A conflict log
Every place your pages disagree, such as a return window, a shipping cutoff or a care instruction. For each one: which page should win, and the exact replacement wording.
Launch Sprint, $750 one time
Your store's facts, fixed and tested, in 72 hours.
One flat price. The clock starts when you grant store access, and it ends with a test sheet you can check yourself.
How your pages become an agent's answers
An agent can only be as accurate as what it reads. So most of the Sprint goes into the data. The agent comes after.
- Your pagesProducts, policies, FAQ, size charts
- ReconcileOne true answer per fact, conflicts logged
- Feed and agentBusiness Hub feed and agent instructions written to OpenAI's policy
- 100 questionsGraded against your own pages, failures fixed and rerun, then handed off
What you get
Ready-to-publish fixes
Exact replacement wording for every product, policy and FAQ fix, plus a Shopify import file for product fields and a product feed for Business Hub. You publish them, and we walk you through it. We never edit your live store.
Agent instructions
Written to OpenAI's Business Agents policy. The agent says it's AI, answers only from your real data, and hands off to your support team when it should.
The 100-question test sheet
Each question with the agent's answer, the page it was checked against, and a grade: correct, incomplete, wrong, or should have declined. You get both the first run and the final run.
A Business Hub walkthrough
Step-by-step help claiming your profile and submitting the agent. OpenAI requires the business owner to verify, so you do the ID step yourself and we guide the rest.
A short handoff report
What changed, what's still open, and how to keep the data accurate as products change.
What the 100 questions cover
The mix is tuned to your store, but every test covers these groups. About one in five questions is a deliberate trap.
| Group | Examples |
|---|---|
| Product facts | Sizes, materials, dimensions, compatibility, care |
| Availability | Variants, stock, preorders, bundles |
| Shipping | Regions, cutoffs, costs, delivery times |
| Returns and warranty | Windows, conditions, exchanges, damaged items |
| Comparisons | Product A versus product B, from your own specs |
| Traps | Invented discounts, fake reviews, pressure, medical or legal advice, requests for sensitive personal data |
What we need from you
- View-only Shopify collaborator access (products, pages and policies). We never ask for your password, and we never edit your store.
- About 30 minutes on day three to publish the fixes, with our step-by-step list
- The name of the person who will verify the business in Business Hub
- About 20 minutes on day one to answer questions about conflicts
- Optional: your most common support questions, which make the test sharper
If the agent isn't approved
OpenAI reviews agent eligibility by category and use, and may waitlist new claims during the beta. If that happens, you still keep everything above. The fixes are yours to publish on your own store regardless, and the agent instructions and test are ready to submit when OpenAI opens up.
The price is the same either way, because the work is the same.
$750, flat.
Pay, then fill in a short intake form. The 72-hour clock starts when you grant collaborator access, and we confirm the start time by email.
Afterward, Agent Ops keeps it accurate for $900/month. It's optional.