AI can help distinguish billable from non-billable work, but it should usually make a classification suggestion, not the final billing decision.
For lawyers, whether an activity is billable often depends on more than the activity itself. A meeting may be billable when it concerns a client matter and non-billable when it is an internal firm meeting. Research may be billable when it supports a client assignment, but general professional development is normally treated differently. Travel, internal conferences, administrative work, and business development can also depend on the engagement terms and client billing guidelines.
The useful role for AI is to combine work activity with matter context and billing rules, then present the suggested status for review.
What Does AI Billable-Time Classification Mean?
AI billable-time classification is the use of software to suggest whether captured work should be recorded as:
Billable
Non-billable
No charge
Internal
Uncertain or review required
A classification engine may consider:
Client
Matter
Activity type
Document or workspace
Participants
Calendar context
Prior approved entries
Task or activity code
Matter-specific billing rules
User corrections
The result should remain reviewable.
Why Billable vs Non-Billable Is Harder Than It Looks
The same activity can have different billing treatment in different circumstances.
Consider an internal conference.
It may be:
Billable when lawyers discuss substantive strategy for a client matter and the client permits internal conferences
Non-billable when it is a general practice-group meeting
No charge when the firm chooses not to bill the client
Restricted when an outside counsel guideline limits internal conferences
The software therefore needs context, not just a label such as “meeting.”
What Signals Can AI Use?
Useful signals can include:
Matter workspace
Document metadata
Email participants
Calendar attendees
Meeting title
Client name
Matter identifier
Browser or research context
Prior approved classifications
User-selected matter
Billing-code history
Client billing rules
A stronger signal is one that connects the work to a specific engagement.
A weaker signal is a generic activity name with no client or matter context.
Usually Billable, Usually Non-Billable, and Context-Dependent Work
Activity | Often billable | Often non-billable | Why context matters |
Client call | Yes | No | Must relate to an active matter |
Legal research | Yes | Yes | Client research differs from general learning |
Document review | Yes | Yes | Matter document review differs from internal review |
Internal meeting | Sometimes | Yes | Client rules may restrict internal conferences |
Travel | Sometimes | Yes | Agreement may limit or discount travel |
Training | Rarely | Yes | Usually supports the firm rather than a client |
Business development | No | Yes | Normally not charged to a client |
Administrative work | Usually no | Yes | Some client rules prohibit clerical work |
Often | Yes | Matter-related email differs from internal administration |
These are general examples. The engagement controls.
Client Billing Guidelines Change the Answer
Outside counsel guidelines may define:
Which timekeepers may bill
What work is compensable
Whether internal conferences are permitted
Research restrictions
Travel treatment
Administrative exclusions
Required task codes
Staffing restrictions
Budget limits
Preapproval requirements
An AI system that ignores matter-specific rules may classify the activity correctly in a general sense but incorrectly for the client.
Classification should therefore be connected to the matter’s current billing requirements whenever possible.
Fee Arrangements Also Change the Meaning of “Billable”
Hourly billing is not the only model.
A firm may use:
Flat fee
Fixed fee by phase
Capped fee
Subscription
Contingency fee
Blended arrangement
Portfolio fee
Under an alternative fee arrangement, the firm may still track work even when individual time entries do not appear on the client invoice.
The status may therefore need to distinguish:
Client-related work
Invoiceable hourly work
Included-in-flat-fee work
No-charge work
Internal work
This is more useful than forcing all activity into only two categories.
What AI Should Not Decide by Itself
AI should not independently determine:
Whether a fee is ethically permissible
Whether a client agreement allows a charge
Whether work should be written off
Whether a lawyer should exercise billing judgment
Whether a disputed activity should be invoiced
Whether a flat-fee matter has become more profitable
Whether confidential information should appear in a narrative
Those decisions require the applicable agreement, firm policy, professional rules, and human judgment.
ABA Model Rule 1.5 requires lawyer fees to be reasonable. That obligation is not transferred to software.
Handling Uncertain Activity
A useful system should be able to say, in effect:
This looks matter-related, but billing status is uncertain.
Uncertain entries can be routed for review when:
Matter confidence is low
Client rules conflict
Activity spans client and internal work
The fee arrangement is unusual
Several matters may apply
Work is administrative but matter-related
Prior classifications are inconsistent
This is better than forcing a confident answer from incomplete evidence.
Human Review Workflow
A practical workflow is:
Capture the work activity.
Suggest the client and matter.
Suggest billable, non-billable, or another status.
Show the evidence supporting the suggestion.
Apply client or matter rules.
Let the lawyer correct the status.
Let the lawyer edit the narrative and codes.
Save the reviewer’s correction.
Release only approved entries.
The goal is to reduce manual entry without removing billing judgment.
What Law Firms Should Measure
Useful metrics include:
Billable-status acceptance rate
Billable-status correction rate
Non-billable-status correction rate
Entries moved to no charge
Entries changed during prebill
Matter correction rate
Narrative edit rate
Client-guideline exceptions
Review time per user
Missing-time rate
A high correction rate may indicate weak rules, poor matter data, or insufficient context.
Implementation Checklist
Before automating classification, define:
Firm-wide billable and non-billable categories
Matter-specific exceptions
Client billing guidelines
Alternative fee treatment
No-charge rules
Administrative-work policy
Travel policy
Research policy
Internal-conference policy
Approval responsibility
Escalation for uncertain entries
Audit and correction process
Technology works better when the policy is explicit.
Frequently Asked Questions
Can AI decide whether legal work is billable?
AI can suggest billing status using matter context and rules, but the lawyer or authorized reviewer should make the final decision.
Can the same activity be billable for one client and non-billable for another?
Yes. Client guidelines, fee arrangements, matter scope, and firm policy can change the treatment.
Should law firms track non-billable work?
Yes. Non-billable time can help firms understand administration, training, business development, capacity, and the true effort required to operate the practice.
What should happen when AI is uncertain?
The entry should be flagged for review rather than automatically classified and released.
Does MIRA automatically bill every activity it captures?
No. MIRA supports a review workflow in which users decide what is edited, excluded, saved, merged, or released.
