LivePublic manufacturer directoryVisit live site

Supply Chain Directory

A searchable public directory of 8,000+ U.S. manufacturers. An AI-assisted enrichment pipeline keeps it current, and human review gates what gets published.

Timeline
2026
Duration
6 months
Client
Catalyst Connection
We Make It Here Supply Chain Directory landing page
We Make It Here manufacturer browse page with filters, results, and map
AI-assisted manufacturer enrichment pipeline diagram for We Make It Here
We Make It Here manufacturer profile page

Landing

The public entry point frames the directory as a resource hub for manufacturers seeking entry and growth in emerging markets.

01 / 04

8,000+ manufacturers

Public search, map markers, profile pages, and sitemap surfaces are backed by the live manufacturer index.

Search + map + profiles

Company, capability, certification, NAICS, location, list, and map views work together for supplier discovery.

AI enrichment pipeline

Intake, record resolution, generated profiles, evidence review, and public publishing stay separated by trust gates.

Outcome

The directory is live with 8,000+ searchable manufacturer profiles. Visitors use the semantic search to find suppliers for real business needs, and Catalyst Connection's team runs the directory day to day through the admin portal, generating reports and approving manufacturers to manage their own profiles.

Challenge

Catalyst Connection, a Pittsburgh nonprofit that supports small manufacturers across southwestern Pennsylvania, needed a directory that helps buyers discover qualified suppliers. Their team also needed a scalable way to ingest, enrich, review, and maintain thousands of company profiles. The hardest part turned out to be the data itself, and roughly 60% of the project went to discovery and the aggregation and enrichment problem.

Approach

Two audiences share one dataset. Buyers need fast public discovery, while the internal team needs trustworthy ways to grow and maintain it.

  • Worked directly with the Catalyst Connection team on requirements and testing, while a Kotech partner designer built Figma mockups and demos to validate the direction early.
  • Kept the public route contract stable around the homepage, manufacturer browse, and indexable manufacturer detail pages.
  • Connected the frontend search experience to Supabase-backed card, marker, profile, slug, and sitemap surfaces instead of static snapshots.
  • Built admin workflows for guided data intake, enrichment monitoring, directory review, claim/listing decisions, and workspace access management.
  • Structured AI-assisted enrichment so reviewed evidence, generated profile fields, canonical directory data, and human edits stay traceable.

Public directory

The public experience is built around a stable homepage, manufacturer browse route, and canonical manufacturer detail pages. Browse requests flow through a single frontend API route that can return cards and map markers, while profile pages use Supabase-backed read models for richer SEO and supplier detail.

  • Search supports company names, capabilities, certifications, NAICS, supply-chain filters, energy segments, geography, and location radius, plus embedding-backed semantic search so buyers can describe what they need in plain language.
  • Map markers and card results load independently so the page can keep list results responsive while map data catches up.
  • Manufacturer profiles expose contact, location, capability, compliance, and enrichment-derived fields in a buyer-friendly structure.
We Make It Here manufacturer browse page with filters, results, and map

Search

The browse surface combines dense filters, result cards, and map context without splitting discovery into separate workflows.

Enrichment system

The backend keeps public search separate from enrichment execution. The frontend stays focused on authenticated workflows and review surfaces, while backend services handle enrichment work, queued processing, and profile/search refreshes.

  • Operators can turn approved inputs into bounded enrichment requests from the admin workspace.
  • Job outcomes, matched records, and row-level results stay inspectable for internal review.
  • Generated summaries and classifications are merged with canonical directory fields and human-authored edits instead of silently replacing them.
AI-assisted manufacturer enrichment pipeline diagram for We Make It Here

Pipeline

The enrichment pipeline makes intake, resolution, AI profile building, human review, and publishing separate gates.

Admin workspace

Authenticated routes under /a give admins and verified manufacturers separate operational paths. Admins can manage enrichment, directory visibility, profile edits, user access, and claim or new-listing requests; verified manufacturers get a focused profile-management workspace after approval.

  • Admin directory search includes archive-aware filtering, exports, public profile links, and reversible archive/restore actions.
  • Request review handles existing-profile claims and new manufacturer listing submissions before anything enters the public directory.
  • Workspace access is resolved from Supabase roles, memberships, allowlists, and access blocks rather than mocked client state.

SEO and operations

The directory keeps crawlable manufacturer pages and operational revalidation explicit. Public pages are static or ISR-friendly, sitemap chunks come from the Supabase slug feed, and writes can revalidate affected browse, sitemap, and detail surfaces after directory updates.

  • Manufacturer detail URLs use stable public slug suffixes, with compatibility for older internal IDs.
  • Turnstile protects paid public text-search paths when configured, while blank browse and hard filters remain low-friction.
  • Launch week brought a flood of bot traffic, likely trained on the unprotected old site. Security hardening dropped it to zero within days without adding friction for real visitors.
  • Vercel Web Analytics tracks public traffic while authenticated workspace routes are excluded.

Next project

Aftercare

An SMS platform that lets clinics check on patients after procedures. AI handles the routine replies; anything uncertain or high-risk fails closed into clinician review.

View case study
Aftercare clinician dashboard with SMS conversation and automation controls