Names everyone understands
Turn status codes into named states, vendor IDs into linked records, and raw columns into typed fields.
Legacy modernisation · Enterprise
Move your legacy systems forward.
Keep control of every step.
Bring AI agents and automation to the systems your business depends on. Inistate adds governed workflows around your existing data, so you can modernise now and migrate in phases.
Start with one process. Connect what you have. Deploy on your own servers if you need to.
Your existing systemsKeep running
ERP & databasesIn-house appsFiles & mailboxes
↓ Controlled sync & write-back ↑
InistateThe workflow decides
The workflow decides
what happens next.
↓ Permitted activities ↑
PeopleWeb & mobileAI agentsMCPAutomationAPI & integrations
Make your existing systems ready for AI
Your data has a history.
Give it a permitted next step.
A status code tells you what is stored. Inistate adds what it means, who can act, and which activity is allowed next. AI agents connect through MCP (Model Context Protocol) to work with those rules.
Existing ERP record
PO_NO 2600412
AMT 18450.00
STS_CD P3
APPR_BY NGSLThe mapping turns codes and fields into a business record, with rules reviewed by your team.
Purchase order
PO-2600412
Allowed next
Issue to vendorAI + confirmation
Put on holdPeople only
Cancel orderPeople only
Illustrative workflow. Your team defines the states, permissions and approval requirements.
The whole case, in view
See the current state, permitted next steps, ownership and history through the access of the connecting identity.
Workflows that outlast the ERP
People and agents perform business activities. When the underlying system changes, adapt and validate the integration mapping while retaining the workflow above it.
A phased migration path
Start small. Prove it.
Move at your business’s pace.
Keep the legacy system as the system of record during the pilot. Move ownership of each process only after your team has validated its data, controls and recovery plan.
- 01
Map one real process
Choose purchase orders, claims or service jobs. Map the fields, states, owners and approvals. AI can help draft the module; your process owners review it.
- 02
Connect and reconcile
Import existing records or sync through an API, database or file integration. Agree which system owns each field and check record counts, identifiers and values before enabling writes.
- 03
Pilot with human oversight
Run representative cases and exceptions. Introduce AI on selected activities, require confirmation where needed, and verify permissions, write-back and the resulting history.
- 04
Expand, then retire what is ready
Roll out by process or department. Agree cutover criteria, backups and rollback with IT. Retire a legacy component when its replacement and retained records are accepted.
Integration and migration scope depends on what your system exposes. Review connectivity, data quality and historical records in the initial assessment.
Governance built into the work
The model reasons.
The workflow controls the action.
People, AI agents and automations act on the same business record, with defined activities and a shared history.
Give agents a defined scope
Agents work with modules, records and permitted activities through MCP. They do not need credentials to write directly to legacy tables. Use dedicated agent connections and grant only the access they need.
Keep important decisions with people
Configure activities for people, AI, or AI with human confirmation. Set confidence thresholds: a below-threshold AI attempt is recorded as an intention for review, without moving the state.
Check before work moves
Access, form requirements and the current state determine whether an activity may proceed. A repeat approval is rejected when approval is no longer allowed from that state.
Keep the evidence on the record
History shows who acted, when, and what changed. AI actions carry reasoning, sources and reported confidence, so reviewers can inspect the decision in its business context.
A more manageable cost of change
Reuse the integration.
Reuse the rules.
A shared workflow layer can reduce repeated integration work as you add teams, agents and automations. New modules use the same schema-driven MCP tools, reducing the need for a separate tool implementation for every table.
What happens as you connect more tools?
Illustrative model: 3 legacy systems, each consumer needs all 3, polling once per minute.
Consumers can be AI agents, automations or apps.
| To maintain | Point-to-point | Shared layer |
|---|---|---|
| Integrations | 24 | 11 |
| Legacy polls / day | 34,560 | 4,320 |
Arithmetic model, not measured savings or a price quote. Integrations: 3 × consumers versus 3 + consumers. With one consumer, the shared layer adds an integration. Actual cost includes mapping, migration, infrastructure, support and any AI usage.
On-premise deployment
Your records.
Your servers.
Your boundary.
Run Inistate Enterprise on servers you control. Records, files and change history stay on your premises.
AI and outside connections are optional. Keep them disabled where required, or agree an architecture that meets your organisation’s data boundary before enabling them.
Your organisation’s boundary
Inistate Enterprise
Records · Files · Change history
Legacy systemsInternal usersApproved integrations
You decide what connects
For an entirely internal AI workflow, the model, MCP client, automation runtime and all connected services must also meet your boundary requirements. An external AI service can receive data you send to it.
Quoted per deployment. Sizing, installation and support are agreed up front; billing is through Inistate or your dealer.
Questions from enterprise teams
Before you begin
Do we have to replace our ERP first?
No. Start with the approvals, exceptions and handoffs around the ERP. Keep it as the system of record while you prove the workflow, then decide which processes to migrate or retain.
What if our legacy system has no API?
Assess supported database access, exports, file drops or mailboxes as integration inputs. The available connection and write-back approach depend on your system and access policies; confirm them during discovery.
Does on-premise mean AI data automatically stays internal?
On-premise keeps Inistate’s records, files and history on your servers. The full data path also depends on your AI provider and connected services. Outside connections are optional; review every endpoint before enabling AI for confidential data.
How do we keep synchronisation reliable?
Map legacy changes to permitted activities, use stable record IDs and idempotent write-backs, and monitor failures. State checks reject invalid repeat transitions; same-state edits and external side effects still need duplicate protection. Read the reliability guide.
What should we bring to the first discussion?
One process, its system and data sources, a sample schema with sensitive values removed, user and approval roles, deployment requirements, and an estimate of records and integration volume. These help scope a pilot and a deployment quote.
A practical first step
Bring one legacy process.
Let’s map its next chapter.
Review your systems, controls and deployment boundary with the Inistate team. Define a pilot your business and IT teams can evaluate together.
Discuss your migration ↗