Software Discovery

Software Discovery is the AI-assisted pipeline that keeps the global catalog growing. It works in two stages: first it discovers vendors (software publishers), then it discovers the products each of those vendors sells. Nothing it finds becomes a catalog entry straight away — vendors and products wait in staging lists on this screen until enrichment succeeds, and a product that enriches successfully is written to the global catalog and approved as part of the same run.

Cloud only

Software Discovery is available on the OpenLM Cloud Platform. On an on-premises deployment the screen is not reachable and SAM Catalog opens on Software Catalogs instead.

OpenLM runs the pipeline automatically on a schedule, so normal use needs no manual start — this screen is mainly for watching what the pipeline is doing and for starting an extra run when you want one.

Overview

The Overview tab summarizes the pipeline and gives you the two trigger buttons, Discover Vendors and Discover Products.

Tile	Shows	Caption underneath
TOTAL VENDORS	Vendors discovered so far	The number of vendor discovery runs behind that total
TOTAL PRODUCTS	Products discovered so far	How many of them were enriched successfully
SUCCESS RATE	Share of enrichments that succeeded	How many enrichments failed
TOTAL API CALLS	AI calls made across every run	—

Recent Jobs lists the latest runs with their Type, Job ID, Status, Started At, and Duration.

Discovered Vendors

The Discovered Vendors tab lists every software publisher the pipeline knows about.

Column	Meaning
Name	The publisher's name.
Aliases	Other names or ticker symbols the publisher is known by.
Categories Processed	The product categories already covered for this vendor.
Discovered At	When the vendor was first found.
Discovery Category	The category the vendor was discovered under.
Headquarters	The vendor's stated headquarters.
Is Included	Whether the vendor is in scope for product discovery.
Product Categories	The categories product discovery covers for this vendor.
Product Discovery Status	Pending, Partial, or Complete.
Source Url / Website Url	Where the information came from, and the vendor's own site.

The toolbar offers Add Vendor, Delete, and Trigger Vendor Discovery, plus Refresh and Search. Select a row with the radio button in the first column to enable Delete, which administrators only see. Hovering a row reveals an Edit Vendor action.

Add a vendor by hand

Add Vendor opens a form for publishers the pipeline has not found. Only Vendor is required. Alias, Discovery Category, Headquarters, Source URLs, and Website URL are optional, but they make the vendor easier to work with later.

Trigger vendor discovery

Trigger Vendor Discovery asks the pipeline to look for publishers in the categories you choose.

1. Tick one or more Categories. Ten are built in:

   CAD and 3D Modeling · Video Editing and VFX · Office Productivity · Developer IDEs and Build Tools · Database and Data Warehouse · Enterprise Resource Planning (ERP) · Cybersecurity and Endpoint Protection · Cloud Infrastructure and DevOps · Business Intelligence and Analytics · Collaboration and Communication

2. To look somewhere else, type a category next to Add custom category and select Add.

3. Set Max vendors per category — between 1 and 30, 15 by default.

4. Select Start. Start stays disabled until at least one category is ticked.

The run is queued rather than executed on the spot, and the dialog closes immediately. Watch it on Job Runs & History.

Discovered Products

The Discovered Products tab lists the products the pipeline found, before they become catalog entries.

Column	Meaning
Software Name	The product's name.
Publisher	The publisher named on the product.
Discovery Vendor	The vendor record the product was discovered under.
Status	The enrichment status: Pending, Enriched, or Failed.
Category	The category the product was discovered under.
Is Package	Whether the product is a suite or bundle.

Status is the column to watch:

• Pending — found, not yet enriched.
• Enriched — enriched successfully and written to the global catalog.
• Failed — enrichment did not produce a usable entry. Failed products stay listed, so you can see what the pipeline was unable to resolve. They never become catalog entries.

Trigger product discovery

Trigger Product Discovery runs the second stage against vendors you pick.

1. Tick the vendors to run against. The list is labelled Categories but holds vendor names, and it only offers vendors whose product discovery is unfinished — so it is empty, and Start is disabled, once every vendor has been processed.
2. Set Max categories per vendor — 1 to 30, 10 by default.
3. Set Max products per category — 1 to 30, 15 by default.
4. Set the API call budget for the run — 200 by default. The run stops when it has used that many AI calls, which is what keeps an over-broad request from running away.
5. Select Start. Start stays disabled until at least one vendor is ticked.

Job Runs & History

The Job Runs & History tab has two sub-tabs.

Active Jobs shows a card per running or queued job, auto-refreshing while you watch. Each card names the job type, its status, and two progress bars: Items (how many vendors have been processed) and Budget (how many API calls have been spent out of the run's budget), plus when the job started or was queued.

Run History lists finished runs — Job ID (shortened to its first eight characters), Status, Type, and Triggered By. Type reads Publisher for a vendor run and Product for a product run. Scheduled runs are attributed to OpenLM's discovery account rather than to a person.

Select a run whose status is Completed to open its detail panel; selecting the same run again closes it. Runs with any other status have no detail panel.

Detail tab	What it holds
Input	The parameters the run was started with — the categories, the per-category limits, and the AI model used.
Results	What the run produced: categories covered, how many were skipped, how many items were discovered, how many were Duplicates of records already held, and the Net New count.
API Metrics	Total Calls, Successful, Failed, and Avg Response time.

Net New is the number worth tracking: it is what the run added once duplicates are discarded.

Where discovered products go

Discovery never writes to your organization's local catalog. Discovered vendors and products sit in their own staging lists — the Discovered Vendors and Discovered Products tabs — and a product only becomes a catalog entry once enrichment succeeds. At that point it is written to the global catalog that OpenLM maintains and approved automatically as part of the same run, so it never appears on Pending Catalogs.

Your organization sees the result on the Global tab of Software Catalogs, alongside every other OpenLM-managed entry.

Related

• Pending Catalogs — the approval queue for the entries that do need review.
• Software Catalogs — the working catalog.
• SAM Catalog overview — concepts and the approval lifecycle.