Independent concept · Not affiliated with Walmart · Unlisted

Walmart taught its agent to shop. Nothing yet proves its stores restored.

Walmart bet on agentic commerce with Sparky, its own AI shopping assistant, and finished Project Gigaton six years early. This concept adds the two things still missing: an open agent surface any AI can read (not just Sparky), and a per-place proof layer under the "regenerative company" pledge: restore work done at a specific store, proven with photo + GPS + signature, settled to a permanent on-chain receipt with its tCO2e. Verified place, verified restored, verified funded.

Read this first. This is an independent, unsolicited concept demonstration by EcoWealth Corporation. It was not requested by, produced with, reviewed by, or endorsed by Walmart Inc. or any affiliate. All observations come from Walmart's public web pages and public DNS only (passive reads, no scanning, no probing, no authentication). Walmart®, Sparky®, and Project Gigaton® are Walmart's marks, referenced descriptively. Every "demo" is a concept mock on EcoWealth's own data.
The whole thing, plainly

We found Walmart shipped its own AI shopping agent, Sparky, but left no open door (the file AI assistants read first, llms.txt) for any other AI to read the store, and its 50-million-acre restore pledge has no per-place proof, only supplier-reported totals.

We already drafted every fix, free: a findings brief, a working demo, and an agent kit with the missing files.

The one ask: try one 10-cent payment call against our live endpoint, a two-minute proof, not a sales conversation.

Four facts from the public record

Sparky, but no door
Walmart shipped an agentic AI assistant (June 2025), yet walmart.com/llms.txt is a true 404 (control-verified). Sparky can read Walmart; no other agent can.
Target ships one
Another major retailer, target.com/llms.txt, returns 200, a real agent map. Walmart's is 404 on both the consumer and the developer domain.
0 × ld+json
The corporate sustainability page carries zero structured data. Project Gigaton's 1.19 Gt is supplier-reported; the 50M-acre pledge has no per-place proof.
p=reject ✓
Credit where due: Walmart enforces DMARC p=reject and publishes a real security.txt, hygiene many peers skip.

The concept, in three pages

Capability brief

An evidence-first read of Walmart's public surface

Baseline credited first: Gigaton, Sparky, and the security hygiene are genuinely strong. Then four evidenced findings: the assistant-without-a-door gap, the missing open agent surface (with the Target contrast), the self-reported-not-verifiable regeneration, and one small HSTS note, each with the exact URL, HTTP status, and re-check command.

Read the brief
Demo, shown, not told

The store-community restore packet

Two acts in Walmart's own blue: Act 1, Sparky plans a cookout and an agent checks out (today, real). Act 2, the store parking-lot bioswale or the creek behind it becomes a restore work packet: authored SDG/311-shaped, claimed by an associate on a step-1 ladder or a local, closed with photo + GPS + signature, settled on-chain with its tCO2e: the per-store proof the 50M-acre pledge is missing.

Run the demo
Agent kit

What agent-legible Walmart would ship

Concept files: an llms.txt that opens the catalog and the store network to every AI reader, an agent card, MCP tool schemas (the standard AI apps call for live actions) for verified store-restore outcomes, and a concierge persona. Drafts, ready to hand to Walmart's own team, the open complement to Sparky.

Open the kit

The proof stack behind this (live today, ours)

Not slides: a running system with reproducible receipts. A live work board reads 7,624 postings / $671,012 posted value at capture across ~90 work types (reproduce: curl https://vealth.net/labor/stats; the counter is the truth, not this page) including place-based restore lanes tree_planting (73), urban_canopy_survey (511), stormwater_maintenance_clearing (48), clean_waterway, build_rain_garden, and native_planting; proof-checked packets (photo+GPS+signature); on-chain settlement on Base: EWP contract 0x76c17C…A14B, first settled packet workId 14; agent-payable tools over x402 (a web-native way for AI agents to pay per use): work-packet $0.10 · work-tco2e $0.005 · work-proof-check $0.01; and a live end-to-end agent walkthrough: the building-agent golden path.
Sparky answers "what should I buy?" This answers "did the store give back?": verified place, verified restored, verified funded. One is the future of shopping; the other is the future of the receipt behind a regenerative pledge.

The ask: put one agent on-chain today

Put one agent on-chain today, the scary part, made small:

  1. Give any AI agent, Claude, a GPT-class model, Llama, or your own Sparky-class agent, a Base wallet holding a few cents of USDC. No account with us, no API key, no sales call.
  2. Point it at a live endpoint: a POST to vealth.net/.well-known/vealth/work-packet answers HTTP 402 with a standard x402 payment schema (scheme: exact · network: base · $0.10 USDC); the agent signs a USDC authorization and pays. A runnable client with a dry-run default is published at vealth.net/recipes.html.
  3. What comes back is a bounded, proof-verified ecological work packet, and every step the agent just took (the payment, the settlement) is an on-chain receipt anyone can verify, forever.

Why this isn't scary: the agent cannot do anything invisible here. Every action embeds its own receipt: permanent, public, verifiable. That is more accountability than most enterprise software offers, not less.

The bet: AI competence is scaling, and Walmart already shipped Sparky, its own agentic AI shopping assistant, betting publicly that shopping is going agentic. Being early on the open side costs $0.10 and one afternoon. Being late means the proof-verified track records belong to someone else.


Then the funded step (pocket change): fund one store-community restore packet at a single Supercenter: a native-tree planting or a creek cleanup, roughly $150–250 of labor plus cents of settlement. Against a company that reports ~$680B in annual revenue and a pledge to protect/restore 50 million acres, that's a rounding error. The deliverable: a permanent on-chain proof receipt: this store, this restoration, this crew, with its tCO2e retired, appended to that store's verified regeneration record. Proof first; the ask is one packet.

Provisioned by EcoWealth · Digital Provisioning Protocol · provenance‑sealed