Stock + rules as the source of truth

Turn messy inbound requests into structured fitment - based on your stock + rules.

Website chat widget that qualifies requests using your rules and stock. Customers get compatible in-stock options. Your team gets a structured sales summary.

Request demo Use cases
RAG is used to ground answers in your docs (FAQ/specs/policies). Compatibility and availability are validated separately.
Example chat:
Customer:I need solar for a small off-grid cabin. I’m not technical. I want fridge, lights, and Starlink. Budget around $4k.
Agent:Quick checks: 120V only or 240V too? Backup goal ~1 day or more?
Customer:120V only. Ideally 1 day backup.
Agent:Typical energy: ~3–5 kWh/day (Starlink 0.8–1.5, fridge 1–2, lights 0.3–1).
Agent:Any microwave / coffee maker / tools / heater?
Customer:Microwave sometimes.
Agent:Microwave ~1–1.5 kW. 15–20 min/day adds ~0.3–0.5 kWh (within margin).
Agent:Peak check: fridge + Starlink + microwave usually < ~2 kW, so 3 kW inverter has headroom.
Agent:Target storage for 1 day: ~6–7 kWh usable.
Agent:Option A: KIT-OFFGRID-3KW-120V + BAT-7KWH-LFP — $3,890 (in stock).
Agent:Option B: KIT-OFFGRID-5KW-120V + BAT-10KWH-LFP — $5,290 (in stock).
Agent:Under $4k w/ more margin: start with A, add 2nd BAT-7KWH-LFP later (compatible).
Agent:Summary: 120V off-grid, ~3–5 kWh/day, 1-day backup → Option A.

Use cases

Same engine. Different verticals. Each starts with a messy human request.
Customer:Need a replacement AC for a small office. Quiet and efficient.
Agent:Room size and cooling-only vs heat pump? I’ll return compatible in-stock units + required accessories.
Customer:Brake pads for my 2016 Civic. Not sure which ones.
Agent:Confirm trim and rotor size. I’ll return fitment-compatible in-stock SKUs + install notes.
Customer:Need a motor for a conveyor. Something reliable.
Agent:Voltage, phase, power/torque, mounting? I’ll match compatible in-stock motors and the right starter/VFD.
Customer:Pipes and fittings for a bathroom remodel. I don’t know the parts list.
Agent:Material and diameter, number of fixtures? I’ll build an in-stock bundle with compatible fittings + summary.
Customer:I want a desk for a small room. Simple, not bulky.
Agent:Max width and storage needs? I’ll return in-stock desks that fit the space + a clean summary.
Customer:Cat tree for a big cat. My apartment is small.
Agent:Cat weight and max height/footprint? I’ll return stable in-stock options that fit your space.

What it solves

Designed for teams handling high volumes of inbound questions, quotes, and "can you recommend..." requests.

24/7 guided qualification

Customers get the right questions and a consistent first response anytime.

Sell what you have - and what you want to push

Stock-aware options with “sell-first” priorities (preferred SKUs, bundles, exclusions).

Handle more conversations without adding headcount

Parallel intake and qualification, with clean routing to the right owner.

Capture real demand signals

Logs show what customers actually ask for, missing fields, and drop-off points.

How it works

Simple pipeline: extract - validate - match - summarize.
Step 1

Website chat input

Customer describes the need in plain language in the website widget.

Step 2

Intent → structured field

Extracts the minimal required fields (plus nice-to-have) and normalizes terms.

Step 3

Validate hard constraints

Applies your fitment rules / exclusions to prevent incompatible options.

Step 4

Ask for missing inputs

If key fields are missing or a constraint can’t be validated, the agent asks targeted follow-up questions (instead of guessing).

Step 5

Match in-stock options

Selects compatible in-stock SKUs and allowed substitutes, using your sell-first priorities.

Step 6

Output + handoff

Shows options to the customer and produces a clean internal log/summary for routing and follow-up.

Integration

Works with any platform and any catalog quality. We adapt on our side; you connect the minimum required inputs.
WORKS ANYWHERE

Cross-platform integration

Shopify, WooCommerce, custom stack - doesn’t matter. The engine runs on our side; you just connect the widget and data.

ANY FEED

Feed normalization

CSV, XML, JSON, API, Google Sheet. We normalize your feed and keep stock and attributes consistent for fitment.

LOW CODE

Seamless site integration

No-code options are available. Add the widget in minutes and iterate without touching your catalog pages.

Demo run

A practical evaluation: you provide small example data, we return a shareable demo run link.

Get a shareable demo run link based on your example data

Your team can paste redacted real-world requests in plain language and see structured fields + in-stock options (with substitutes when needed) + a sales-ready summary.

Minimal inputs (any one is enough to start)

  • a small stock extract (CSV/Sheet) or a read-only feed
  • 1-2 typical inbound requests (redacted)
  • your basic hard constraints (even a bullet list)

What you'll see in the demo run

  • extracted structured fields + confidence/missing inputs
  • rules-checked compatibility decisions
  • in-stock options + substitutes (when needed)
  • sales-ready summary for routing and handoff
Request a demo run link See the pipeline

Trust

Designed to reduce risk and avoid "hallucination-driven" outputs.

Operator controls

You control constraints, priorities, and the questions asked - without rebuilding your site.

Stock-aware by design

Options are filtered by availability and sell-first priorities (preferred SKUs, bundles, substitutes)

No guessing when data is missing

If key inputs are missing or a rule can’t be validated, the agent asks targeted questions before proposing options.

Logged and auditable

Every session produces a structured record: intent, extracted fields, rule checks, and options shown.

Contact

Get a demo run link based on your example data.

Request a demo

Send a short note with any of: stock extract/feed, 1-2 redacted requests, or your hard constraints. We'll reply with a shareable demo run link.

Email→b2b@airankforge.com
Subject→Demo run link request
Include→CSV/Sheet OR feed + 1-2 requests

Founder

Vadim Donskoy