Connect Data Octopus Brain to the AI you use every day.
Octopus Brain connects to the AI assistant you already use: Claude, Codex, ChatGPT, Cursor, ClickUp or another one. The assistant talks and works with Data Octopus Brain based on your store's data, observations, recommended actions, business context and connected data sources, without switching to our app.
E-commerce store
- Google Ads campaigns
- Products
- Search terms
- Categories
- Brands
MCP Data Octopus Brain
- Context
- Observations
- Actions
- Memory
- Domain knowledge
- Data sources
Your AI assistant
- Claude
- Codex
- ChatGPT
- Cursor
- ClickUp
Ask in your own words. Brain answers with your store's data.
A simulated conversation in an AI assistant with MCP Data Octopus Brain connected. Pick the next question you're interested in.
Which AI assistant do you use today?
Ask about your store…
From talking about data to working on your business.
You don't have to prepare data for every conversation.
The assistant can pull available observations, analysis details and recommended actions from Brain. No more exports and pasting tables into the chat.
AI gains context that isn't in the report itself.
Business goals, seasonality, promotion rules or constraints saved in Brain help assess whether a recommendation fits your store's situation.
You can go from a summary to specifics.
Start with “what needs my attention?”, then check the campaigns, products or search terms available in the analysis.
You don't have to accept recommendations without checking.
Ask the assistant to compare proposed actions with the data and business context. Separate what's justified from what needs further verification.
A conversation can end with more than an answer.
Within your permissions you can save agreed business context or your decision on a recommended action.
You work in your assistant, not in yet another interface.
MCP makes Brain's capabilities available to compatible AI apps. You choose where you want to have the conversation.
Data → interpretation → decision → action plan. In one conversation.
What the assistant can read and what it can save.
What the assistant can read
Stores and companies
All stores and companies you have access to. With multiple companies, the assistant always says which store the answer refers to.
Observations
The store's list of observations, the details of a single observation and its rows, for example the products or search terms it concerns.
Recommended actions
What to do and why, with priority and category.
Store context
Questionnaire answers and notes about the store, also broken down by category.
Data sources
What's connected (Google Ads, Meta, Shopify, etc.) and in what state.
What the assistant can save
Always only after your approval in the conversation.
- Complete an answer in the store context, for example add missing information about the store.
- Add a note to the store context or to the store itself.
- Vote on a recommended action (accept / reject). A vote can't be undone, so the assistant always asks.
Strategy and operations. Connected by the same context.
Some need to know where to direct attention and budget. Others: what exactly to check in campaigns and where to start. MCP Data Octopus Brain supports both perspectives.
Choose your perspective
C-level, owner, e-commerce director
Budget, efficiency and priorities
Which campaign issues need my attention today?
Reviewed observations
Three areas stand out in the available observations: spend on products without conversions, slowing momentum of former best-sellers and product overlap between campaigns. I'd start by checking the scale of the first issue. A lack of conversions alone doesn't justify switching products off, though, especially for new arrivals and seasonal demand.
- Products without conversions$4,380
- Declining best-sellers$2,210
- Assortment overlap$1,460
Strategy and business context
Summarise our business model and check what Brain doesn't know about us yet.
Checked store context
I'll summarise the saved information about customers, priorities, prices and how campaigns are run. Separately, I'll point out unanswered questions and information that needs clarifying. A high level of context completion alone doesn't mean all assumptions are still current.
Oversight of stores and decisions
Compare the key risks across the stores I have access to.
Compared stores
For each store I'll list the most important available observations, the scale of the issue and the limitations of the analysis. I'll separate risks requiring action from situations that first need more data. I'll account for differences in analysis periods and won't sum results in different currencies without conversion.
| Store | Risk | Scale |
|---|---|---|
| outdoor-store.com | high | $3,100 |
| outdoor-shop.de | medium | €2,100 |
| outdoor.cz | no data | — |
More questions by category
Budget, efficiency and priorities
Where should we look for budget to reallocate before increasing spend?
