Fihub Overview

Fihub is the agentic intelligence platform for consumer brands. We connect to the tools you already use — your store, marketplaces, fulfillment provider, accounting system, and ad platforms — then automatically unify and model that data so you can answer any business question in real time: in Slack, in your LLM, or in the Fihub app.

This page explains how integrations work — how to connect them, what data we access, and how to manage them. The flow is the same for every integration.

Fihub connects to your tools using OAuth, the industry-standard, secure authorization method used across the web. You never share a password with Fihub — you authorize access directly with each provider, and you can revoke it at any time.

Connecting any integration follows the same steps:

1. 2.1

   Choose your integration: In the Fihub app, go to Settings → Integrations and select the tool you want to connect (e.g. Shopify, Amazon, ShipBob, QuickBooks, Meta Ads).

2. 2.2

   Authorize on the provider: You'll be redirected to the provider's own login and consent screen. Sign in and approve the access Fihub requests.

3. 2.3

   Return to Fihub: After you approve, you're redirected back to Fihub and the connection is confirmed.

4. 2.4

   Initial sync: Fihub imports your historical data and unifies it with your other connected sources. The first sync can take from a few minutes to a few hours depending on data volume.

5. 2.5

   Ongoing sync: After the initial import, Fihub keeps your data current automatically.

A live connection shows as Connected on the Integrations page.

Fihub requests read access to the operational data that powers your analytics — for example orders, products, inventory, fulfillment and shipping costs, expenses, and ad spend. Fihub reads your data; it does not modify, create, or delete anything in your connected systems. We only request the specific permissions each integration needs, and every provider's consent screen shows exactly what you're approving before you confirm.

This holds for every Fihub surface, including our LLM connector — see "Connecting your LLM" below.

Fihub connects across the systems that run a consumer brand:

Some integrations may require specific plan or access levels on the provider's side. Multiple accounts can be connected for advertising platforms and marketplaces.

Don't see your stack? Email support@fihub.ai — we add integrations based on customer demand.

Connecting a fulfillment provider such as ShipBob lets Fihub calculate your true per-SKU fulfillment cost and contribution margin, track inventory levels and turnover across fulfillment centers, and alert you before you run low on stock — all unified with your sales, ad spend, and accounting data.

Once connected, Fihub proactively delivers insights and answers your questions in:

1. 7.1

   Fihub's connector lets you query your business data from an AI assistant, in plain language. Instead of a dashboard, you ask: "What was net sales last week, split by subscription and one-time?" or "Why did CAC go up in August?" — and your assistant answers from your actual, reconciled data. Fihub uses the Model Context Protocol (MCP), an open standard for connecting AI assistants to external data.

2. 7.2

   How to connect: In the Fihub app, open Integrations and use the Connect button on the LLM connector. That flow carries the current step-by-step instructions for each supported assistant — Claude, Gemini and others that support remote MCP servers. You'll need an active Fihub account with at least one integration connected.

3. 7.3

   Access and revoking it: Access is authorised with OAuth: you sign in to Fihub from within your assistant, and no Fihub password is shared with the assistant. A session reaches only your own Fihub account, within your existing permissions. You can revoke it at any time from your assistant's connector settings or from Integrations, and revocation takes effect immediately for new sessions.

4. 7.4

   Asking your first question: Start broad and let the assistant narrow down:

   1. (a.)

      "Using Fihub, what was net sales last week? Break that down by one-time and subscription."

   2. (b.)

      "What drove the change in net sales last quarter?"

   3. (c.)

      "How is CLTV-to-CAC trending over the last 12 months?"

   4. (d.)

      "Show me retention by acquisition cohort for subscription customers."

   5. (e.)

      "Do our books match what Shopify is reporting for last month?"

   6. (f.)

      "Which SKUs are about to run out of stock?"

   7. (g.)

      "Using Fihub, how was the business doing last month?"

5. 7.5

   Getting better answers: Three things worth knowing:

   1. (a.)

      Name the source in your first message: Something like "Using Fihub, give me a business summary" helps the assistant reach for your data rather than answering from general knowledge. After the first tool call it usually continues on its own.

   2. (b.)

      Ask about the source when it matters: Many Fihub metrics exist in two versions: what the platform reported (Shopify, Amazon, Meta) and what your books say (QuickBooks, Xero). These legitimately differ — bookkeeping lag, fulfillment-based recognition, refund timing. Your assistant will tell you which it used, and you can ask it to switch.

   3. (c.)

      Your settings shape the answers: Fiscal calendar, attribution model, churn basis and reporting currency all come from your Fihub configuration, and the assistant is told to cite them when they shaped a number.

6. 7.6

   What the connector can do:

   1. (a.)

      Retrieve any of 230+ metrics and the three financial statements, over any date range — monthly, weekly, or as a single figure for the period

   2. (b.)

      Segment by product, channel, campaign, sales platform, SKU, customer type, subscription vs one-time and more

   3. (c.)

      Explain why a metric moved, and which segments drove the change

   4. (d.)

      Pull retention, cohort and lifetime-value analyses, and subscription conversion

   5. (e.)

      Reconcile your books against platform-reported figures, and audit what was ingested

   6. (f.)

      Show your inventory position by SKU and fulfillment network, and report integration sync status

   7. (g.)

      Answer ad-hoc questions against your modelled data when no standard metric fits

7. 7.7

   What it can write — inside Fihub only:

   1. (a.)

      Saving a cohort definition for reuse

   2. (b.)

      Correcting how a product category, sales channel, marketing channel or accounting line is mapped in Fihub's taxonomy

   3. (c.)

      Your assistant will ask you to confirm before any of these

8. 7.8

   What it cannot do:

   1. (a.)

      Modify anything in Shopify, Amazon, QuickBooks, Xero, your ad platforms or your 3PL

   2. (b.)

      Move money, place orders or change account settings anywhere

   3. (c.)

      Reach another customer's data

   4. (d.)

      Access your end customers' personal data, which Fihub does not hold

9. 7.9

   Privacy: When the connector returns data, that data goes into the AI assistant you've chosen and is then handled under that provider's privacy policy and retention terms, not Fihub's. Review your assistant provider's own privacy policy and its retention and model-training terms before connecting. Our Privacy Policy covers the connector in Section 13.

Need help connecting or using an integration? Email us at support@fihub.ai and we'll get you sorted.