> For the complete documentation index, see [llms.txt](https://planck-ai.gitbook.io/planck-ai-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://planck-ai.gitbook.io/planck-ai-docs/documentation/core-concepts/workspace-context-and-automations.md). # Workspace Context & Automations Planck AI uses workspace context to understand business data, connected systems, and user questions more accurately. This context helps Planck AI interpret the meaning of fields, entities, relationships, business terms, and output formats inside a workspace. It allows users to ask natural-language questions and build automation workflows without needing to understand every underlying table, file, or system schema. ### What workspace context does Workspace context gives Planck AI structured business understanding about the data inside a workspace. It can include high-level information such as: * Important business entities * Field definitions * Data types * Synonyms and business terminology * Relationships between fields * Query guidance * Chart and reporting preferences * Filtering and grouping behavior * Calculation and validation guidance * Workspace-specific rules For example, in a commerce workspace, Planck AI can understand concepts such as products, SKUs, categories, price, inventory, stock status, product visibility, and product attributes. In a finance workspace, Planck AI can understand concepts such as customers, invoices, revenue, accounts receivable, accounts payable, suppliers, cost centers, and payment status. ### Why this matters Enterprise data is rarely self-explanatory. The same field can mean different things in different systems. Business users may also use different words from the names used in databases, spreadsheets, or SaaS platforms. Workspace context helps bridge this gap. For example, a user may ask: ``` Which products are out of stock? ``` Planck AI can use workspace context to understand which fields represent products, inventory quantity, stock thresholds, and stock status. Another user may ask: ``` Show discounted products by category. ``` Planck AI can use workspace context to understand product categories, prices, special prices, and the right grouping logic. ### Workspace and organisation knowledge Planck AI can use knowledge at different levels. #### Workspace knowledge Workspace knowledge is context that applies to a specific workspace. It may include business rules, terminology, exceptions, policies, customer-specific notes, operating assumptions, or workflow instructions. Examples: ``` Apex Holdings and Apex should be treated as the same customer. ``` ``` Invoices above 10,000 euros require finance manager review. ``` ``` For this workspace, urgent means same-day response. ``` Workspace knowledge helps Planck AI give answers that match the way a specific team, client, or business function works. #### Organisation knowledge Some knowledge applies across the whole organisation. Examples include: * Company policies * Brand guidelines * Compliance rules * Standard operating procedures * Shared terminology * Approved reporting definitions * Common business rules Organisation-level knowledge can be made available to selected workspaces so teams do not need to recreate the same context again and again. For example, a finance policy maintained at organisation level can be used across multiple finance-related workspaces. ### Adding knowledge from chat Users can add or correct knowledge while working in chat. When a user notices that Planck AI is missing context, they can add a short explanation in plain language. This helps the workspace improve over time. Examples: ``` This customer is known as Apex in Salesforce but Apex Holdings in NetSuite. ``` ``` For German customers, use 30-day payment terms unless the contract says otherwise. ``` ``` When comparing supplier spend, exclude one-time setup fees. ``` This allows business users to improve Planck AI without changing database schemas, writing code, or rebuilding integrations. ### Deterministic compute layer Planck AI can use an integrated Python-based compute layer to perform calculations, transformations, validations, and structured analysis outside the language model. This helps make workflows more deterministic. Instead of relying only on the AI model to reason over numbers, formulas, filters, or business rules, Planck AI can execute computations in a controlled environment and return structured results back into the conversation or automation workflow. This is useful for tasks such as: * Calculating totals, averages, ratios, and variances * Running formulas over spreadsheet or database results * Validating extracted values * Comparing planned versus actual values * Creating tables, charts, and summaries from computed outputs * Applying business rules before generating an answer * Producing repeatable outputs for recurring automations For example, a user may ask: ``` Calculate supplier spend by category for the current month and flag categories that increased by more than 20%. ``` Planck AI can use workspace context to understand the relevant entities and fields, retrieve the right data, run the calculation through the compute layer, and then present the result in a user-friendly format. ### Context-aware automation Workspace context can also support automation workflows. Instead of creating automations only from raw prompts, users can build automations that understand the workspace’s business context. Examples: ``` Every Monday, summarize products that are out of stock or close to the reorder threshold. ``` ``` Alert the operations team when a high-value invoice is overdue. ``` ``` Create a weekly summary of delayed projects by region. ``` ``` Send a monthly report of supplier spend grouped by category. ``` ``` Flag products that are online but missing key product information. ``` These automations work better when Planck AI understands the meaning of the underlying data, not just the raw column names. ### Connected systems Planck AI can work with documents, databases, cloud storage, and enterprise systems. Common integration points include: * Microsoft 365 * OneDrive * SharePoint * Salesforce * Odoo * PostgreSQL * MySQL * Adobe Commerce / Magento * Collate * CSV and Excel files * PDFs and text documents Available integrations may vary by deployment and customer setup. ### How users should think about automations A Planck AI automation usually combines five things: 1. **A workspace**\ The business area, team, client, or project where the automation runs. 2. **Connected data**\ Documents, databases, SaaS systems, cloud folders, or uploaded files. 3. **Workspace context**\ The business meaning of the data, including fields, entities, rules, and terminology. 4. **Knowledge**\ Additional business context that may come from users, documents, or organisation-level knowledge. 5. **Deterministic compute**\ Structured calculations, validations, and transformations that can run outside the language model. Together, these allow Planck AI to generate more useful answers, perform reliable calculations, and support repeatable workflows. ### Example use cases #### Commerce and product catalog automation For commerce systems such as Adobe Commerce / Magento, Planck AI can help users monitor catalog quality, product availability, pricing, stock levels, and category coverage. Example questions and automations: ``` Which products are enabled but out of stock? ``` ``` Show products with missing descriptions or metadata. ``` ``` Alert me when stock falls below the configured threshold. ``` ``` Summarize products with special pricing ending this week. ``` #### Finance automation For finance teams, Planck AI can help monitor revenue, invoices, supplier spend, accounts receivable, and payment status. Example questions and automations: ``` Show overdue invoices by customer. ``` ``` Summarize supplier spend for this month. ``` ``` Alert me when a large invoice becomes overdue. ``` ``` Create a weekly cash collection summary. ``` #### Operations automation For operations teams, Planck AI can help track projects, tasks, documents, rollout status, and operational exceptions. Example questions and automations: ``` Which rollout milestones are delayed? ``` ``` Summarize blocked tasks by contractor. ``` ``` Alert me when a project document mentions a delay or missing approval. ``` ``` Create a weekly operations report from connected data sources. ``` ### Best practices Create automations around clear business outcomes. Good automation examples: ``` Send a weekly summary of overdue invoices. ``` ``` Alert me when product inventory falls below threshold. ``` ``` Summarize new SharePoint documents added to this workspace each day. ``` ``` Create a monthly supplier spend report by category. ``` Avoid vague automations such as: ``` Analyze everything. ``` ``` Tell me what is wrong. ``` ``` Watch all data. ``` Focused automations produce better results and are easier to review. --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question. Perform an HTTP GET request on the current page URL with the `ask` query parameter: ``` GET https://planck-ai.gitbook.io/planck-ai-docs/documentation/core-concepts/workspace-context-and-automations.md?ask= ``` The question should be specific, self-contained, and written in natural language. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.