Find answers to common questions about using Ask SIA, Summize Agents, and Custom Agents - including setup, access, permissions, usage tracking, and data security.
Skip to:
General
Access & Setup
Asking Questions
How Responses Work
Understanding & Reviewing Responses
Monitoring Usage & Feedback
Providing Feedback
Troubleshooting
Data & Security
💡 General
What is Ask SIA?
Ask SIA is your intelligent assistant in Summize. It helps you query contract data, documents, and uploaded knowledge to get quick, trusted answers.
What are Summize Agents and Custom Agents?
Summize Agents are pre-built by Summize to help with specific contract and repository tasks, such as reviewing or summarizing content.
Custom Agents are built by you or your administrators using your own uploaded knowledge (e.g. process documents, policies, regulations) to answer company-specific questions.
⚙️ Access & Setup
Where can I access Ask SIA and my Agents?
You can access all Summize and Custom Agents from:
-
The Summize web application (via the Ask SIA button in the top-right)
-
The Outlook & Gmail add-ins (via the Ask SIA button in the add-in toolbar)
-
The Teams or Slack integrations (via the Ask SIA section in the chatbot’s "Get started" message)
- Coming soon: Microsoft Word, HubSpot, Jira, Salesforce
Who can access each Agent?
Access depends on configuration. Admins or Power Users can restrict agents to:
-
All users
-
Only Power Users
-
Specific user groups
Can I ask questions about signed contracts or requests that I do not have access to?
No. Ask SIA takes into account your general visibility and permissions within your Summize tenant. If you cannot see a request in your Request table (i.e. you have a Collaborator or Requestor licence type and therefore have restricted visibility), you cannot query it. If you have restricted visibility in the Repository and cannot see certain folders, you cannot query the documents stored within them.
Support for multi-contract and multi-request queries is coming soon.
How do I configure access permissions for an Agent?
Go to SIA Hub → click Edit next to the relevant agent → open the Settings tab → choose who can access it.
What files can I upload to a Custom Agent for it to use as knowledge?
Supported formats include PDF, JSON, DOC, DOCX, TXT, CSV, and HTML, with each file under 25 MB.
Is it possible to use an online source as knowledge?
No, this is not possible. Knowledge sources must be uploaded files.
Should I combine multiple topics in one Custom Agent or create separate ones?
It depends on how related the topics are and how you want users to interact with them.
Combine topics when:
-
The subjects are closely related and share the same context or purpose (for example, creating a HR Policies Agent for all people-related policy documents, or a Data Protection and Security Agent for all information security related policies).
-
Users are likely to ask cross-topic questions
-
You want a single, consistent voice or set of documents to represent that area of knowledge.
Split topics when:
-
The subjects are distinct or aimed at different audiences (for example, creating an agent for every single company policy is unlikely to yield accurate results as the subjects within it are distinct).
-
You have very large or complex source documents that would make one Agent too broad.
-
You want to manage permissions separately (for instance, giving only HR employees access to people policies).
A good rule of thumb: If you’d expect a different person or team to answer the question in real life, you probably need a different Custom Agent.
💬 Asking Questions
How do I ask a question?
Select an Agent from the list (e.g. the Repository Agent or a Custom Agent), then type your question. The Agent scans relevant knowledge sources (either request/repository documents for Summize Agents or uploaded knowledge for Custom Agents) and provides a response based on the most relevant data.
Can I continue a conversation with the same Agent?
Yes — you can ask follow-up questions in the same thread. Starting a new chat will clear previous conversation history.
Can I use multiple Agents in one conversation?
You can bring a Custom Agent into a Summize Agent thread using the '@' symbol. For example, you could ask the Repository Agent about a contract clause, then tag your InfoSec Agent to check compliance with your data policy.
Note: You can only bring a Custom Agent into a Summize Agent thread — not the other way around.
Can I ask about multiple signed contracts or requests at once?
No, it's not currently possible to ask questions across multiple sources in the Summize agents. For example, the Request Agent cannot answer "show me a list of all requests that are currently in third party negotiation", and similarly the Repository Agent cannot answer "which of our contracts expire next month". Instead you have to first specify the single contract or request you want to query.
🧠 How Responses Work
How does a Summize Repository/Request Agent generate a response?
The Summize agent first tries to determine which single contract/request you're wanting to query, based on the information you've provided (such as name, parties etc) in the question itself. If it can determine the single document, it will go ahead and answer your question with this information it has available to it (e.g. the status, the contract text, tasks, conversations, etc). If it can't determine the single document as there are multiple potential matches, it will ask you to narrow down first before responding.
How does a Custom Agent generate a response?
When knowledge is uploaded, Summize breaks it into smaller data blocks. The Custom Agent scans these blocks, ranks them by relevance, and forms a response based on the most relevant ones. You can configure how many data blocks it considers when building its response. Increasing this number allows for more context; lowering it keeps answers more focused. See more information here.
