Legal AI from the ITECS Dallas team
AI Solutions for Law Firms
ITECS builds governed AI for law firms of 50 to 300 people: a pre-bill review agent that checks time entries against client billing guidelines before invoices go out, intake and conflicts assistants, and matter knowledge agents that respect ethical walls. Privileged data stays inside the firm's boundary, and an attorney approves every decision.
Firms do not need another chatbot. They need AI that reads the systems they already run, follows each client's billing guidelines and the firm's matter permissions, and leaves professional judgment where the Model Rules put it: with a supervising lawyer.
Legal signal map
Matters + billing + ethics
Matters
Intake, conflicts, and matter knowledge
Billing
Time entries, guidelines, and realization
Ethics
Privilege, walls, and attorney supervision
Dallas IT discipline behind legal AI
Governed AI for firms, from a team that already runs law-firm IT.
ITECS has managed IT for Dallas organizations since 2002, and its managed-services practice already supports law firms and their legal stack, from Clio and NEOS to iManage and NetDocuments. ITECS AI builds on that operating base: identity, document security, and backup discipline come before any model touches a matter file. Read about ITECS managed IT for law firms.
83% and 84%
Average realization and collection at mid-sized firms (21 or more employees) in 2023.
Clio, 2024 mid-sized firms report11% to 18%
Rise in e-billing invoice rejections during 2025 as clients review invoices with AI, across about 400 firms.
Elite, May 2026Sept. 3, 2024
Since then, N.D. Tex. Local Civil Rule 7.2(f) has required briefs prepared with generative AI to say so on the first page.
N.D. Tex. Special Order 2-98
Operating pressure
Legal AI Starts With Realization, Capacity, and Confidentiality
The first step is not picking a model. It is finding where attorney time leaks between the work and the invoice, and where client confidences need firmer boundaries before AI touches them.
Realization pressure
Work is done and recorded, then reduced: written down at pre-bill, rejected under a client's billing guidelines, or discounted at collection.
- Block-billed and vague entries that billing guidelines reject
- Rate, staffing, and task-code exceptions found after the invoice goes out
- Partners reviewing proformas line by line at month end
- Write-downs recorded without a reason anyone can learn from
Capacity pressure
Associates and paralegals spend hours assembling what already exists: prior work product, intake facts, chronologies, and deadlines.
- New-matter intake and conflicts details rekeyed across systems
- Prior briefs, clauses, and memos hard to find across matters
- Medical-record and deposition chronologies built by hand
- Deadlines pulled from filings and email without a second check
Confidentiality pressure
AI is only usable when privileged material stays inside matter permissions, client terms, and the firm's own policies.
- Ethical walls and matter-level permissions that AI tools must respect
- Client billing guidelines that restrict or condition AI use
- Consumer AI accounts outside the firm's control
- Citations and quotations that must be verified before filing
Example billing signal board
The Metrics a Legal Billing AI System Should Make Visible
A pre-bill agent should show the managing partner where realization leaks, what is recoverable before the invoice goes out, and where guideline exceptions concentrate. These figures are illustrative for a fictional 150-person firm in one billing cycle; discovery replaces them with the firm's own billing data.
Realization bridge
Where billed time leaks
Realization leakage this cycle
$312K
12% of $2.6M worked at standard rates, flagged against client guidelines
Recoverable before invoice
$118K
Fixable by narrative, task-code, or approved-rate updates with partner approval
E-billing rejection rate
14%
Invoices rejected or reduced last cycle, against a firm target under 8%
Entries cleared without exceptions
88%
8,660 of 9,800 entries passed every guideline check
What leadership sees
- Ties every flag to the client's guideline clause, rate letter, and timekeeper
- Separates recoverable fixes from true write-downs before partner review
- Routes every change to the timekeeper and billing partner; nothing reaches the client unapproved
Legal AI use cases
Where AI Creates Measurable Value for Law Firms
Start where the data already arrives and a partner already owns the decision: billing, intake, matter knowledge, and deadline-heavy work.
Pre-Bill & Billing-Guideline Compliance
Check proforma entries against each client's guidelines, rate letters, and staffing rules, flag block billing, vague narratives, and clerical work, and propose fixes before release.
Write-down decisions made before the client challenges a bill
E-Billing Rejection Triage & Appeals
Read rejection and reduction reasons from client e-billing portals, map each to the guideline clause behind it, and draft the appeal or corrected resubmission.
Rejected invoices worked, not written off
Intake & Conflicts Preparation
Normalize party and affiliate names from intake forms and email, run them against the conflicts database, and draft the hit memo for the conflicts attorney.
Faster, consistent clearance with a decision trail
Ethical Wall Provisioning & Checks
Draft the screen when a lateral hire or waiver requires one, then confirm it holds across the document system, Teams, and AI tools before the risk partner signs off.
Walls applied consistently, including to AI
Permission-Aware Matter Knowledge Base
Answer questions over matter files, prior work product, and firm forms, inheriting document and Microsoft 365 permissions, with each answer citing its source.
Less time re-finding the firm's own precedent
Medical-Record & Deposition Chronologies
Build page-cited treatment chronologies, gaps-in-care lists, and transcript digests that a paralegal verifies entry by entry.
Paralegal time moves from assembly to verification
Docketing & Deadline Extraction
Pull dates and trigger events from scheduling orders, ECF notices, and correspondence into proposed docket entries for the docketing clerk to verify.
Fewer manual keying steps and a verification log
Contract Review Playbooks
Apply firm or client playbook positions to inbound drafts and propose redlines with clause-level reasons the responsible attorney accepts or rejects.
