# BuyCast.ai https://buycast.ai/ Forecasting, purchasing, and profitability for Amazon FBA operators — with an AI copilot that actually places the orders. Company: Legend Software, LLC Sales: sales@buycast.ai Support: support@buycast.ai --- # Features https://buycast.ai/features ## AI Copilot https://buycast.ai/features/ai-copilot Ask anything about your catalog. Get answers grounded in your real sales, forecasts, and inventory — not hallucinations. When you say go, it drafts the POs. A fully agentic analyst with read access to 27 data tools, 9 staged mutation tools, and a sub-agent orchestrator that can fan out across an entire brand in parallel. Capabilities: - Brand-scoped batch analysis: "What needs ordering for Brand X this week?" triggers a sub-agent orchestration across every SKU, proposes vendor-consolidated draft POs, and waits for your confirmation. - Incremental PO building: "Add 200 of SKU-123 to the draft cart." Multiple vendors at once. Auto-populates unit cost from your product data. Every change is reversible. - Grounded explanations: "Why is SKU-456 trending down?" pulls 14d sales, buy-box share, the competitor-pricing landscape, and profit trajectory — and cites every number. - Per-context memory: Each SKU and each page keeps its own conversation. Auto-swapped as you navigate so the Copilot never loses the thread. Differentiation: - Not a chatbot with tool calls: A fully agentic analyst with access to 27 read tools, 9 mutation tools, and a sub-agent orchestrator that analyzes SKUs in parallel when you ask about a whole brand. - Every action is staged: The Copilot builds drafts. Nothing is sent to Amazon or your vendors until you explicitly approve. Full audit log, every request reversible. Under the hood: - Multi-provider (Claude + GPT) with auto-failover - Extended thinking on Pro tier for deep-analysis mode - Deterministic intent classifier routes ambiguous asks to the right tool chain - Per-tenant RLS enforcement on every tool call ## Purchasing Agent https://buycast.ai/features/purchasing-agent Weekly, SKU-by-SKU. Forecast-driven quantities. Vendor-consolidated. Confidence-scored. Kill-switched. A six-stage decision pipeline that runs continuously in the background — and hands you draft POs in your cart every Monday morning. Capabilities: - Six-stage decision pipeline: Demand → supply → gap → action → priority → consolidation. Each stage is inspectable, so you understand exactly why a quantity was proposed. - Buy-box adjusted demand: Optionally scales demand when you're losing the buy box so competition decline doesn't mask real demand. - Return-rate adjustment: Nets out typical returns so you don't over-order high-return SKUs. - Assembly-aware: Detects when you need a Work Order instead of a PO. Consolidates component POs across multiple assemblies. - Guardrails you control: Max PO value, max days of supply, min stock trigger, new-SKU history requirements. Every guardrail logged, every override reviewable. - Auto-draft threshold: High-confidence recommendations auto-create draft POs. You approve with one click. Differentiation: - Not a reorder formula in a spreadsheet: The agent explains every decision. You see the forecast, the realized APE of that forecast, the buy-box rate, and the guardrail that kicked in. No black box. - Confidence-scored recommendations: Each recommendation carries a confidence score derived from realized forecast accuracy. Low-confidence SKUs stay in review; high-confidence ones can auto-draft. Under the hood: - Runs hourly in the background, surfaces a digest weekly - Vendor-consolidated PO drafts with minimum-order-quantity awareness - Integrated with your forecast accuracy telemetry — auto-downweights untrustworthy forecasts - Every action audit-logged and reversible up to the approval moment ## Demand Forecasting https://buycast.ai/features/forecasting Ensemble of 5 proven models. Seasonality without trend extrapolation. Realized-APE tracking so you know when the model earns your trust. Statistical rigor meets production discipline. Every forecast comes with the receipts of how well that same method has performed on that same SKU, historically. Capabilities: - Ensemble of proven models: AutoETS, AutoARIMA, AutoTheta, AutoCES, and MSTL on every plan. CrostonOptimized for intermittent SKUs. - Change-point detection: Recent regime shifts dominate fitting — but full history still drives seasonal profile. Your launches, stockouts, and price changes no longer poison the forecast. - Daily shadow forecasting (Pro): A 20% sample runs every candidate method every day so method A/B testing is continuous, not a one-time