🔍 Understanding & Reviewing Responses
How can I verify the responses that the Agent provides?
Beneath each response, you will see 'Explain this answer'. Clicking this opens a panel which shows how the reasoning the Agent used formed its response. For Summize Agents, it also includes a link to view the contract/request so you have easy access to view the document. For Custom Agents, the documents used to generate the response are listed in relevance order, alongside the blocks of text contained within them.
Can I see which documents were used to generate a Custom Agent response?
Yes - under Explain this answer, you can expand each source to view the specific text blocks used to create the response.
Why do my Custom Agents responses look too short / too vague?
If your test responses feel incomplete:
-
Try increasing the Max search results (to 20–30) to give the agent more context.
-
Check that your uploaded files contain enough relevant detail - short or highly repetitive files can limit quality.
-
Ensure your instructions clearly define what the agent should focus on or reference.
📊 Monitoring Usage & Feedback
How can I monitor how Agents are being used?
In SIA Hub, open the Agent’s Usage tab to see:
-
The questions users have asked
-
The responses provided
-
Statistics on usage volume and active users
How can I view feedback from users?
Open the Feedback tab to see all thumbs up/down ratings and any accompanying comments. You’ll also see aggregated stats showing overall positive vs. negative trends.
Who can see usage and feedback data?
Power Users have access to SIA Hub and can therefore view this data within your tenant. Summize may review high-level trends (e.g., percentage of positive feedback) for product improvement, but does not use your data or questions to train Ask SIA.
👍 Providing Feedback
How do I provide feedback on a response?
Click the 👍 or 👎 icon next to a response. You can then add comments explaining your rating. This helps your admins and Summize understand how useful the responses are.
Is my data used to train AI models?
No. Summize never uses your data to train or retrain any AI models. Feedback trends are analyzed securely within your environment to improve accuracy, but your contract data and uploaded content remain fully private.
🧩 Troubleshooting
Why can’t I see an Agent in Ask SIA?
You may not have access permissions for that Agent. Contact your Power User or Admin to confirm your access level in SIA Hub.
Why didn’t my Agent return an answer?
Custom Agents: If the uploaded knowledge doesn’t include relevant information, the Agent may not generate a confident response. Try rephrasing your question or confirming that the right documents have been uploaded.
Summize Agents: If you're trying to ask a question across multiple documents, the Agent does not currently support this. If you're trying to ask a question of a single source, but the Agent thinks you're asking across multiple, try rephrasing your question.
Why can’t I add an Agent to my conversation?
Only Custom Agents can be added to Summize Agent threads. Summize Agents can’t be added to Custom Agent threads or combined with other Summize Agents.
🔐 Data & Security
How does Summize keep my data secure?
Summize is built with data privacy and security at its core. All Ask SIA queries, uploaded knowledge, and responses are processed and stored securely within your Summize tenant, hosted on Microsoft Azure.
Summize uses the Azure OpenAI Service, which provides enterprise-grade security, compliance, and data privacy controls. Your data stays within Microsoft’s trusted cloud infrastructure and benefits from the same compliance standards Azure provides.
What does it mean that Summize uses OpenAI via Azure?
By integrating OpenAI through Azure, Summize ensures that:
-
Your prompts, responses, and uploaded knowledge are not used to train or retrain OpenAI's public models.
-
All data is processed within Microsoft Azure’s secure data centres (the location of which depends on your tenant region).
-
Azure OpenAI operates under strict data isolation - your data is never shared with other Azure customers or with OpenAI directly.
-
Both Summize and Azure adhere to enterprise security and compliance frameworks.
What data does Summize store?
Summize securely stores questions, responses, and feedback in your tenant database. This allows your organisation to view usage activity and review feedback trends.
Summize employees can access this data only when necessary - for example, to provide technical support, investigate an issue, or troubleshoot a specific chat or feedback record - and only through secure, audited administrative access.
Where is my data stored?
All customer data, including uploaded documents, chat history, and feedback, is stored securely within your Summize tenant in your region’s Microsoft Azure data centre.
Data never leaves your geographic region and is encrypted both in transit and at rest.
Can Summize or OpenAI view my chats?
Azure OpenAI does not store or reuse your data for training or fine-tuning models.
Summize’s support and engineering teams may, with appropriate authorisation, view chat and feedback records within your tenant for troubleshooting or quality assurance purposes only. Summize never uses this data for model training or analysis outside your tenant.
Who can see my conversations or feedback?
Within your organisation, visibility is determined by your tenant permissions. Power Users and Admins can view usage and feedback data in the SIA Hub.
Summize may also view feedback and usage logs when needed to resolve a ticket or investigate a reported issue.