Asking questions of your firm's own data with an AI assistant
By the ChamberOne team 5 minute read
Every partner has a list of questions that the firm's records could answer, if only someone had an afternoon to dig. Which matters have a deadline in the next fortnight? Which clients owe us the most? How much unbilled time sits on that big dispute? Traditionally the answer meant a report request, a spreadsheet and a wait. An AI assistant that reads the firm's own records promises the answer in a sentence. The promise is sound only if a few conditions hold.
Key points
- Questions about deadlines, workload, billing and matter status suit an assistant that reads your records.
- Insist on traceable answers, respect for permissions, accurate numbers and honest failure.
- Data quality sets the ceiling, so fix recording habits first.
- Check where data goes, and confirm a source record before you act on an answer.
What 'asking your own data' means
The idea is simple. Instead of building a report, you type or say a question in plain words, and the assistant looks through your records and replies. The records might be matter details, diary entries, tasks, time entries, invoices and notes. The assistant works out which records are relevant and writes an answer.
This differs from asking a general chatbot, which answers from what it learned in training. A question about your firm should be answered from your firm's data, and the answer should say where it came from.
Questions that suit it
- Deadlines: what hearings, filings and limitation dates fall within the next two weeks, and who is responsible for each?
- Workload: which fee earners have the most open matters, and which matters have had no activity for a month?
- Billing: which matters have unbilled time above a level I choose, and which invoices are overdue?
- Clients: what matters have we done for this client, and who was the partner on each?
- Status: summarise where this matter stands, based on the notes and tasks.
Questions that do not
Be wary of asking the tool to do things records cannot support. 'Will we win this case?' is a judgment, not a lookup. 'What is the limitation period?' is a legal question, answered from law and not from your database. 'Why is revenue down?' invites a plausible story that may not be true. For these, use the assistant, if at all, to pull together the facts, and let a person do the reasoning.
- Which hearings and filings fall in the next two weeks?
- Which matters have unbilled time or overdue invoices?
- What matters have we done for this client?
- Where does this matter stand, from the notes and tasks?
- 'Will we win this case?' is a judgment, not a lookup
- 'What is the limitation period?' is answered from law
- 'Why is revenue down?' invites a plausible story
Four tests before you trust the answers
Whatever product you use, check the following before relying on it for firm management.
- 1
Traceability: does every answer point to the records it came from, so that you can click through and check? If the tool cannot show its sources, treat each answer as a guess.
- 2
Permissions: does it respect who may see what? A junior should not learn about a restricted matter by asking a question. Ask how the tool treats confidential or restricted matters and test with a real example.
- 3
Accuracy on numbers: totals, counts and dates need to match the underlying reports. Compare a sample of answers against a standard report.
- 4
Honest failure: does it say 'I do not know' or 'there is no record', or does it make something up?
Data quality sets the ceiling
An assistant can only find what is recorded. If deadlines live in people's heads, time is entered a fortnight late and matters are never closed, then the answers will be wrong in a confident tone. Before you judge the tool, look at your own data: are matters assigned to the right people, are statuses current, is time up to date?
In practice, introducing a question-asking tool often shows up gaps in recording habits that were always there. That is a benefit, provided someone then acts on it.
Confidentiality and where the data goes
Your records hold client information, so check how the assistant gets at them. Does the question and the relevant record text leave your system and go to a third-party AI provider? In which country? Is it retained, and can it be used for training? What does the contract say? Ask for the answers in writing, and check whether the firm can switch the feature off or choose the model.
Check also the firm's own duties and client terms. Some clients restrict the use of AI on their data. A firm-wide tool should let an administrator apply such restrictions to particular clients or matters.
Rolling it out sensibly
- Start with the partners and practice manager, who know what the right answers look like.
- Make a list of twenty real questions. Ask them in the tool and compare against known answers from reports.
- Record where it was right, wrong or vague. Share the results.
- Write a short guide for staff: what to ask, what not to ask, and how to check an answer.
- Review an audit record of use, at least at first, to see what people are asking and whether it is appropriate.
Keep a person in the loop
A convenient answer is still a prompt for action by a person. If the assistant says a hearing is next Tuesday, glance at the diary entry. If it says a client owes a large sum, check the ledger before you phone them. The goal is to reach the right record quickly, not to stop looking at records. Used that way, a good question-asking tool shortens the path from a partner's question to the evidence, which is worth having.
Keep the scope modest at first. Ask the assistant for lists and summaries, and leave decisions, advice and client communication to people. As trust grows and you see where the tool is dependable, you can widen its use. If it gets something badly wrong, record it, tell the supplier and tell your colleagues, so the whole firm learns the same lesson at once.
Asking AI in ChamberOne
- Request a matter brief, a document analysis, suggested tasks or an answer from Ask AI.
- Each output stays a draft until someone accepts it, and every AI call is recorded in the audit log with the model and token counts.
What it does not do
- AI output can be wrong or incomplete. It does not replace a lawyer's review.
- With AI features on, the text needed for a request is sent to the AI provider for that feature, outside the country where ChamberOne is hosted.
This article is general information about the law, legal practice and technology, current when it was published. It is not legal advice. Laws and guidance change: check the sources linked above and take advice on your own circumstances from a lawyer qualified where you are.
Read next
Onboarding a new client matter: an AI-assisted step-by-step checklist
Walk through every step of opening a new matter — from conflict check to task assignment — showing where AI saves time and where only a lawyer can decide.
6 minute readUsing AI matter briefs to pick up an unfamiliar file quickly
How a generated brief speeds up handovers and cover, and how to check it against the file before acting.
5 minute readAI task suggestions and matter phases: standardising how work starts
How matter-type templates and suggested tasks help new files start consistently, and why each suggestion needs review.
5 minute read