Flagged areas to review
The first area to verify is products and search terms that generated cost without recorded conversions in the analysed period. I'll prepare a list of areas to review along with their scale. This isn't yet an amount of possible savings: conversion delay, campaign goals and the role of these products in the business need to be taken into account.
| Area | Cost | Conv. |
|---|---|---|
| Products w/o conversions | $4,380 | 0 |
| Search terms w/o conversions | $1,870 | 0 |
| Overlapping campaigns | $1,460 | 12 |
Should we increase the budget for best-sellers?
Compared best-seller momentum
The available analyses can show whether best-sellers share budget with non-converting products and how their momentum is changing. That helps pick an area to test. Without confirming margin, availability and the profitability target, I don't yet recommend a specific budget increase.
- Nord jacket+34%
- Alp boots+12%
- Urban backpack-8%
Strategy and business context
Which recommendations might conflict with our seasonality?
Checked the seasonal calendar
I'll compare the recommended actions with your saved seasonal calendar. If a product is building demand ahead of its sales peak, limiting its exposure only because it had no conversions recently may be premature. I'll flag such actions as needing further review.
2 recommendations to cut exposure before the peak → needs review
We're shifting priority from revenue to profitability. How do we tell Brain?
Drafted a context entry
I suggest this entry: «Profitability of sales is the priority. Revenue growth doesn't justify higher costs without assessing margin and acquisition cost». I'll point to the relevant question or context category and show the full text for approval before saving. Updating the context alone doesn't change campaign settings.
- 1Priority: sales profitabilitynew
- 2Assess costs against marginnew
- 3Campaign settings unchangedinfo
Oversight of stores and decisions
Which recommendations did I accept and which did I postpone?
Listed your decisions
I'll show your decisions assigned to recommendations: accepted, postponed and rejected, with the available rationale. Acceptance means a decision saved in Brain, not confirmation that the change was implemented in Google Ads.
- Accepted14
- Postponed6
- Rejected3
Prepare a short summary for the board meeting.
Prepared the brief
I'll prepare a three-part brief: the key signals from the analyses, decisions to be made and missing information. For each conclusion I'll give the period and the observation it's based on. I'll separate data from interpretation and won't present potential savings as results already achieved.
The answers below are demonstration examples. The scope of a real answer depends on the available observations, data, context and user permissions. The assistant can combine and interpret them, but should point out missing information.
The same permissions as in the app. Full control over writes.
- Sign in with your Octopus Brain account. No separate passwords or API keys to copy.
- You only see your own companies and stores, with the same permissions as in the app.
- The consent screen shows whether the assistant gets read-only or also write access.
- Writes only on the user's explicit request. Every call is logged, without data content or tokens.
- Third-party content, such as product names or search terms, is marked as data, not instructions. This protects against prompt injection attacks.
- Access can be cut off for one user or for everyone at once. There are also request limits.
Three steps. Then ask in your own words.
MCP server address
https://mcp.octopusbrain.ai/mcp- 01
Add a connection to Brain
In your AI assistant, open the integrations or MCP servers settings. Add a connection named “Data Octopus Brain” and enter the server address.
- 02
Sign in and approve access
Go through the Data Octopus Brain sign-in. Check the scope of access you're granting. Available data and operations depend on your account permissions and assistant settings.
- 03
Pick a store and ask your first question
Turn on the integration in the conversation and write your first question.
Example first question
For the store [store name], show the most important available observations. Suggest the order in which to verify them and point out what context is missing to make a decision.
You don't need to know tool names or write technical queries. You describe the goal and the assistant picks the available MCP operations.
Choose your assistant
- 1In Claude, open Customize → Connectors, then choose + Add → Add custom connector.
- 2Name the connection “Octopus Brain”, paste the MCP server address and select Continue twice, keeping the detected settings. Under Authentication choose Sign in now, keep the suggested OAuth client and select Add.
- 3Sign in to Octopus Brain and approve access. If the connector still shows Connect, select it. Enable the connector in a conversation via + → Connectors.
