New ResearchCorpBenchSee the results ↗We tested AI models on real GTM tasks.
The Spice House
Food & specialty retailOperations

Warehouse availability and production planning

The Spice House keeps channel availability and marketing in step with upcoming restocks.

Agents investigate stock discrepancies, decide which sources to recheck and use browser access where APIs are unavailable. Verified findings guide channel allocations and restock updates, helping operations reduce order delays and marketing plan ahead.

Summarize with AI

The impact

68%

less time reconciling channel availability

Warehouse availability and production planning

Case study facts

Team
Operations
Systems connected
3
Time to live
3 weeks
Engagement
AI implementation

The impact

Results for the team

less time reconciling channel availability
68%
less time reconciling channel availability
Operations spends less time reconciling warehouse balances with online availability and channel allocations.
fewer stock-related order delays
27%
fewer stock-related order delays
Channel updates reflect eligible stock, reservations and fulfillment rules, reducing order delays caused by stock availability mismatches.
of visibility into planned restocks
3 weeks
of visibility into planned restocks
Marketing can see planned restocks across a three-week horizon and adjust campaign timing when production schedules change.

Situation

Stock in a warehouse is only part of the picture.

Stock across warehouse locations, availability shown online and allocations to other channels can tell different stories. Meanwhile, planned production holds the next restock dates. Operations and marketing need those views connected to understand what can sell now and what is coming next.


Challenge

Channel availability and restock plans need to move together.

A product can appear unavailable online while eligible stock sits in another location, or remain promoted when its channel allocation is running low. Production dates can also shift. The agents need to decide which source to check next when records disagree, including systems without APIs. They must adjust channel availability without counting the same stock twice and give marketing timely notice of planned restocks and delays.

What we built

Agents investigate before updating availability.

We built agents that investigate differences between warehouse stock, reservations and channel availability. They choose the relevant sources, use APIs where available and operate systems through Browserbase where APIs are unavailable. When findings are stale or conflicting, they decide what to recheck and reassess the evidence.

Stock calculations and allocation rules validate updates based on the checked findings. The agents also check planned production quantities and completion dates, keeping expected restocks separate from sellable stock. They revisit sources when dates change and notify the team of upcoming restocks or delays so marketing can adjust campaign timing.

AI tech stack

  • Cloudflare Agent CloudAgent infrastructure
  • OpenRouterModel access
  • BrowserbaseBrowser execution
  1. Choose and access sources

    Select the warehouse, reservation and channel records needed for the check. Use APIs where available and Browserbase for systems without APIs.

  2. Investigate and recheck

    Compare the findings, decide which stale or conflicting source needs another check and reassess the evidence.

  3. Update verified allocations

    Use checked findings to adjust channel availability within stock, reservation and fulfillment rules; send unresolved discrepancies to operations.

  4. Reassess production plans

    Check planned quantities and completion dates, revisiting the relevant source when restock timing changes.

  5. Notify the team

    Share checked restock changes and unresolved availability issues with the affected products and channels.

  6. Coordinate marketing

    Give marketing the stock and timing context to adjust promotions, campaign dates and back-in-stock messaging.

Evals and guardrails

Reconcile stock. Prevent duplicate allocation.

  • Stock evals

    Replay stock and reservation changes against expected totals. Test whether agents choose relevant source rechecks when records conflict, including outdated records and systems accessed through the browser.

  • Schedule evals

    Measure missed production-date changes and false alerts. Test delayed restocks, repeated updates and notifications arriving out of order.

  • Guardrails

    Check stock totals and record versions before updates; prevent duplicate allocation. Bound source rechecks and send unresolved mismatches to operations. Keep planned production separate from sellable stock.

Results

Teams can act on available stock and upcoming restocks.

Operations spends less time reconciling warehouse balances with online availability and channel allocations. Channel updates reflect eligible stock, reservations and fulfillment rules, reducing order delays caused by stock availability mismatches.

Marketing can see planned restocks across a three-week horizon and adjust campaign timing when production schedules change.

less time reconciling channel availability
68%less time reconciling channel availability
fewer stock-related order delays
27%fewer stock-related order delays
of visibility into planned restocks
3 weeksof visibility into planned restocks

From The Spice House

“An invaluable resource to have on your side. They work quickly, find solutions tailored to how the business actually runs, and consistently deliver high-quality work. Beyond the results, they are a pleasure to work with.”
Allyson LewisCEO at The Spice House

Build capacity to grow. Own your intelligence.

Bring the workflow slowing your pipeline, campaigns or renewals. We'll map the bottleneck and show you the AI agent concept.