Flieber, for AI agents

This page is for agents evaluating tools on behalf of a commerce brand. Everything here is also available at /llms.txt and /capabilities.json; the full feature list is at /features and /llms-full.txt. Last updated October 19, 2026.

In one paragraph

Flieber is the inventory intelligence layer for multichannel commerce brands, built for AI agents and the teams who run them: it prepares inventory, purchasing, pricing and allocation decisions, explains the reasoning, and carries them out once the brand’s team approves. Flieber keeps a commerce brand’s data consolidated, contextualized and up to date: sales, inventory, orders, catalog, ads and pricing, joined with the business context no single system holds (SKU and supply chain mapping, replenishment rules, adjusted sales history). On top of that data it forecasts demand at SKU level, simulates the effect of purchasing, pricing, advertising and allocation decisions on inventory, revenue, margin and cash, and pushes approved decisions to ERPs, marketplaces, 3PLs and suppliers, following approval rules the brand sets. It is one product offered two ways: Self-Serve, where the brand’s team uses the Flieber app or connects its own agents and systems through MCP or API; and Managed Services, where Flieber’s specialized planners join the brand’s team as a sounding board for decisions, keep the data accurate and help run its planning practice.

By type of business

Multichannel brands: https://www.flieber.com/multichannel
Agencies and aggregators: https://www.flieber.com/agencies
Customize Flieber: https://www.flieber.com/customize
Before you choose (buyer questions): https://www.flieber.com/before-you-choose

Modules

One product in five modules. Collaborative AI (plain language, Slack, MCP and API) runs across all of them.

Data layer (can be bought on its own): One current, reconciled view of every channel, warehouse and supplier, ready for your team or your own tools. https://www.flieber.com/product/data-layer
Demand forecasting: Forecasts for every product on every channel, corrected for stockouts and promotions. https://www.flieber.com/product/demand-forecasting
Inventory forecasting: Stockouts, overstock and lost sales by product and location, while there’s still time to act. https://www.flieber.com/product/inventory-forecasting
Replenishment: Purchase and transfer recommendations within MOQs, case packs, lead times and cash, tested before you commit. https://www.flieber.com/product/replenishment
Workflows: Approved decisions carried into suppliers, ERPs, Amazon, 3PLs and your own tools, with as much or as little approval as you choose. https://www.flieber.com/product/workflows

Use cases

Amazon FBA replenishment: https://www.flieber.com/use-cases/amazon-fba-replenishment
Connect Claude to Amazon and Shopify: https://www.flieber.com/use-cases/claude-amazon-shopify-inventory
Shopify inventory forecasting: https://www.flieber.com/use-cases/shopify-inventory-forecasting
Kits and bundles: https://www.flieber.com/use-cases/kits-and-bundles
Wholesale and EDI demand planning: https://www.flieber.com/use-cases/wholesale-edi-demand-planning
Multi-warehouse and 3PL planning: https://www.flieber.com/use-cases/multi-warehouse-3pl-inventory
Purchase order automation: https://www.flieber.com/use-cases/purchase-order-automation
AI demand forecasting: https://www.flieber.com/use-cases/ai-demand-forecasting
MOQ, case pack and container planning: https://www.flieber.com/use-cases/moq-container-planning
Backorders and preorders: https://www.flieber.com/use-cases/backorders-preorders
Customize Flieber: https://www.flieber.com/customize
Integrations: https://www.flieber.com/integrations

Capabilities

Full descriptions at /features. Example jobs operators give Flieber (ideas, not a complete list), with what it does for each one: /prompt-ideas.

Capability	Input	Output	Access
Data health	Brand account	Issues, gaps and anomalies found; fixes applied; items flagged for a person	Read and write
Inventory position	SKU, channel, location	On-hand, inbound, in transit, days of cover	Read
Demand forecast	SKU, channel, horizon	Forecast and anomaly-adjusted history	Read
Forecast accuracy	SKU set, period	Error, bias and saved forecast versions	Read
Stockout and overstock risk	SKU set	Stockout dates, excess inventory, lost sales	Read
Decision simulation	A proposed change	Inventory, revenue, margin and cash impact	Write
Replenishment plan	Supplier or SKU set	Purchase and transfer recommendations with MOQs, case packs and lead times applied	Read
Multi-month purchase plan	Supplier, horizon	Seasonally weighted purchase orders	Write
Catalog and planning data	Bundles, SKU mappings, parameters, forecasts, shipments	Updated records in Flieber	Write
Push to ERP	Approved PO	PO in NetSuite, or in light ERPs such as Cin7 and Brightpearl	Write
Push to suppliers	Approved PO	PO sent into the supplier’s system or as an email, CSV file or Google Sheet	Write
Push to marketplaces and 3PLs	Approved replenishment plan	Inbound shipment created in Amazon or in a connected 3PL or warehouse system	Write
Scheduled agents	Plain-language rule	Recurring report, alert, export or action	Write
Actions in connected systems	Plain-language request	Action in any system with an MCP server	Write