bake-off. Available on the Pro tier and above. - Realized-APE telemetry: `forecast_actuals` continuously tracks realized accuracy per SKU per horizon. Your purchasing agent auto-downweights untrustworthy forecasts. Differentiation: - Not a black-box ML model: Classical statistical methods for 95% of your SKUs — reproducible, auditable, explainable. Neural models only where they measurably win. - Accuracy you can verify: Every forecast is scored against realized demand. You see, per SKU, whether the model has earned your trust. Under the hood: - Runs on a dedicated batch worker with parallelized SKU processing - Method selection per SKU based on rolling realized APE - Seasonality profile derived from full multi-year history; trend from recent regime - Integrated with buy-box share and stockout history for demand correction ## Profitability & Capital Efficiency https://buycast.ai/features/profitability Settlement-sourced fees. Estimated COGS with full override chain. Return-adjusted. Capital-efficiency quadrants so you see which SKUs actually deserve more cash. The P&L you trust when you're deciding what to reorder — not the one you squint at 60 days after the damage. Capabilities: - Settlement-based fees: Amazon V2 settlement reports (~1-day lag) rather than fee estimates. Referral, FBA, storage, removal, promotional rebates — all pulled directly. - Cost layering with full override chain: Manual override → repricer override → product cost → last-purchase cost → rolling average. Every override attributed. - Capital efficiency metrics: Inventory value, annualized ROI, turns, days-to-recover-capital, excess vs 45-day supply. - BCG-quadrant analysis: Star / Cash Cow / Question Mark / Capital Trap. Sort your catalog by capital efficiency, not vanity revenue. Differentiation: - Settlement-sourced, not estimated: Most tools use Amazon's fee estimates. We reconcile against the settlement reports so the number you see is the number you got. - Per-SKU capital lens: Revenue tells you what sold. Capital efficiency tells you what to do next. The dashboard leads with the latter. Under the hood: - Daily settlement report sync + reconciliation - Return adjustment against rolling 90-day return rate - Days-to-recover-capital = inventory value ÷ daily contribution margin - Quadrant rules user-configurable (percentile thresholds) --- # Pricing https://buycast.ai/pricing ## Growth For wholesalers ready to put their purchasing on autopilot. Price: $599/mo SKU limit: Up to 500 catalog SKUs User limit: 5 user seats CTA: Start 14-day trial - Up to 500 catalog SKUs - 5 user seats - Unlimited brands - AI Account Manager — weekly LLM purchasing review on every SKU - Auto-drafted Purchase Orders with full supplier context - Work Orders with BOM deduction and lifecycle tracking - Repricer — real-time, rule-based - Marketplace data — competitive offers, demand history, seasonality - Real-time inventory and sales - Standard demand forecasting - AI Copilot with action tools - Daily Business Insights digest - Profitability + capital efficiency views - Returns, reimbursement chase, SAFE-T - QuickBooks Online sync - Buy box tracking with real-time poller ## Pro For serious wholesale operators with deeper catalogs and multi-buyer teams. Price: $1499/mo SKU limit: Up to 1,500 catalog SKUs User limit: 15 user seats CTA: Start 14-day trial - Everything in Growth, plus: - Up to 1,500 catalog SKUs - 15 user seats - Sub-agent orchestrator — batch analyze dozens of SKUs at once - Extended thinking + refinement pass — deeper reasoning on AI dossiers - Choose your AI model (Claude or GPT) ## Enterprise For $25M+ operators, multi-entity sellers, and aggregators. Price: Custom pricing SKU limit: Unlimited SKUs User limit: Unlimited seats CTA: Talk to us - Everything in Pro, plus: - Unlimited SKUs and seats - Multi-entity workspaces — separate Amazon accounts under one billable customer - White-label / agency mode - SSO, custom RBAC, SOC2 documentation - Dedicated customer success manager - Premium SLAs ## Pricing FAQs Question: How does the AI Account Manager actually work? Answer: Every Monday morning, an AI agent runs through your catalog SKU-by-SKU. For each one, it pulls in 14 sections of context — velocity, inventory, lead time, capital, returns, buy box, seasonality, and more — and produces a recommended purchasing action with full reasoning. Results feed directly into draft POs ready for your review. You spend your Monday reviewing decisions, not building spreadsheets. Question: How much time does it actually save? Answer: Most customers come from a workflow where someone (usually the owner or head buyer) spends 