How to add Slack to Baan IV / Baan V
Short answer
Create a small Slack app backed by a service that reads Baan through read-only database views or exported files. Use alerts for events, slash commands for lookups such as item or order status, and interactive buttons for approvals that are imported into Baan by a reviewed job.
Slack tends to show up at Baan sites through engineering, IT or a recently acquired business unit, rather than the plant itself. A shared channel for supply chain or customer service is where the pain is felt: people paste Baan screenshots into Slack to ask who knows the status of an order.
That screenshot habit is the opportunity. A slash command like /baan order 40213 that returns status, ship date and open lines removes the screenshot, the delay and the need for everyone to hold a Baan licence for look-ups.
Because Baan is old, the integration is deliberately conservative: read-heavy, write-light, and always through paths that Baan itself validates.
What teams use it for
Order status slash command
Customer service types the order number and gets status, lines, planned and promised dates, and shipment references, taken from a read-only view.
Stock availability lookup
Engineers and sales query an item and warehouse for on-hand, allocated and on-order quantities without opening an inventory session.
Shortage alerts to the planning channel
After each MRP run, post a threaded message of the critical shortages by buyer code, with a link to a report for the full list.
Machine-down production escalations
A manual command creates an alert thread tying a work centre to open production orders, so planners can re-sequence while maintenance works.
Supplier delivery watchlist
Daily message listing purchase orders due this week that have no receipt, grouped by buyer, with a button to snooze.
Month-end blockers
Finance posts counts of unposted receipts and invoices from an exported summary so owners can clear them before close.
Baan IV / Baan V integration surface
- Read-only database views over sales order, purchase order, inventory and production order tables, accessed from the Slack backend over ODBC or the native client of the underlying RDBMS.
- Scheduled report printing to ASCII or CSV through Baan devices, picked up by a file watcher for sites that block database access.
- Baan item, business partner and order numbering conventions, used to validate Slack command inputs before any query runs.
- AFS (Application Function Server) and data exchange imports for the few write actions you choose to allow, executed as a defined integration user.
- Baan user codes and authorisation groups, which you can mirror in the Slack backend to decide who may run which command.
How it works
- 1
Pick three commands
Start with order status, item availability and PO status. They are read-only, quick to build and cover most screenshot requests.
- 2
Create the Slack app
Define slash commands, a bot user and the minimal scopes (chat:write, commands). Use signing secrets to verify every request.
- 3
Build the query service
A small service receives the command, parses and validates the order or item number, runs a parameterised query on a read-only view and formats a Block Kit message. Never concatenate user input into SQL.
- 4
Map Slack users to Baan roles
Join Slack user IDs to Baan user codes so a customer service user cannot view cost data that only purchasing may see.
- 5
Add scheduled alerts
Run the extract after MRP or nightly, compute what changed since the last post, and send threaded summaries to the right channels.
- 6
Optionally add actions
Add Approve or Acknowledge buttons that write to a staging table, then import through a supervised job into Baan.
Example flows
Order lookup by customer service
Trigger: A rep types /baan order 40213 in the support channel.
- Slack sends the command to the service with the user ID.
- Service checks the user's role and the format of the order number.
- Query returns header, lines, dates and shipment status.
- Service replies in the channel or as an ephemeral message.
Result: The rep answers the customer in seconds, with no Baan login and no screenshot.
Post-MRP shortage thread
Trigger: The MRP session finishes at night.
- Scheduled job queries planned orders flagged late.
- Items are grouped by planner code.
- Service posts one message per planner with a threaded list.
- Planner replies in thread with the plan and marks it with an emoji reaction.
Result: Exceptions are triaged visibly in the morning, and decisions are searchable later.
Keeping data on-prem
Slack is a SaaS product, so treat it as an outbound notification and query surface only. Keep Baan on-prem and host the Slack app backend inside your network or DMZ, reachable by Slack through a published HTTPS endpoint or by Socket Mode, which uses an outbound connection and avoids opening inbound ports. Return only what the asker is allowed to see, and use ephemeral replies for sensitive values. If your policy forbids business data in a US-hosted SaaS, restrict Slack use to alerts that hold identifiers and links, and keep details in an internal page. A local language model can translate free-text questions into the three or four safe queries without sending data outside.
Security checklist
- Verify the Slack signing secret on every request and reject stale timestamps.
- Use read-only database credentials and parameterised queries only.
- Enforce per-command role checks, particularly for cost, price and supplier terms.
- Prefer ephemeral responses for individual lookups so data is not left in public channels.
- Retain logs of who ran which command with which parameters.
Ways to get it done
Home-grown Slack app with a few commands
Effort: One to three weeks for three commands, including role mapping.
Teams with a developer comfortable with Baan table structures and a Python or Node service.
Low-code integration platform
Effort: Days for alerts; slash commands with validation take more custom code.
Sites that already own a platform such as Zapier-class or enterprise iPaaS tools with database access.
Netray build with natural-language lookups
Effort: Short project; Netray adds a grounded query layer, shows the SQL it ran and keeps it read-only by default.
When users want to ask free-form questions rather than memorise commands.
Pitfalls
- !Slack channels are broader than Baan authority groups. Posting price or cost detail in an open channel is a leak, not a convenience.
- !Slash commands time out in a few seconds. Acknowledge immediately and post results asynchronously when queries are slow.
- !Old Baan tables rarely have clean indexes for your lookup. Test query speed against production volumes before launch.
- !Item and order numbers may have leading blanks or fixed-width padding. Normalise inputs or lookups will silently fail.
- !Free workspace limits and retention policies can delete the audit evidence you thought you had; keep your own log.
Frequently asked questions
Is there a Slack app for Baan?
No official one. The standard approach is a custom Slack app that queries Baan data via read-only database views or exported files. That keeps Baan unchanged and gives you control over which data is exposed and to whom, and it avoids the risk of third-party connectors that expect modern APIs Baan does not provide.
Can Slack users update Baan records?
They can request updates, but a prudent design stages them. An Approve button writes to a staging table, a reviewer sees the queue, and a controlled import applies the change through AFS or data exchange. That way Baan's validations run and you have a human check on anything touching quantities, prices or statuses.
Do we need to open firewall ports to Slack?
Not necessarily. Socket Mode lets your service open an outbound connection to Slack, so no inbound port is required. Alternatively publish an HTTPS endpoint in a DMZ behind a reverse proxy. Your security team will choose based on policy, and either way the Baan server itself never connects to Slack.
How is this different from the Teams option?
The pattern is the same: relay, read-only access, staged writes. Slack puts more weight on slash commands and Block Kit messages, while Teams often centres on adaptive cards and Entra ID. Choose the one your people already use. Running both from one relay is common in mixed organisations.
Can an AI model answer questions in Slack from Baan data?
Yes, with guardrails. A model can map a question to an allowed query, run it read-only and format the answer, showing the query used. For sensitive plants the model runs on-prem so no order data leaves. Without those guardrails, free-form SQL generation is risky on schemas as large as Baan's.
Want Slack working on your Baan IV / Baan V?
A free 30 minute assessment: we map the integration path for your version and setup, and where it should run.
Get a free assessment