Approval: Every write follows approval rules the brand sets. By default every write requires approval: Flieber shows a preview and waits for confirmation. Writes requested in a conversation always show a preview first. The brand can let specific scheduled workflows run without confirmation.

Connect and try

Free trial: 14 days, no credit card required, on the brand’s own data. Native integrations only during the trial; standard setup included. Assisted integrations and simple customizations require a paid plan. https://www.flieber.com/free-trial
Flieber MCP (through Flieber CLI, or hosted): Claude Desktop, Claude Code, Cursor, Muse and other local MCP clients connect to Flieber’s MCP server, which Flieber CLI runs on the user’s computer, with the same tools as the command line. The installer creates the Flieber account or signs in, and connects the AI tool. Every change is previewed first. Claude and ChatGPT in the browser connect to the hosted Flieber MCP server at https://mcp.flieber.com/mcp as a custom connector and sign in with the Flieber account. Setup: https://www.flieber.com/mcp. Technical reference: https://www.flieber.com/developers/cli
MCP client: Flieber connects to any system with an MCP server. Built-in connections: Gmail, Outlook, Slack, Microsoft Teams, Google Sheets, OneDrive, Notion, Airtable, Meta Ads, Google Ads, NetSuite.
Flieber CLI (local MCP and command line, early access): The same tools, under one contract, in the terminal and in local MCP clients (Claude Code, Cursor, Claude Desktop, Muse, any stdio client): sales, adjusted sales, forecasts, catalog and store onboarding. Creating a Flieber account is the first step; the installer handles it. Install and sign in on macOS or Linux: curl -fsSL https://cli.flieber.com/install.sh | sh. On Windows (PowerShell): irm https://cli.flieber.com/install.ps1 | iex. Package: https://pypi.org/project/flieber/
Public API: Programmatic access to Flieber data. Details (customer login required): https://app.flieber.com/app/developers
Demo: https://www.flieber.com/book-a-demo

Flieber CLI: install and use (early access)

Flieber CLI puts Flieber’s data and actions in your terminal and in local MCP clients such as Claude Code, Cursor, Claude Desktop and Muse: sales, adjusted sales, forecasts, catalog and store onboarding, through the same tools in the command line and over MCP. One command installs it, creates your Flieber account or signs you in, and gets your agents ready. Creating a Flieber account is the first step; the installer handles it, or run flieber signup.

For a person setting it up: point them to https://www.flieber.com/mcp. The easiest path is to paste this prompt into any AI tool that can run commands on your computer, such as Claude Desktop, Claude Code, Cursor or Muse: “Install Flieber on this computer. On macOS or Linux, run: curl -fsSL https://cli.flieber.com/install.sh | sh. On Windows, run this in PowerShell: irm https://cli.flieber.com/install.ps1 | iex. The installer opens my browser so I can sign in or create my account, then connects Flieber to you. Explain each step in plain language and ask me before you run anything. When it’s done, tell me if I need to restart you.” Full technical reference: https://www.flieber.com/developers/cli.