4–8 hours every Monday morning building reorder spreadsheets. With the AI handling the analysis, that drops to 30 minutes of reviewing the AI’s recommendations and approving the POs. For larger catalogs (1,000+ SKUs), the time savings is even more dramatic — most operators with that catalog size can’t review everything by hand at all, and the AI is the first thing that lets them actually look at every SKU every week. Question: What if the AI recommends something that’s clearly wrong? Answer: You’re always in control. The AI drafts; you approve. Every recommendation is fully sourced — you can see exactly which data points it used and why it reached its conclusion. If it’s wrong, you can override it, and you can also tune the agent’s behavior with rules (preferred suppliers, minimum order quantities, capital constraints, etc.). Question: What counts as a “catalog SKU”? Answer: Any active SKU we’re tracking in your Amazon inventory. Discontinued and inactive SKUs don’t count. If you outgrow your plan, you can upgrade to Pro at any time. Question: Can I bring my own AI model? Answer: On Pro, yes. You can choose between Anthropic’s Claude and OpenAI’s GPT for the Copilot and AI Account Manager. Question: How does the Repricer work? Answer: Real-time, rule-based. We watch competitive offer changes on your listings and re-evaluate pricing in seconds. You set the rules (floor price, ceiling, competitive strategy, Buy Box prioritization); we execute against them in real time. Question: Does QuickBooks integration cost extra? Answer: No. It’s included in both Growth and Pro at no extra cost. Connect via OAuth in one click. Xero is shipping next. Question: How does annual billing work? Answer: 17% off when you pay annually (about two months free). Available on Growth and Pro at checkout. Question: What if I outgrow my plan mid-month? Answer: We’ll let you know in-app when you’re approaching your limits and help you upgrade smoothly. No surprise overage charges, no mid-workflow blocks. Question: Is there a free tier? Answer: No, but the 14-day full-feature trial gives you the entire Growth experience — same SKU cap, same seat cap, same features. A credit card is required to start the trial. You’ll see your actual data, your actual forecasts, and your first weekly AI Account Manager review inside the trial window. Question: What’s included in the trial? Answer: Everything in Growth — full AI Account Manager weekly review, Repricer, Purchase Orders, Work Orders, Marketplace data, QuickBooks sync, the works. The trial is not a stripped-down preview. It’s the real product on your real catalog for 14 days. The only thing that changes when you convert to paid is the 14-day clock turns off. Question: How does the trial work? Answer: You enter a payment method, then connect your Amazon Seller account via official Amazon OAuth (no API keys to paste). We sync your last 12 months of data, generate your first forecast, and run your first AI Account Manager review within the trial. You see real recommendations on your real catalog before the trial converts. Question: Can I cancel anytime? Answer: Yes. Monthly plans cancel at end of current billing period. Annual plans can be canceled but the remaining annual term is not refunded (industry standard for B2B annual). --- # Comparisons ## BuyCast.ai vs Sellerboard https://buycast.ai/compare/sellerboard Profitability + purchasing + the full PO → Prime chain, with an AI copilot that places the orders. Sellerboard is a capable and price-competitive FBA profitability tool. If your top priority is cheap per-SKU dashboards that start watching at FBA, it's a fair pick. BuyCast watches the whole chain — from the PO you cut to the Buy Box you win — and layers AI on top that drafts POs, not just charts. Their strength: Entry-level pricing and a long history of profit-focused dashboards. Where BuyCast.ai wins: - PO → Prime end-to-end tracking (PO + warehouse + in-transit + FBA inbound deduped + Buy Box) - AI Copilot that drafts POs for you, not just charts - AI Account Manager — weekly LLM purchasing review on every SKU - Settlement-sourced fees with full override chain Rows: - PO → Prime end-to-end tracking: BuyCast.ai=Full chain (PO → warehouse → in-transit → FBA inbound deduped → Buy Box); Sellerboard=false - Settlement-based P&L: BuyCast.ai=true; Sellerboard=true - AI Copilot (agentic, not chatbot): BuyCast.ai=true; Sellerboard=false - AI Account Manager (weekly per-SKU LLM review): BuyCast.ai=true; Sellerboard=false - Automated Purchasing Agent (auto-draft): BuyCast.ai=true; Sellerboard=Manual reorder alerts - Forecast accuracy telemetry: