AI supplier intelligence for procurement

Every SOW already describes its supplier shortlist. We extract it — with evidence.

SpiderNet converts an uploaded Statement of Work into structured, source-cited requirements, matches them against a supplier database enriched with real tender history, and produces an explainable shortlist where every claim is backed by evidence. Minutes and hours instead of days and weeks.

Two categories. Your real SOWs. A three-month pilot to a verdict.

One SOW, processing
110 kV substation reconstruction Uploaded · 78 pages · mixed AZ/EN · parsed in 2 minutes
Technical capabilities 14 requirements · 8 mandatory
Project experience ≥ 3 comparable in 5 years
Certificates & licenses 5 mandatory requirements
Delivery & geography 18 months · local support
Structured requirements Extracted with source citations, confirmed by you
Evidence-backed shortlist 10–20 suppliers with proof per requirement
Explainable ranking Every score traceable to its source — and never for sale
The problem

Both sides of sourcing operate on incomplete information

Supplier databases are large but full of stale, self-reported data. Qualification is still done by hand — in Excel and in the memory of individual buyers.

Procurement teams

Have the database — not a reliable answer.

  • ✕Supplier data is fragmented, outdated and mostly self-reported.
  • ✕Technical qualification is manual every time, and the same suppliers get re-researched again and again.
  • ✕Supplier selection depends on individual buyers’ memory and legacy relationships.

Qualified suppliers

Have the capacity and the certificates — no way to be seen.

  • ✕Invisible outside the tenders they happen to enter.
  • ✕Rebuild the same qualification paperwork for every single buyer.
  • ✕No way to prove capability to a buyer who has never met them.

Buyers cannot verify; suppliers cannot be discovered.

Supplier directory

The live supplier directory is the product surface buyers return to

SpiderNet first makes the market visible: approved, known and newly discovered suppliers are assembled into one evidence-backed directory. A profile is not just a name — it carries contacts, capabilities, experience, certificates and source confidence.

The shortlist comes from the directory; the directory improves with every SOW and every source check.

Browse the directory
HVAC and electrical pilot directory 100 candidates · 30 contact-enriched · source confidence labeled
HVAC contractors Project work, service, ventilation, cooling
50 candidates
Electrical contractors Substations, panels, cable, testing
50 candidates
Contact enrichment Phone, email, website, address, source URL
30 profiles
Identity Deduplicates companies and name variants

Legal name, tax ID, branch, brand and trading name are separated so the same supplier does not appear three times.

Contacts Phone and email are prioritized for B2B outreach

The minimum useful record is company, category, phone or email, source URL and last verified date.

Evidence Every capability is tied to a source

Tender results, accreditation records, company websites, directories and manufacturer lists are stored with separate confidence.

AI scope builder

Turn a short brief into a scope of work

Describe an HVAC maintenance need in a few lines. The builder asks only for what is missing, shows you the open items it could not resolve, and hands back a scope and a requirements file you can download.

Open the scope builder

HVAC maintenance today. Runs in your browser, the draft stays in this tab, and you can work through it without AI if you prefer.

  1. 1

    Describe the need in your own words

  2. 2

    Answer only the gaps the AI finds

  3. 3

    Download the scope and the requirements

From the directory

The market is already in the directory

Every profile is assembled from public sources, and every claim is tied to one. The examples below are examples, not recommendations.

21categories
543suppliers
543with contact details
Browse the full directory See your company? Claim the profile.

Examples are chosen alphabetically from the rows whose contact details we resolved. This is not a quality ranking — the directory does not rank.

The approach

The shortlist is already written inside the SOW

Every SOW effectively describes who qualifies: technical requirements, experience, certificates, financial thresholds. We don’t guess any of it — we extract it from the document, with citations.

Statement of Work · extracted requirements

110 kV substation reconstruction

78 pages · mixed AZ/EN · scanned tables included

Power transformers 110/35/10 kV 2 units
ISO 9001 and ISO 14001 certification mandatory
Comparable projects, last 5 years ≥ 3
Annual turnover ≥ ₼5 mln
Delivery and commissioning 18 months
Technical capabilities Products, technologies, equipment
Experience Projects, industries, references
Qualifications Certificates, licenses, authorizations
Commercial & financial Turnover, insurance, contract scale
Delivery Geography, schedule, capacity, support
Compliance & risk Sanctions, HSE, regulatory requirements

We don’t guess requirements — we extract them, with citations.

The product

Not another supplier directory — an SOW-to-shortlist matching product

The enriched supplier database is the substrate. Three outputs are what the customer actually buys.

📄

Structured requirements

Requirements extracted from the SOW with section and page citations and mandatory/optional status. No mandatory criterion excludes anyone until you confirm it.

🎯

Evidence-backed shortlist

10–20 candidates drawn from approved, known and newly discovered suppliers — with proof, gaps and contradictions listed per requirement.

🧾

Explainable recommendation

Not one opaque percentage: requirement-by-requirement status, evidence sources, risk flags and missing information — in an audit-ready form.

The database is the substrate. The evidence-backed match is the product.

How it works

AI does the research; you make the decision

Human control is preserved at every step: no supplier is excluded by a criterion you didn’t confirm, and no recommendation ships without a source.

1
You

Upload the SOW

PDF, DOCX or scan — in Azerbaijani, Russian or English, mixed-language documents included.

2
SpiderNet

Extracts requirements with citations

Every requirement is tied to its source section and page, with type, importance and extraction confidence.

3
You

Confirm the mandatory criteria

Only a human-confirmed requirement can become an exclusion rule.