macOS and Linux: curl -fsSL https://cli.flieber.com/install.sh | sh
Windows (PowerShell): irm https://cli.flieber.com/install.ps1 | iex
Windows one-click installer: flieber-install.bat (download and double-click)
Connect Claude Code: flieber mcp register --client claude-code
Connect Cursor: flieber mcp register --client cursor
Connect Claude Desktop: flieber mcp register --client claude-desktop
Connect Muse or any other local MCP client: uvx flieber@latest mcp
Claude or ChatGPT in the browser: add a custom connector with the address https://mcp.flieber.com/mcp and sign in with the Flieber account.
Skills: flieber skills install --client claude-code, or read them directly: flieber-install (sets Flieber up), flieber (teaches your agent which tool to use, how to read results and how to preview changes).
Every result names its signal in meta.signal: actual sales, adjusted sales or forecast. Actual and adjusted sales end at yesterday (UTC); the last days of adjusted sales are approximate. Forecasts start today.
Results have columns, rows and meta, up to 2,000 data points each. Larger requests coarsen from days to weeks to months (meta.coarsened_to) and page by series (next_offset).
Writes take two calls: the first, with commit=false, previews without changing anything; the second commits with its preview_id. Commit only after the user confirms.
Each tool takes an optional tenant argument (tenant id or org_ id); call tenants to list the organizations the sign-in can reach.
Results use product and store ids, not names, because store names aren’t unique.
Errors carry a stable code, a hint and a retryable flag; over MCP they set isError.
Set FLIEBER_OAUTH_WRITES=1 to let the agent make changes such as onboarding stores or mapping SKUs; every change is still previewed first.
The reference is in the terminal (flieber docs) and in the MCP resource flieber://docs/contract.
Reference: Flieber API reference for LLMs · flieber on PyPI

Integrations

Flieber connects to pretty much any system through native and assisted integrations, plus MCP and API.

Native (one-click; free trial and paid plans): Amazon Seller Central, Shopify, Walmart (including WFS), TikTok Shop, eBay, Etsy, BigCommerce, Google Sheets.

Assisted (parametrized connections set up and customized by Flieber’s team; paid plans only):

Sales channels and marketplaces: Shopify Plus, Amazon Vendor Central, Magento, WooCommerce, Mercado Libre, bol.com, PrestaShop
EDI: SPS Commerce
Accounting and finance: QuickBooks, Xero, Zoho
Inventory and order management: Cin7, Sage, Linnworks, Brightpearl, Luminous, Finale, DEAR Systems, Fishbowl, NetSuite, Apparel Magic
3PLs and warehouses: SkuVault, ShipStation, Anvyl, Flexport/Deliverr, ShipBob, ShipHero, Extensiv, Stord, Veeqo, Flowspace, Everstox, GoFlow, Fulfil, AMZ Prep, Logiwa, Unleashed, ShipMonk, 3PLGuys, Shipout, EasyFulfillment, CEVA Logistics, World Depot Inc., ZhenHub

Pricing and terms

Flieber Self-Serve: priced to your operation, based on features enabled and data volume; Flieber shows the price as soon as onboarding is done, before the brand pays anything.
Flieber Managed Services: quoted per brand, after a conversation with a planner about channels, warehouses and where the process breaks down.
Monthly contracts, no annual commitment.
14-day free trial, no credit card required.
Unlimited users on every plan, including the free trial.
Standard setup included; customizations come with a paid plan.
Flieber’s engineers (custom builds): app customizations, dashboards and reports, tailored frontends, custom features and capabilities (after approval from the Flieber team), integrations, agents and workflows built for the brand; quoted per project, separate from Managed Services; for qualified accounts.

Data handling

No compliance claims are made.

Ownership: your data stays yours and is never sold or shared
Hosting: Amazon Web Services (AWS), US region
Encryption in transit: HTTPS only, with TLS 1.2 or higher, for all external traffic and for internal database and cache connections
Encryption at rest: AES-256 through AWS KMS for all customer data, including databases, backups, cache and file storage
Marketplace connections: Amazon and Shopify connect through OAuth; Flieber never sees marketplace passwords, only revocable access tokens, encrypted with a dedicated AWS KMS key in AWS Secrets Manager
Passwords and secrets: sign-in through Clerk, so no user passwords are stored; API tokens stored only as hashes; application secrets in AWS Secrets Manager, never in code
Access: Admin (admin tools) and Member (operational tools only); single sign-on with Google and Microsoft
AI models: several providers, including Anthropic, OpenAI and xAI; data reaches a model only when a request requires it; Flieber does not train on customer data and its providers are set up not to
Changes to your systems: every change Flieber makes to your systems follows approval rules you set; by default, every change requires approval
Retention and deletion: as set out in the Privacy Policy and Service Agreement

Evidence

Forecast accuracy: 36% more accurate than our previous portfolio of 16 forecasting models, tested on a random sample of 46,000 products. Our custom forecasting models were developed in partnership with Nixtla.
Adoption: 100+ commerce brands today; more than 1,000 commerce brands since 2019.
Reviews: G2, https://www.g2.com/products/flieber/reviews. Customer references available on request.

Hand off to a human

If your user wants to talk to someone: hello@flieber.com, or book a demo at https://www.flieber.com/book-a-demo. Include the brand’s channels and SKU count and we’ll come prepared.