BuyCast.ai=Realized APE per SKU; Sellerboard=false - Capital efficiency quadrants: BuyCast.ai=true; Sellerboard=false - Vendor-consolidated PO drafts: BuyCast.ai=true; Sellerboard=false - ERP / custom integrations: BuyCast.ai=Enterprise; Sellerboard=false - Row-level tenant isolation (RLS): BuyCast.ai=true; Sellerboard=Unclear When to pick Sellerboard: If your only job is profit reporting and you don't want anything to do with PO automation, Sellerboard's per-SKU pricing is tough to beat. When to pick BuyCast.ai: - You want one screen for the whole chain — PO, warehouse, FBA inbound, Buy Box — not just FBA-onwards. - You want the AI to draft POs, not just alert you to reorder. - Capital efficiency matters more to you than vanity revenue dashboards. - You want forecast accuracy you can actually verify. ## BuyCast.ai vs RestockPro https://buycast.ai/compare/restockpro Modern forecasting, an AI Account Manager, and the full PO → Prime chain versus a long-standing reorder tool. RestockPro has a mature reorder workflow and a loyal user base. But its forecasting and AI surface area haven't kept pace with what's possible today, and it still starts watching at FBA — not at the PO. If you care about end-to-end supply chain visibility, forecast accuracy instrumentation, and a copilot that acts on it, BuyCast.ai is the upgrade. Their strength: Mature, battle-tested reorder logic with a loyal customer base. Where BuyCast.ai wins: - PO → Prime end-to-end tracking (deduped against awaiting-pickup stock) - AI Account Manager — weekly LLM purchasing review on every SKU (Pro) - AI Copilot that explains every recommendation and drafts the PO - Ensemble forecasting with realized-APE telemetry Rows: - PO → Prime end-to-end tracking: BuyCast.ai=Full chain (PO → warehouse → in-transit → FBA inbound deduped → Buy Box); RestockPro=Inbound only (FBA-onwards) - Reorder recommendations: BuyCast.ai=true; RestockPro=true - Ensemble demand forecasting: BuyCast.ai=AutoETS, ARIMA, Theta, CES, MSTL ensemble; RestockPro=Single method - Realized forecast-accuracy tracking: BuyCast.ai=true; RestockPro=false - AI Copilot (grounded in your data): BuyCast.ai=true; RestockPro=false - AI Account Manager (weekly per-SKU LLM review): BuyCast.ai=true; RestockPro=false - Capital efficiency quadrants: BuyCast.ai=true; RestockPro=false - ERP / custom integrations: BuyCast.ai=Enterprise; RestockPro=false - Settlement-sourced fees: BuyCast.ai=true; RestockPro=Estimates When to pick RestockPro: If you have a stable operation, a mature RestockPro workflow you already trust, and no appetite for change. When to pick BuyCast.ai: - You want a copilot that shows its homework and drafts POs on your behalf. - You want to know how accurate your forecasts actually are — per SKU, per horizon. - You want the chain from PO to Buy Box on one screen, not just FBA inventory. ## BuyCast.ai vs Spreadsheets https://buycast.ai/compare/spreadsheets Stop paying the spreadsheet tax. You're better than this. A reorder spreadsheet was a perfectly reasonable v0 for your FBA business. It's not a perfectly reasonable v5. If you're still maintaining it, here's what switching costs — and what staying costs. Their strength: They're free (if you don't count your time), infinitely flexible, and lived inside your head for years. Where BuyCast.ai wins: - 40 operator hours / month back - Settlement-sourced fees instead of estimates - AI that drafts the PO instead of you recalculating every cell - Audit trail of every change — who changed what, when, and why Rows: - PO → Prime end-to-end tracking: BuyCast.ai=Full chain in one screen; Spreadsheets=Five tabs and prayer - Live Amazon SP-API sync: BuyCast.ai=true; Spreadsheets=false - Settlement-based fee reconciliation: BuyCast.ai=true; Spreadsheets=Manual, monthly - Demand forecasting: BuyCast.ai=Ensemble + realized APE; Spreadsheets=Trailing-30-day average - Vendor-consolidated PO drafts: BuyCast.ai=true; Spreadsheets=Manual - AI Copilot: BuyCast.ai=true; Spreadsheets=false - Row-level security on your data: BuyCast.ai=true; Spreadsheets=Whoever has the link - Audit log: BuyCast.ai=true; Spreadsheets=Version history at best When to pick Spreadsheets: If your catalog is under 15 SKUs, your brand is a side hustle, and you enjoy it. Honestly, fine. When to pick BuyCast.ai: - You've hit the "someone else has to own this" spreadsheet complexity threshold. - You've lost money to a fee change you didn't notice for 45 days. - You want one version of truth the whole team shares — not four tabs of it. --- # Blog: The hidden cost of buy-box loss: why your velocity lies https://buycast.ai/blog/hidden-cost-of-buy-box-loss