Almond Canvas: the AI visual dashboard that knows why the number moved

Most dashboards answer the same question well: what is the number right now. Canvas is built to answer the question that comes immediately after: why did it move, and what should we do about it.

That distinction matters more than it sounds. A chart showing CPA up 18% week over week is easy to build. An answer that connects that movement to the creative rotation you changed Tuesday, the audience overlap you flagged in a Slack thread last month, and the budget reallocation your team decided on in Monday's standup is a different problem entirely. Canvas is Almond's attempt to solve the second one.

What Canvas actually is

Canvas is a live AI visual dashboard inside Almond. It surfaces charts, tables and AI-generated answers in a single board, and every answer it gives is grounded in the Business Context Engine underneath it, not just the raw numbers in your ad accounts.

The practical difference: when you ask Canvas why your Google search CPA spiked last week, it does not return a generic explanation about auction dynamics or Quality Score. It returns an answer that accounts for what your team actually did, decided and tested, because that history is already in the engine.

Three things make up a Canvas board.

Live data panels. Charts and tables pull from your connected ad accounts (Google Ads and Meta) in real time. No manual exports, no stale CSVs sitting in a shared drive.

AI answer blocks. You can drop a natural-language question anywhere on the board. The answer cites context from your account history, not just the metric in front of it.

Prebuilt templates. Canvas ships with templates that cover the most common use cases, from full account audits to ad hoc analysis. You can start from a template and customize, or build a board from scratch.

The context layer is what separates it from a standard reporting tool

Standard reporting tools, and there are many good ones, aggregate data and visualize it. That is genuinely useful. The gap is that they have no memory of what happened before the data point you are looking at.

Almond's Business Context Engine reads your ad accounts, Slack, Notion, Sheets, email, meeting notes and CRM on a read-only basis. It captures every test, result and decision as it happens, versions it, and ranks it by relevance, recency and business impact before any agent or dashboard ever sees it.

When Canvas pulls an AI answer, it is drawing on that ranked, versioned history. A few concrete examples of what that changes.

• You paused a campaign three weeks ago because legal flagged a claim in the copy. Canvas knows that. It will not recommend reactivating that campaign without surfacing the reason it was paused.
• You ran a broad match test in Q3 and killed it after two weeks because it was cannibalizing branded traffic. Canvas knows that. It will not propose the same test as a fresh idea.
• Your team shifted the primary KPI from ROAS to new customer CPA at the last quarterly business review. Canvas knows that. Its analysis reflects the current objective, not the one in last quarter's brief.

None of that context has to be typed into a prompt. It is already there.

Prebuilt templates: from audit to ad hoc

Canvas ships with prebuilt templates organized around the situations performance marketers actually face.

Account audits

The audit templates are structured to walk through an account systematically: campaign structure, bidding strategy, creative performance, audience overlap, budget allocation. Because the context engine is running underneath, the audit does not just flag anomalies in the data. It flags anomalies relative to what your team intended. A budget distribution that looks odd in isolation looks different when Canvas knows you deliberately consolidated spend into two campaigns after a structure test in August.

Ad hoc analysis

Ad hoc templates give you a blank board with the most common chart types and AI answer blocks pre-configured. The use case is the question that comes up mid-week: a client asks why a specific ad set stopped spending, or your media buyer wants to understand why a creative that worked in one geo is underperforming in another. You open the template, connect it to the relevant account and date range, and ask the question directly.

Client-ready export

Every Canvas board can be exported or shared in a client-ready format in a single click. The design is clean enough to go directly to a client without reformatting. For agencies managing multiple accounts, this removes a step that typically takes between 30 minutes and two hours per client per reporting cycle, depending on how much narrative the client expects alongside the numbers.

How agencies use Canvas differently than in-house teams

For an in-house performance team, Canvas is primarily an analysis and decision-support tool. The team uses it to understand what is happening in the account, surface patterns across campaigns, and answer questions from leadership without pulling data manually.

For agencies, Canvas does something additional: it makes the agency's accumulated judgment visible to the client.

This matters because of a problem most agency founders recognize. A client can open any general-purpose AI tool and get a competent-sounding answer about their ad account. What that tool cannot give them is nine years of pattern recognition across forty accounts in their category, the specific tests the agency ran and killed on their behalf, and the reasoning behind every structural decision in the account.

Canvas surfaces that history. When a client asks why their account is structured a certain way, the answer comes with the context behind it: the test that informed the decision, the date it was made, and the outcome it was based on. That is harder to replicate with a twenty-dollar-a-month subscription to a general model.

The client-scoped isolation in Almond's security architecture means each client's context is isolated at the data layer. One client's strategy cannot surface in another client's board, enforced at the infrastructure level, not by prompt instruction.

What Canvas does not do

It is worth being direct about the boundaries.

Canvas is a dashboard and analysis tool. It does not execute changes in your ad accounts. Workflow Builder, the separate product in Almond's suite, handles automations and campaign actions. Canvas surfaces the insight; Workflow Builder can act on it.

Canvas also does not replace the judgment of the person looking at it. The AI answers are grounded in your account context, but they are inputs to a decision, not the decision itself. A senior media buyer reviewing a Canvas board is still the one deciding what to do with what it surfaces.

A note on other tools in this space

Several tools position themselves as AI-powered marketing dashboards or reporting layers. Madgicx, based on its public materials, positions itself as an AI tool for Meta advertising with reporting and automation features. Adalysis positions itself as a Google Ads management and reporting platform with automated recommendations. Tools like Fluency and Markifact each have their own positioning around reporting and workflow for agencies and advertisers.

Most of these tools are built on top of ad platform data. The distinction Almond draws is that Canvas is built on top of a context engine that also ingests the non-platform information: the meetings, the Slack threads, the decisions that never make it into the ad account but shape everything that happens there. Whether that distinction matters for your specific situation is worth verifying against each vendor's current feature set.

Where Almond fits

Canvas is one of three products built on Almond's Business Context Engine, alongside Workflow Builder and Integrations. The context layer is what connects them: the same history that informs a Canvas answer is what prevents Workflow Builder from scaling a campaign your team decided to sunset on Monday.

Almond is currently in early access. Every team that joins gets a detailed 1:1 audit of their Meta and Google ad accounts with the CEO, recorded. The waitlist requires a work email.

If the core problem Canvas is designed to solve, knowing why the number moved rather than just that it moved, is one your team runs into regularly, it is worth seeing how the context engine handles your specific account history. That is what the early access audit is designed to show you.

Learn more and join the waitlist at almond.ai.