These reports were produced by asking Claude a plain-language question against a live Sync database through the Model Context Protocol - no exports, no report builder, no BI project. Every figure below comes from the Sync demonstration dataset. No customer data appears on this page.
None of the analysis below is intellectually difficult. Every one of these reports is something a competent analyst with database access could have produced. The reason they usually don’t exist is that assembling numbers across invoices, stock, sales orders and credit notes takes days, so it happens at a quarterly review rather than in the middle of the season when it is still actionable.
What changes is the cost of asking. Each report started as a sentence typed in plain language; Claude queried the live Sync database through MCP and returned a structured answer with every figure traceable to the query behind it. The reports run in minutes, which means the questions get asked in-season, which is the entire point.
Each began as a question typed in plain language. Figures are from the Sync demo brand - a synthetic apparel label running wholesale and D2C.
“How are we tracking this year against last, and where did it move?”
The headline is a decline; the structure underneath is more interesting. Units fell faster than revenue because average selling price rose 25.5% - a mix shift, not a pricing win, and worth knowing before anyone builds next season’s buy on the revenue number alone. Quarter over quarter the year is recovering (+29% Q1 to Q2) but still trails the prior year heading into peak. By category the split is sharp: one-piece up 127.9% and swim tops up 38.9%, while boardshorts fell 19.5% and T-shirts 22.4%. Two categories growing fast off a small base, two large categories carrying the decline.
YoY sales by division and category, Jan to Jun. Flag the biggest movers.
Claude · MCP| Category | 2026 | 2025 | Change |
|---|---|---|---|
| One Piece | $96.7K | $42.4K | +127.9% |
| Swim Top | $204.7K | $147.4K | +38.9% |
| Boardshorts | $2.50M | $3.11M | −19.5% |
| T-Shirts | $922.9K | $1.19M | −22.4% |
| Short Sleeve | $1.99M | $3.41M | −41.6% |
Which top sellers are about to run out? Rank by weeks of supply.
Claude · MCP| Style | YTD units | Units/wk | On hand | Wks supply | |
|---|---|---|---|---|---|
| 006362 | 3,838 | 137 | 85 | 0.6 | CRITICAL |
| 006253 | 2,682 | 96 | 76 | 0.8 | CRITICAL |
| 006229 | 1,875 | 67 | 61 | 0.9 | CRITICAL |
| 006311 | 9,041 | 323 | 1,057 | 3.3 | CRITICAL |
| 006192 | 2,112 | 75 | 396 | 5.3 | TIGHT |
Style 006311 is the number-one seller year to date and holds 3.3 weeks of cover.
“Which wholesale accounts grew, which shrank, and who stopped ordering?”
Growers and decliners are the easy half. The column that changes behaviour is the fifteen accounts that have gone dark - customers who bought last year and have placed nothing this year, which no sales report surfaces because an absence produces no row. The top grower added $51.6K on an 83.4% increase; the point of the report is that a rep can see both lists before the next territory review rather than after it.
“Which accounts are in trouble - and not just the obvious ones?”
This is the report that is hardest to reproduce by hand, because the signal spans tables. Declining revenue is visible on any sales report. Four of these eight accounts were flagged for something else: cancellation value exceeding actual invoiced sales. An account that keeps ordering and keeps cancelling looks healthy in the order book and is quietly consuming allocation, production slots and credit. One further account was flagged for outsized returns against volume. D2C channels were excluded so the wholesale signal is not diluted.
Which wholesale accounts are cancelling more than they buy?
Claude · MCPThe 1.8× account is the largest wholesale customer on the book. Ratio, not revenue, is what surfaces it.
“What stock should we be marking down right now, and what is it costing us to hold?”
Run mid-peak-season, filtered to styles still active in the system rather than discontinued - the distinction matters, because these are not legacy lines nobody expected to sell. The worst single position carries around twenty-one years of supply at current velocity: 923 units, $85K, eleven sold in the last quarter. Nobody decided to hold that. It accumulated, and no report existed to name it.
What stock should we discount now? Rank by dollars tied up.
Claude · MCPStyle 008366 is the largest single stock position in the warehouse by unit count, with no invoiced sales in twelve months.
“What is actually sitting in the warehouse, and is it there for a good reason?”
Ranked by units on hand, then cross-checked against sales velocity so that high stock is labelled healthy or a warning rather than left ambiguous. The largest single stock position in the warehouse - 4,505 units - had recorded zero sales in twelve months. Seven of the twenty largest positions were dead stock and five more were multi-year supply, meaning the majority of the warehouse’s biggest holdings were there for no current commercial reason.
“What sells where? Rank the top ten by units for each coast.”
A regional split of the same catalogue, ranked by units and reported with revenue alongside, because the two orders differ - the highest-unit style is rarely the highest-dollar one. For a brand allocating limited inventory across regions, or planning a rep’s trip, the question is trivially easy to ask and was previously slow enough to answer that it mostly went unasked.
Sync demonstration dataset, July 2026. Synthetic brand, synthetic accounts, no customer data.
The data model is apparel-native. Every report above depends on it. A question about a broken size curve is unanswerable on a system that flattens a size run into independent SKUs - the fact that mattered was never stored. Sync holds styles, colourways, size runs, prepacks, seasons and forward availability as first-class structures, which is why the analysis has something worth reading.
Your account, your keys, your permissions. You connect your own Claude account to your own Sync instance. No third-party data store sits in the middle, no separate AI contract is required, and every query runs scoped to the permissions of the person asking - a sales user cannot pull margin data they could not open in Sync itself. That is what makes this deployable rather than a demo.
Every number traces back to its query. Each figure on this page can be followed to the exact request that produced it. That matters less on the impressive first run and enormously on the Tuesday when a number looks wrong and somebody has to establish whether the analysis erred or the underlying record did. Auditable output is the difference between a report you can act on and one you have to double-check by hand.
No. Every figure here comes from the Sync demonstration dataset - a synthetic apparel brand with wholesale and D2C channels, built to exercise the same structures a real brand has: styles, colourways, size runs, seasons, cancellations and credit notes. No customer data appears on this page. On a demo the same reports run against your own data.
Claude connects to Sync through a native Model Context Protocol server and queries the live database using tools like query_invoices, query_stock, query_sales_orders and query_credit_notes. No export, no data warehouse, no separate AI contract - the brand connects its own Claude account, queries run scoped to the asker's permissions, and every figure traces back to the query behind it.
Minutes rather than days. The work these replace isn't intellectually hard - it's assembling numbers across several tables, which is why it usually waits for a quarterly review rather than happening in-season when it's still actionable. The value is that the cost of asking has fallen far enough that people ask.
Yes. These aren't fixed templates - they're answers to questions typed in plain language, and the same connection answers whatever else you ask of the data in Sync: a specific account, category, season, warehouse or rep. Once the shape is right, a report can be re-run on a schedule.
Every figure traces back to the exact query that produced it, so any number on a Sync report can be followed to source and checked in seconds rather than rebuilt by hand. Reports also flag their own weak signals - a week-over-week rule will mark most of a wholesale catalogue as collapsing, because wholesale ships in batches, so the report says so and points to the four-week trend instead. You get analysis you can verify, not a black box with a confident tone.
No. You connect your own Claude account to your own Sync instance. There's no third-party data store in the middle and no separate AI vendor relationship, and queries are scoped to each user's existing permissions - a sales user can't retrieve margin data they couldn't open in Sync itself.
No generic slides. Bring a question you have been meaning to answer - a category, an account, a season - and we’ll run it live against your data in the demo.