Due | Priority |
|---|---|---|---|
| 3391 | P-114 | 28 Jul | 1 |
| 3402 | P-118 | 30 Jul | 2 |
| 3417 | C-22 | 01 Aug | 2 |
Raise a fault report on P-114 — mechanical seal leaking
Confirm before writing to IFS
- Object
- P-114
- Site
- 100
- Fault code
- LEAKAGE
- Priority
- 2
Fault report 41207 created
Written to IFS as the signed-in user.
Why another assistant
Most ERP assistants stop at the answer
Answering a question is genuinely useful. But the work is not finished until something in the system has changed — and that is where nearly every ERP chat product stops.
What a read-only assistant does
It tells you the pump has a leaking seal and three overdue work orders. Then you open IFS, find the screen, and do the thing you already knew you needed to do.
- Answers questions
- Explains how the software works
- Does not create the record
- Does not run the action
What NgageChat does
It tells you the same thing, then raises the fault report, adds the operation, updates the promised date or releases the order — after showing you exactly what it is about to write.
- Answers from live IFS data
- Explains how IFS works, from official documentation
- Creates, updates and deletes real records
- Runs IFS actions — release, approve, complete
On IFS’s own AI: IFS Cloud includes IFS.ai Copilot, an in-product assistant. IFS’s own 25R2 documentation describes it as designed for informational and advisory purposes and states that it does not perform direct transactions, answering from IFS documentation. It is good at what IFS documents it as doing, and many customers will run both. NgageChat does the other half: your live records, and the transaction. See the full comparison, or read what separates the three kinds of IFS Cloud chatbot. Quoted from IFS Cloud 25R2 documentation and checked on 3 August 2026. IFS ships two release updates a year; we re-check this every quarter.
How it works
From a sentence to a committed IFS transaction
Four steps, all visible on screen while they happen. You can watch which projection it searched, which endpoint it called, and what came back.
You ask
Type it the way you would say it. "Which work orders at site 100 are overdue?" "Raise a fault report for pump P-114, the seal is leaking." No screen to find, no field names to know, no training.
See the flowIt finds the right place in IFS
NgageChat searches its index of the IFS Cloud data model for the projection and entity set that actually handle the request, then resolves the real required fields, enums and list-of-values — adding your custom fields, read live from your environment. Not a guess, and never an invented dropdown.
How discovery worksIt fills in the blanks with you
Defaults come from IFS itself via the entity’s _Default() function. Anything still missing comes back as a short form with the real choices from your system — so you answer three questions instead of hunting through twelve tabs.
The confirm stepYou confirm, it commits
Before anything is written, NgageChat shows you exactly what it is about to do. You approve, it calls the IFS OData API under your own login — and every message and tool call is logged for audit.
Security & permissionsThe IFS engine
Why the answers are right, and the writes succeed
NgageChat does not guess at IFS. 27 native tools resolve the right projection, learn the exact payload each endpoint expects, and execute against your IFS OData REST API.
Grounded in the real IFS data model
NgageChat carries a curated index of IFS Cloud itself — projections, entity sets, key and required fields, enums and list-of-values — so it knows the shape of the system before it touches yours. Your custom fields are then read live from your environment, and your data always is.
Read moreIt writes, not just reads
Create, update and delete records; run bound actions like release and approve; call action imports for wizard-style flows such as expense sheets and invoices. Read-only chatbots stop at the answer — this one finishes the job.
Read moreSigned in as you
Sign in through the browser with your own IFS account over OpenID Connect and every IFS call carries your token, so IFS decides what you can see and change. Which sign-in mode a deployment uses is a decision worth making deliberately — the security page sets out the options.
Read moreAnswers from official IFS documentation
Ask how something works, not just what the data says. A semantic search over IFS Cloud documentation returns the explanation with the relevant diagrams and screenshots — so new starters stop queueing at a super-user’s desk.
Read moreCharts on request
"Break that down by supplier as a pie chart." Results come back as bar, line, area, pie or donut charts rendered from your live IFS data — no report builder, no BI licence, no extract.
Read moreAttachments that do something
Photograph a leaking seal and NgageChat reads the image to draft the fault report. Attach a spreadsheet and it extracts the rows. Link a document to a work order or an asset without leaving the conversation.
Read moreSecurity by construction
Write access, without widening access
The thing that stops most ERP AI projects is not whether it works — it is the meeting where someone asks who exactly is allowed to change what. NgageChat answers that with your existing permission sets rather than a new model to design and defend.
-
Sign in with your own IFS account
With browser sign-in, IFS sees the request arrive from you. Your existing permission model applies without anyone configuring it a second time.
-
Confirm before every write
The complete payload is shown as a readable summary. Nothing reaches IFS until you approve it, and you can change a value there rather than starting again.
-
A log you can reconstruct from
Every message, tool call, parameter, result, model, duration and error is recorded — and IFS keeps its own audit, because the call arrives as the real user.
-
Your cloud, if you want it
Self-host so IFS traffic never leaves your boundary, or let EX10 run it. No core modifications either way.
Close work order 3391 and book 6 hours against it
Two changes to make. I need the work type for the time booking — your IFS enum offers Normal, Overtime, Travel.
Normal
Two changes — approve both?