4
SpiderNet

Discovers, enriches, qualifies

Search across three pools — approved, known and new suppliers; deep research is spent only on the strongest candidates.

5
You

Approve the shortlist

Inspect the evidence, add or remove candidates, and hand the approved list to your sourcing event.

The AI never decides on its own — it makes your decision faster.

The engine

Similarity is not qualification

Semantic similarity is only the beginning. Every recommendation must clear hard qualification gates first — and commercial promotion never touches the ranking.

Every recommendation clears these gates

  • ✓Human-confirmed mandatory criteria
  • ✓Certificates, licenses and their validity
  • ✓Sanctions and compliance screening
  • ✓Capability fit on the taxonomy
  • ✓Semantic match on real project experience
  • ✓Financial and capacity thresholds
  • ✓Geography and delivery requirements
  • ✓Configurable, evidence-weighted scoring

Qualification first. Similarity second. Payment never.

Every attribute carries provenance

The status of every fact is shown openly: verified, supplier-declared, externally observed or AI-inferred. An AI inference is never dressed up as a verified fact.

Real contract and tender outcomesTender participation historyCertificates & validityProject referencesRegistry and sanctions dataSupplier-declared profileWebsite-observed claimsBuyer feedback
The funnel

Deep research only where it pays off

Instead of continuously enriching an entire database, research deepens stage by stage against live demand. The result: speed, low cost and data that is always fresh.

Stage 1

Broad discovery

~100 candidates across three pools: approved, known and new suppliers.

Stage 2

Low-cost enrichment

Registries, sanctions and internal history narrow the list to ~30.

Stage 3

Deep enrichment

Websites, documents and references — for the strongest candidates only.

Stage 4

Qualification

10–20 suppliers assessed requirement by requirement.

Stage 5

Shortlist

Ranked, explainable, ready for your sourcing event.

Depth where it matters — instead of a stale database everywhere.

The difference

Search shows similarity. Evidence proves qualification.

Traditional databases answer “who do we already know?”. SpiderNet answers a different question: who can actually deliver this specific scope — and what evidence supports it?

Directory model SpiderNet
The database answers: which suppliers do we know? SpiderNet answers: who satisfies this specific SOW?
Self-reported profiles Evidence from real tenders, contracts and deliveries
One opaque match percentage Requirement-by-requirement status with sources and gaps
A category search when the buyer remembers Automatic extraction the moment the SOW arrives

Any search can find similar profiles; only evidence finds qualified ones.

Who it’s for

Built for buying teams; open to suppliers

Each role gets its own outcome: the research grunt work goes to the machine, the decision stays with the human.

Buyer

Procurement specialists

Upload the SOW, confirm the requirements, approve the shortlist — hours instead of weeks of manual research.

Buyer

Category managers

Taxonomy, qualification templates and scoring weights in your hands — with a full market view of your category.

Buyer

Vendor management

A living Golden Record: duplicates merged, provenance on every attribute, contradictions surfaced.

Supplier

Suppliers

One claimed profile and reusable qualification documents — visible to every buyer on the platform.

Plans

Buyers subscribe; suppliers join free

Network completeness is the point, so supplier participation is free. And ranking and qualification results are never for sale — to anyone.

Supplier — Free

Coverage. Make the network complete and correct.

  • ✓A profile pre-built from tender history — claim it
  • ✓Capabilities, certificates and project references
  • ✓A correction and dispute channel for your data
  • ✓Qualification documents uploaded once, valid with every buyer
  • ✓Position based on evidence only — no paid boosts exist
Claim your profile

Buyer — Platform

Scale. Every category, the whole team.

  • ✓Annual subscription per organization (tenant)
  • ✓Unlimited sourcing cases in enabled categories
  • ✓ERP and e-procurement integrations
  • ✓Tenant-private data, shared supplier intelligence
  • ✓Configurable taxonomy, rules and weights
Talk to us
Trust & data

Shared intelligence; private data; provenance on everything

On a multi-tenant platform the boundaries are strict: the supplier layer is shared, your commercial information is yours alone.

🌍

Shared supplier layer

Company identity, capabilities, certificates and public tender outcomes — one profile for every buyer. Claimable and correctable.

🔐

Tenant-private data

Your contracts, prices, evaluations and decisions never leave your tenant.

🧠

Provenance on every attribute

Every fact has a source, a date and a status. An AI inference is never presented as a verified fact.

The pilot

Validate on your own cases, against your own metrics

The pilot runs in two categories on your real SOWs and is measured on a golden set built from your historical cases: extraction accuracy, recall of known-good suppliers, net-new candidates and acceptance rate. Targets are agreed up front — so the verdict leaves no room for argument.

2 Pilot categories
≥15 Real procurement cases
10–20 Suppliers per shortlist
3 Months to a verdict
North-star metric

An evidence-backed shortlist your buyers accept — in half the hours of the current process.

SpiderNet
spai-der-net
the web feels the event
The name

Where the name comes from

A spider does not patrol its web looking for something to happen. It sits still and reads the strands — one disturbance, and it knows exactly where in the structure the event occurred and what to do about it. A new SOW is that disturbance. SpiderNet is the web: every supplier a node, every tender a strand, and every piece of evidence a direction that carries the signal to exactly the right place.

Get started

Start with your own SOWs

Send us two or three recent Statements of Work and we will come back with the extracted requirements, the evidence-backed supplier shortlist behind them and a pilot scope.

  • ✓No integration needed to start — uploading the document is enough.
  • ✓No mandatory criterion excludes anyone until you confirm it.
  • ✓Success is measured on your own baseline: hours, coverage, acceptance rate.