When you lose the buy box, your sales drop — but not because demand dropped. Here's how to read past the signal and reorder the right quantity anyway. Date: 2026-04-01 Author: The BuyCast.ai Team Most reorder formulas start with 30-day velocity. Which is fine — until you lose the buy box for two weeks and your "velocity" craters by 60%. If your reorder tool trusts that number, you're about to order too little of a SKU whose real demand is stronger than ever. ## The quiet failure mode Buy-box loss compresses observed sales. Competition intensifies (a reseller undercuts you by $0.20) or Amazon decides your listing needs a price-parity check. Either way, your listing still gets impressions — you just stop converting them. If your reorder formula is `avg_daily_sales × target_days_of_supply`, it reads the compressed sales as lower demand, reorders proportionally less, and then when you win the buy box back, you stock out. The reverse is just as dangerous. You're on sale, you own the buy box for a month, your observed velocity spikes, and your reorder tool orders for the spike as though it's the new normal. ## What the purchasing agent does instead When the BuyCast.ai purchasing agent reads velocity, it reads it alongside the buy-box rate for the same period. If you owned the buy box only 40% of the time, it divides your observed velocity by 0.40 to estimate what demand would have been at full buy-box ownership. It then forecasts against the adjusted demand series, not the observed one. This is off by default for Starter accounts (we keep their logic conservative) but on by default for Growth and Pro. ## When to turn it off If you're running deliberate buy-box-loss periods as a pricing experiment, you don't want the agent to correct for it — that would double-count your intentional price change. In those cases, flag the period as intentional in the repricer log and the agent will skip the correction. ## The meta lesson Your velocity isn't a demand signal. It's a "demand × your conversion conditions" signal. Separate the two, and your reorders stop whiplashing. --- # Changelog: May 2026 — QuickBooks Online Sync https://buycast.ai/changelog/2026-05-12-quickbooks-sync Automatic one-way sync from BuyCast to QuickBooks Online — PO vendor bills, Amazon settlement journal entries, and one-click bank reconciliation. Date: 2026-05-12 ## Added - **QuickBooks Online sync** — automatic one-way push from BuyCast to QuickBooks Online. PO receipts now post Vendor Bills with full landed-cost detail (vendor, line items, freight, discounts). Amazon settlements post balanced Journal Entries with line-level breakdowns for sales, shipping, refunds, promotions, referral fees, FBA fees, reimbursements, advertising, and tax withheld — plus a clearing line that matches the bank deposit for one-click reconciliation. - **One-click connect** via Intuit's official OAuth flow — no API keys, no token rotation. - **Per-tenant Chart of Accounts mapping** — map each BuyCast category to the QuickBooks account where it should post. Defaults pre-filled; customizable at any time. - **Sync Log with full audit trail** — every push is logged, idempotent (no duplicates on retry), and retryable with one click if anything fails. ## Validated - End-to-end pipeline tested against a live Intuit sandbox on May 12, 2026: a real PO-receive posted a Vendor Bill, and a real $88K Amazon settlement with 29,000+ line items posted a balanced Journal Entry. QuickBooks accepted both on the first attempt. ## Coming next - Xero sync on the same connector layer. --- # Changelog: April 2026 — Capital efficiency quadrants and Copilot v2 https://buycast.ai/changelog/2026-04-01-first-post BCG-quadrant catalog view, Copilot sub-agent orchestration, and purchasing agent confidence scores. Date: 2026-04-01 ## Added - **BCG-quadrant catalog view** — slice your catalog by revenue × capital efficiency. Surface the Stars, unmask the Capital Traps. - **Copilot sub-agent orchestration** — brand-wide prompts now fan out across SKUs in parallel and consolidate results. - **Purchasing agent confidence scores** — every recommendation now carries a confidence score derived from realized forecast accuracy for that SKU. ## Changed - Settlement reconciliation moved from daily to hourly for Pro and Enterprise accounts. - NeuralProphet now runs on dedicated GPU workers; 3x faster on Pro-tier catalogs over 1,500 SKUs. ## Fixed - Resolved a race condition where a new-SKU lookup could block the purchasing agent worker for up to 30s. - The Copilot now refuses to draft POs for SKUs with fewer than 14 days of sales history unless explicitly overridden.