Exploration Map v2: problems worth chasing

Revision of Yuval Peleg's Exploration Map after a working pass by Tom. Israel first, US later. A map of where to look and what to learn. Nothing is chosen or ruled out.

Every area is framed as options: what the stack could offer, and what we still need to learn.
Correction to v1: the agent stack has no paying clients yet. GoMobile, OctoMonic's paying client, runs custom systems built from the same parts (price scraping, a live sales dashboard, ERP integration, approval steps). That proves the parts, not the product.
Boundaries: no automated price monitoring for Israeli consumer-electronics retailers (an existing client commitment). Audio listening is not on this map. Retail SMB leads are shared space with an existing retail partnership, to discuss case by case.
Door-openers for any area: review mining (show a business what its customers already say, then how it gets fixed), the hiring signal (job ads show where people are thrown at a problem), and a free lead-research sample.

Areas

Area	What the stack could offer	To learn, and caveats
1. Importers	Order intake from any channel into draft ERP orders. Supplier price and exchange-rate alerts, shelf prices, public competitor prices. Supplier shortlists. Order status answers. Per-shipment document checklists, full compliance help with a licensed partner.	Ask every Priority shop: "do you already have the API module?" WhatsApp pilots run internally on a separate number; customer-facing needs Meta's official API. Leads: Yuval's network, trade associations, order-clerk job ads.
2. Tech companies	Account and pipeline briefs, ticket triage, lead research, inbox routing, a searchable company knowledge base. A managed service that runs existing tools.	Lead with the architecture: isolated machine per client, the client's own AI account, human approval. No SOC 2 or ISO yet. What did past buyers demand before signing?
3. B2C service	Furniture, appliance repair, car leasing, developers' after-sales, solar. Review mining is the method and the opener.	No sample report until an area looks promising.
4a. Customs brokers	As clients: document reading, status answers, back office. As partners: the licensed sign-off that unlocks compliance help for importers.	Which customs software they file through (Shaar Olami), and where re-keying hurts.
4b. Spare parts	Parts lookup from the dealer's own catalogue and stock. Counting and locating parts: barcode scanning, bin maps, photo counting.	A dataset shared across dealers only with each one's consent. Photo counting to be tested first.
4c. Food and cosmetics	Automating per-product compliance work for regulatory consultants, who also become the expert partner.	Find them by asking each importer who handles their compliance today.
5. Retail	Dashboards, data scraping and mining (prices outside consumer electronics), restock orders after approval, stock questions, a daily branch summary.	How leads are handled alongside the existing retail partnership; possibly weighted to larger retailers.
6. E-commerce	WooCommerce, Shopify and Wix. A WhatsApp support agent on store and courier data, listings, reviews, price checks against AliExpress, Temu and Shein.	AliExpress has an official API; Temu and Shein only by scraping, case by case. Which couriers and aggregators stores use.
7. Proof	GoMobile (nationwide phone and electronics retailer, stores plus a WooCommerce shop) runs these parts in production. It came through a family connection.	Look-alikes of the first stack client, once there is one. GoMobile's own look-alikes are inside the boundary.
8. Freight tracking (new)	Client-owned tracking of high-value shipments: one view, SLA proof, live status. Buyers: shippers first, then cargo insurers and logistics firms.	AirTags and Google tags can't be read by a business and alert nearby phones. Commercial trackers with APIs do it (Tive, Roambee; in Israel Hoopo, Wiliot). Insurers already use such data (Parsyl, Loadsure). Drivers must be informed.
9. Hiring: AI proficiency test (new)	A test HR runs as a required hiring step, for any business and any role. The candidate does real work tasks with AI tools on an isolated, logged machine, and HR gets a scored report: how well they prompt, check output and use the tools.	Who buys first: HR teams, recruiting agencies or assessment firms. What HR uses today for skills tests. Keeping it fair and legal (equal-opportunity hiring rules, candidate data under Amendment 13). Demand signal: 1 in 12 US job titles now involve AI, 63% outside tech (Indeed, Jul 2026).

Integrations

ERP: Priority REST API (billed per write, 10,000 transactions a month per package, reads free) and Priority's official AI-agent connector (Cloud, version 26.0+).
Messaging: WhatsApp (internal bridge for pilots, Meta's official API for customers), email, phone and voicemail transcription, SMS, Telegram, web order forms.
Data: exchange rates, supplier price lists, public web scraping, AliExpress API, reviews (Google, Zap, Facebook, G2, Capterra), job-ad sites, supplier directories, paid lead lists later (Apollo, Ocean, BDI, Dun's 100).
Operations: customs software, parts catalogues, barcode scanning, WooCommerce, Shopify, Wix, couriers and Israel Post, freight trackers, applicant tracking systems (Comeet, Greenhouse).
Already in the stack: Google Drive, Notion, Monday, Slack, databases, invoicing. For tech companies: HubSpot, Salesforce, Zendesk, Intercom.

Working together, and questions

Proposed: Yuval carries most of the selling end to end, Tom most of the building, specs and design together. One stack pilot at a time. Structure open.

Any non-compete, non-solicit or ongoing ties with Vetric or past employers? Could Vetric be a client or pilot?
What did security and OSINT buyers demand before signing?
Does current research already use review mining, hiring signals or lead samples?
Which importers can be introduced first?

Sources: Yuval Peleg, Exploration Map (Nov 2026). Priority REST API, Priority MCP server, Apple Find My FAQ, Google Find Hub, Tive, Hoopo, Parsyl, Indeed Hiring Lab.