This product is currently in early access.
Overview
While reviewing your firm management reports, you can use Ask Acumen to gain a deeper, more robust understanding of your data. Like other AI-based chatbots and agents, you can submit natural language questions about firm or timekeeper performance and receive contextualized, domain-aware explanations grounded in your Actionstep data. This should make it easier to interpret complex financial metrics while quickly discovering opportunities for your firm.
Please note the following when using Ask Acumen:
- Ask Acumen is powered by Amazon Quick AI Assistant. To learn more, see https://docs.aws.amazon.com/quick/.
- Ask Acumen only ever sees your firm's data — the same information available in your dashboards and reports. Every firm has its own dedicated AI with its own isolated data, so there's no way for your information to be seen by or mixed with another firm's data. (See Ask Acumen FAQ (below) to learn more about how the AI is managed.)
- Questions or prompts you submit through Ask Acumen can be written in plain English / natural language.
- The questions you ask the chatbot must be related to the data displayed on the Firm Management dashboard.
- When submitting your query, it might be helpful to define the time period you want covered in your analysis. Or, if needed, you can always use follow-up queries to narrow your response by a date or date range.
- The Ask Acumen panel includes some sample prompts you can use. These prompts may be updated based on your chat history, meaning the questions may change based on the types of questions you're asking. These questions may also be suggested based on what you're viewing on the dashboard.
- As the agent is processing and answering your question, it records the different tasks it completes and logs them under Response Events, which you can view to see the process that was taken to reach the response.
Completing Your AI-Assisted Search
To use Ask Acumen:
- In Practice Management, go to Insights > Business Intelligence.
- In the Insights panel on the left side of the page, click Firm Management. The page is updated to show your Firm Management Dashboard.
- Click the Ask Acumen button, located in the top-right corner of the page. The Ask Acumen panel is expanded on the right side of the page.
- Enter your question or prompt in the Ask Acumen field and press Enter. The search is submitted and the answer is displayed in the panel. [SW1]
- Optionally, rate the response by clicking the thumbs up or thumbs down button.
- Continue to fine-tune your results by asking additional questions or using the options included with the response to explore further results.
- Complete any of the following tasks:
- Click Sources to see which databases were used to come up with your results.
- To view a table in your response in full detail, either click the View more icon or link.
- Click the Copy icon at the end of the response to copy and paste the response in another application.
- Start a new chat session by clicking New conversation above the chat prompt field. (See "Using Chat History" (below) to learn more about how Actionstep keeps a record of previous chats and how you can access those sessions again.)
Using Chat History
Each time you close the Ask Acumen panel and reopen it, a new chatbot session is started. However, if you want to continue an existing session, you can click History (located in the top-right corner of the panel), find the specific session you want to revisit, and continue with the conversation. [screenshot]
For example, let's say you are using Ask Acumen to analyze a particular data point in one of your Firm Management cards, but you need to review something specific on your dashboard. When you close the panel and then reopen it to continue with your analysis, you will need to view your History and relaunch that session.
You can remove saved sessions by clicking the Delete icon next to the session name.
Ask Acumen FAQ
Actionstep's Ask Acumen beta adds a conversational analytics assistant to the Actionstep Business Intelligence (BI) experience. The beta is built on Amazon Quick, AWS's generative business intelligence service, and runs entirely within Actionstep's existing AWS environment. This document answers common questions about how data is handled, secured, and governed during the beta.
- Is any of our data used to train AI models?
- Which LLM powers the AI assistant?
- Where is our data processed and stored?
- Is data sent to another country?
- Is data retained?
- Are audit logs available for AI activity?
- Is data encrypted at rest and in transit?
- Is customer data used for model improvement?
- Is a data processing or AI privacy statement available?
Is any of our data used to train AI models?
No. AWS does not use customer data — including the questions you ask the assistant and the answers it generates — to train or improve the underlying AI models. This is AWS's published policy for Amazon Quick and applies automatically; it is not an opt-out setting. Actionstep does not use your data to train AI models either.
Which LLM powers the AI assistant?
The assistant is powered by Amazon Quick, which uses large language models hosted and managed by AWS through Amazon Bedrock. The models operate entirely as a managed service within the AWS environment — no third-party AI vendor receives or processes your data. AWS manages model selection and updates as part of the service.
Where is our data processed and stored?
Your data is stored and processed in the AWS region that matches your firm's geography, and each firm's data is isolated within its own region:
- Australian firms are hosted in the AWS Asia Pacific (Sydney) region (ap-southeast-2).
- New Zealand firms are hosted in the AWS Asia Pacific (Sydney) region (ap-southeast-2). Amazon Quick is not yet available in the AWS New Zealand (Auckland) region, so NZ firms are served from Sydney — meaning their data resides in Australia and is processed within the Australia geography (Sydney and Melbourne).
- UK firms are hosted in the AWS Europe (London) region (eu-west-2).
- US firms are hosted in the AWS US East (N. Virginia) region (us-east-1).
Each firm has its own dedicated data warehouse, BI datasets, AI assistant, and AI user records within its region — firms' data is never co-mingled.
Is data sent to another country?
No — your data stays within your own geography. For AI processing, AWS uses "cross-region inference," but processing always remains within the same geography as your data, and the geographies never mix:
- The United States geography uses only US AWS regions (N. Virginia, Ohio, Oregon).
- The Europe geography uses only European AWS regions (Frankfurt, Ireland, London, Paris, Stockholm).
- The Australia geography uses only Australian AWS regions (Sydney, Melbourne).
So a US firm's data stays within the United States, a UK firm's data within Europe, and Australian and New Zealand firms' data within Australia. All data in motion is encrypted across AWS's private network.
Is data retained?
AWS does not retain your AI inputs or outputs for model training or improvement. Your underlying BI data remains within Actionstep's QuickSight environment under Actionstep's control, subject to existing data management practices.
During the beta, Actionstep retains chat interactions (questions asked and responses given) for evaluation purposes. This allows us to assess answer quality, identify gaps, and improve the assistant before general release. These records are handled under Actionstep's standard data protection controls, remain within the same AWS region as your firm's data, and are used solely for product evaluation and improvement — not for AI model training.
Are audit logs available for AI activity?
Yes. The platform is integrated with AWS CloudTrail, which records actions taken in the QuickSight environment — including who made a request, when, and from where. In addition, the chat interactions retained during the beta provide a record of AI assistant usage. If you have specific audit or reporting requirements, let us know and we can discuss what can be made available.
Is data encrypted at rest and in transit?
Yes, both. Data at rest is encrypted using AWS Key Management Service (KMS), including data held in QuickSight's SPICE in-memory engine, which uses hardware block-level encryption. Data in transit is encrypted using TLS, covering transfers between data sources, the BI platform, and your browser.
Is customer data used for model improvement?
No. AWS explicitly states that no data, inputs, or outputs from Amazon Quick are used to improve the underlying models. Actionstep's beta evaluation of chat interactions (described above) is limited to assessing and improving the assistant's configuration and knowledge base — it does not involve training any AI model on your data.
Is a data processing or AI privacy statement available?
The platform inherits AWS's compliance certifications, including SOC, ISO 27001, ISO 27018, ISO 9001, and GDPR-aligned controls. AWS publishes detailed documentation on data protection for the service. For Actionstep-specific terms covering the beta — including the chat retention described above — please contact your Actionstep representative, who can provide the relevant beta terms and privacy documentation.
Updated