- Work order
- 3391
- State
- Finished
- Hours
- 6.0
- Work type
- NORMAL
- Reported by
- j.okafor
Work order 3391 finished
Grouped into a single changeset so the two writes could not half-succeed.
Nothing reached IFS until the summary was approved.
What you can ask
Named transactions, not adjectives
“Transactional” only means something when it is specific. Here is what NgageChat actually reads and writes, area by area.
Work orders & maintenance
“Which work orders at site 100 are overdue?”
5 reads · 6 writesPurchasing & suppliers
“Which purchase orders from Acme are still awaiting delivery?”
5 reads · 5 writesCustomer orders & sales
“What is the status of customer order 5512?”
5 reads · 5 writesInventory & stock
“What is the stock balance for part 74821 across all warehouses?”
5 reads · 4 writesManufacturing & shop orders
“What is the status of shop order 40122?”
5 reads · 5 writesProjects & finance
“Which supplier invoices are awaiting authorisation?”
5 reads · 5 writesHSE & field reporting
“Report a near miss — pallet fell from a rack in bay A-12, nobody hurt”
4 reads · 4 writesIFS how-to answers
“How does IFS calculate work order scheduling?”
4 answer typesDomain expertise
Ten IFS areas it already knows its way around
NgageChat loads the domain knowledge relevant to your request — the entity quirks, the action names, the order operations have to happen in. Asking about a shop order and asking about an expense sheet are different jobs, and it treats them that way.
Why NgageChat
Built only for IFS — which is the whole point
Built only for IFS
Not a general assistant pointed at an API. NgageChat is built around IFS projections, OData conventions, _Default() resolution, ETag concurrency and bound actions — the details that decide whether a write succeeds or 400s.
It will not invent a value
Dropdown options come from your enums, then your list-of-values, then a plain text box. If NgageChat cannot find real options in your system, it says so rather than offering plausible-looking ones.
You watch it work
Every step is on screen — which projection it searched, which endpoint it called, what came back. When something fails you get the actual IFS error and what to try next, not a shrug.
By people who live in IFS
Built by EX10, a boutique consultancy founded by former IFS leaders — the same people who implement and support IFS Cloud for a living.
Use cases
Where it earns its place
IFS is strongest in asset-intensive and service-heavy industries. So is conversational access.
The hard questions
The things people actually ask before they say yes
Usually about fifteen minutes into the first conversation. Here they are up front, answered specifically.
What stops it hallucinating a record?
It never writes from memory. Every field is resolved from your live IFS metadata — required fields, enums, list-of-values — and every write is preceded by an on-screen summary you have to approve. If a value cannot be verified against your system, NgageChat asks instead of guessing.
Are you giving an AI write access to production?
You are giving it exactly the access the signed-in user already has. Each person authenticates with their own IFS account; NgageChat calls IFS with that token. Someone who cannot release a work order in IFS cannot release one by asking.
Can we prove what it did afterwards?
Yes. Every user message, assistant reply and tool call is logged with timestamps, the model used and the outcome, and IFS records its own audit as normal because the calls arrive as that user.
Does it need a data warehouse or a sync?
No. NgageChat queries IFS Cloud live over the OData REST API at the moment you ask. There is no replica to go stale and no nightly extract to maintain.
Will it survive our next IFS upgrade?
Nothing is installed in your core, so there is nothing to re-apply after an upgrade. The data-model index we maintain tracks IFS releases, your custom fields are re-read live from your environment, and all access is through standard IFS REST APIs.
Our data cannot leave our environment.
Self-host NgageChat in your own cloud so IFS traffic never leaves your boundary, or let EX10 run it for you. Either way, your IFS credentials stay between the user and IFS.
People already have IFS. Why chat?
Because most of a workforce never learns it. A planner who lives in IFS is fine. Someone who opens it twice a month is not, and there are far more of the second than the first. Conversation serves them without taking the screens away from anyone who prefers them.
What happens when it cannot do something?
It tells you what it tried, which call failed, the exact IFS error, and what you could do instead. A dead end you can act on beats a confident wrong answer.
FAQ
Frequently asked questions
What is NgageChat?
Can I chat with my IFS Cloud data?
Does NgageChat only answer questions, or can it change data in IFS?
How does NgageChat handle IFS security and permissions?
How is this different from a generic AI assistant connected to an API?
How do you stop it inventing data?
Do we need to modify IFS Cloud to use it?
Can NgageChat answer questions about how IFS works, not just what the data says?
Ngage Suite
NgageChat is the conversation. The suite is the rest.
Five AI products for IFS Cloud from EX10, designed to work together — chat for people, agents for processes, forms for the frontline, code for the gaps, advice for the programme.
NgageFlow
Agentic AI workflows for IFS Cloud
Visit siteNgageEase
No-code AI forms — mobile and offline
Visit siteNgageCode
AI app builder for IFS Cloud
Visit siteNgageConsult
AI co-pilot for IFS implementation
Visit siteBrowse the full range on Ngage Suite, or read about the team at EX10 Global.
Early access
See NgageChat answer a question from your own IFS environment
A 30-minute session: bring one real scenario — an overdue work order list, a fault report, a supplier lookup — and watch it run end to end. No slideware.