Go through the Data Octopus Brain sign-in. Check the scope of access you're granting. Available data and operations depend on your account permissions and assistant settings.
Can’t find the option to add a server?
- On Team and Enterprise plans, an Owner or Primary Owner first adds the connector under Organization settings → Connectors → Add → Custom → Web. On Enterprise, people with a role including Manage access to Libraries can also do this. Members then find it under Customize → Connectors and select Connect.
- The Free plan allows one custom connector.
- This is a remote connector: it works in the browser, Claude Desktop and mobile apps. Do not edit claude_desktop_config.json.
Frequently asked questions about MCP Data Octopus Brain
What is MCP Data Octopus Brain?
It's a connection that lets a compatible AI assistant use selected data and features of Data Octopus Brain. MCP provides tools for retrieving observations, analysis details, recommendations, business context and data source information, as well as selected write operations.
How is this different from pasting a report into a chat?
A report is a copy prepared at a specific moment. Through MCP the assistant can pull available information from Brain during the conversation, check observation details and read the saved context. You don't have to manually prepare a new dataset for every question.
Do I need to know how to code?
Not to ask questions. You use natural language. The one-time setup depends on the assistant: it may mean adding an integration in settings or editing a configuration file.
Can I use Brain without opening its app?
Within the features available through MCP: yes. You can read observations, analyse details and perform permitted operations without going to the Brain interface. That doesn't mean MCP replaces all of the app's features.
What information can the assistant retrieve?
Among other things: the list of available companies and stores, published observations, analysis results and their details, recommended actions, your decisions on recommendations, business context and data source statuses. The scope depends on the account's data and permissions.
Can I ask about any report and any date range?
No. The current MCP provides the results and details of analyses prepared in Brain. It isn't universal access to all raw Google Ads data or a tool for creating any report. The assistant should state the analysis period and say when data is missing for an answer.
Is the data available in real time?
You shouldn't assume so. Answers are based on the data and analysis results available in Brain. The analysis date, time range and source delays matter. Advertising data may also be corrected retroactively, e.g. after conversions are added.
Can the assistant change a budget or switch off a campaign on its own?
Not through the currently available set of MCP Data Octopus Brain tools. It can help assess the situation and prepare a plan, but this MCP doesn't provide operations to change budgets or bids, or to enable or pause campaigns in Google Ads.
What can be saved through MCP?
With the right permissions you can update answers and notes in the store's business context and save decisions on recommendations: accept, postpone or reject. Before saving context, the assistant should show the exact text for approval.
Does an accepted recommendation mean the change has been implemented?
No. Acceptance is a decision saved in Brain. It doesn't confirm that the change was made in the advertising system or that it delivered the expected result.
Can I change a previously saved decision on a recommendation?
The current MCP tool doesn't allow undoing or changing a saved vote. That's why you should check the specific recommendation and decision before accepting, postponing or rejecting.
Does Brain know my business's seasonality, margin and goals?
It can use the information saved in the business context. If important data is missing or outdated, the assistant should point that out. Campaign results shouldn't be equated with business profitability without the right data.
Can I use several stores?
Yes, if your account has access to them. In your question you can name a specific store or ask for a comparison of available observations across stores. The scope and comparability of answers depend on each store's data.
Does my data go to the AI assistant's provider?
Yes. Information retrieved by the connected assistant is also processed in its provider's environment. Rules on storage, use and processing location depend on the provider, plan, settings and your agreement with them. Use accounts and configurations approved by your organisation.
Does signing out of Brain disconnect the assistant?
No. Signing out of the Brain app is not the same as disconnecting the MCP integration. The connection is managed separately in the assistant's integration settings.
Is MCP available on my plan and how much does it cost?
Confirm the terms of access to MCP Data Octopus Brain with the Data Octopus team. Your AI assistant's provider may have separate requirements and fees.
Start with a question that matters to your business.
What needs my attention?
Which recommendations does the data support?
What should we check before the next campaign change?
Connect your assistant to Data Octopus Brain and move from observations to informed action.