Every system you run, plugged in your way

Modern SaaS, an old ERP, a desktop program from 2009 — if your team can use it, there is a path to connect it. You do three human things; the platform does the technical rest.

Three ways in

One registry, however the system arrives

Pick a system and MatePlug works out which path fits. Ready connectors authorize in one click. A system with an API but no connector gets one drafted for you. A system with no API at all still has three fallbacks — including Agent Link working its screen.

Connector catalogOne-click authorization for systemswith a ready connectorAI connector builderGive it the API docs — it drafts a private connectorfor your tenant; you confirm the capability listNo API? Three fallbacksRead-only database view · file & report import ·Agent Link works the screenOne tool registryEvery path lands here — after that,agents use it the same way
Fig. — Three ways in, one registry · design diagram

PATH A SHARED BASE · LIVE

Connector catalog

Search by the name you call the system, sign in, approve access. Proposed field mappings are shown in plain words with one real record to confirm — never a field-name puzzle.

PATH B IN DEVELOPMENT

AI connector builder

Point it at the system's API documentation and give it credentials (held in a vault, never in chat or logs). It drafts a private connector for your tenant only, test-reads real data, and shows you a plain-language capability list to confirm before it registers.

PATH C IN DEVELOPMENT

No API at all

Read-only database views for looking things up. File and report imports for systems that only export. And Agent Link for the rest — it works the screen, while writes still wait for your approval. Databases are never written to directly.

Across systems IN DEVELOPMENT

The same order is one order

Agents don't think in systems; they think in customers, orders, tickets and inventory. MatePlug matches those objects across the systems they live in, so “where is my order?” is answered from the whole business, not one database.

CustomerCRM · store · emailOrderstore · warehouseTickethelpdesk · inboxInventorywarehouse · storeOne business-object layerThe same order in your store and your warehouse is one order to the agent
Fig. — Objects matched across systems · design diagram

Two gates before writing

Connected is not the same as trusted

A connection alone never earns write access. Every agent passes a readiness check on your data and knowledge, then a replay of your real history — before it may write anything at all.

Gate 1 · Readiness check

Data access verifiedread test passed
Knowledge coveragedocs & history indexed
Gaps listed in plain wordsyou fix, we re-check

A plain-language report: what's ready, what's missing, what to fix first. DEMO

Gate 2 · History replay

Overall score≥ 85 required
Write-action score≥ 90 required
Mistaken irreversible actions= 0 required

Your past tickets and orders, re-run in a sandbox. Fail the numbers, stay read-only. DEMO

The shared connection base

What already exists — and what we're building

MatePlug's connections run on the same connection base as EDIpipe and the Docklify platform: the connector catalog, the EDI engine and the mapping tooling already exist there and are shared, not rebuilt. What's in development is MatePlug's self-serve layer on top of it — the three-step onboarding, the AI connector builder, the object layer and the agent gates on this page. A connector that works on the base will work for MatePlug; new industry connections are added to the base once and show up on both sides.

Your connection assets stay yours: your system identifiers, mappings and message history export with you. Nothing about connecting locks you in.

Bring us your hardest system

The waitlist asks one question: which system would you connect first? The oldest, strangest one is the most useful answer.

Join the waitlist