Consistent first-pass review for transactional groups
Collections & Receivables Follow-Up
Rank aged receivables by client and matter, draft reminders and payment-plan follow-ups, and trace disputes back to their pre-bill causes.
Earlier attention to aging matters
Flagship enterprise agent
Pre-Bill & Billing-Guideline Agent: Invoices Clients Will Pay
An illustrative solution design, not a reported client result. It fits firms that bill hourly to institutional clients; for a mostly contingency-fee practice, a medical-record chronology agent usually comes first.
Turn each client's guidelines into rules
The agent extracts rate caps, approved timekeepers, staffing limits, non-billable tasks, and AI-use clauses from each client's guidelines, links each rule to its clause, and the billing manager approves the set.
Human approval point
Check every proforma entry
Fixed checks catch rate, timekeeper, increment, task-code, and duplicate problems. The model flags block billing, vague narratives, clerical work, and extra attendees, citing the clause each time.
Propose the fix, never more time
It drafts narrative rewrites from facts already in the entry or matter record and suggests splits, task-code changes, or write-downs. It does not add hours or tasks, or change a date or timekeeper.
Route to the timekeeper and billing partner
Timekeepers confirm that each rewrite is accurate. The billing partner approves write-downs, staffing exceptions, and release, and rate questions go to the billing manager.
Human approval point
Release, then learn from rejections
Approved changes post to the billing system, which still produces the invoice and LEDES file. E-billing rejections flow back, map to the clause behind them, and become draft appeals for partner approval.
What it reads
- Proforma entries: timekeeper, hours, rate, narrative, and UTBMS codes from the billing system
- Client guidelines, rate letters, and approved-timekeeper lists from the document system
- Rejection and reduction reasons from client e-billing portals
- Matter records for narrative context, within each user's matter permissions
What it never does on its own
- Adds hours or tasks, or changes an entry's date or timekeeper
- Changes an entry or releases an invoice without timekeeper and billing-partner approval
- Submits an invoice to a client's e-billing platform
- Shows walled matters to screened users, or lets a vendor train on firm data
Two kinds of legal agents
Firm-Wide Agents and Desktop Agents
ITECS builds two kinds of agents for law firms. Each has its own owner, controls, and rollout path.
Engineered by ITECS
Enterprise hosted agents
Engineered by ITECS, connected to the firm's practice, document, and billing systems, and run in a governed hosted environment with a review portal and audit trail. Firm leadership owns them: the managing partner, COO, or CFO.
- Pre-bill review against client billing guidelines
- New-matter intake and conflicts preparation
- Matter knowledge search that honors ethical walls
Built with your team
Personal desktop agents
Built with attorneys and staff in Claude, ChatGPT, or Microsoft Copilot on their own machines, then rolled out through hands-on training under the firm's AI policy. The practice group or department head owns them.
- Discovery-response formatting, exhibit lists, and privilege-log first drafts for paralegals
- LEDES and UTBMS code-check helpers for billing coordinators
- Research-memo outlines with mandatory citation checks for associates
Systems legal agents read
- Clio
- NEOS
- iManage
- NetDocuments
- Aderant
- Elite 3E
- Intapp
- Juris
- PCLaw
- Microsoft 365
- LEDES and UTBMS e-billing
Technologies ITECS evaluates and integrates; listing a product does not imply a formal partnership. Discovery confirms which system holds the authoritative record.
Start here
Legal AI Readiness Assessment
ITECS starts by mapping the business case, data readiness, ethics requirements, and governance model before recommending an agent build.
Map firm priorities
Identify the billing, intake, knowledge, and deadline work where attorney and staff time leaks or risk surfaces too late.
Review systems and permissions
Assess practice, document, billing, and conflicts data for ownership, quality, matter permissions, and ethical walls.
Rank use cases by economics
Prioritize by realization recovered, hours returned to billable work, risk reduced, and feasibility, starting where data already arrives.
Write the governed path
Document the AI policy, client-guideline constraints, approval points, audit logging, and vendor terms before any production build.
Governance
Built Around the Rules Lawyers Already Follow
ITECS designs legal AI around the professional-responsibility rules, client terms, and firm policies that already govern the work.
Professional responsibility by design
ABA Formal Opinion 512 and Texas Ethics Opinion 705 apply existing duties to generative AI: competence, confidentiality, communication, supervision, candor, and reasonable fees. The designs follow those duties.
- Lawyers verify and remain responsible for AI-assisted work product
- Informed client consent before client information enters a self-learning tool; boilerplate is not enough
- Hourly entries bill actual time, not time a tool saved or time spent learning it
- Court-filing checklists cover disclosure rules such as N.D. Tex. Local Civil Rule 7.2(f)
Matter-level boundaries
Agents inherit the firm's permissions instead of creating new ones, so an ethical wall in the document system is a wall for the agent too.
- Matter and ethical-wall permissions enforced from the source systems, denied when unconfirmed
- Client guideline clauses on AI use tracked, so the firm knows who restricts or requires disclosure
- Vendor data-use, retention, and training terms verified before production
- SOC 2 Type II applies only if Promus managed cloud hosting is selected
Security
Security for Legal AI Workflows
Legal AI touches privileged communications, client confidences, and work product. ITECS designs these systems with matter-scoped access, audit logs, and attorney approval before anything leaves the firm.
FAQ
Legal AI FAQ
Is it ethical for our firm to use generative AI on client matters?
Yes, within existing rules. ABA Formal Opinion 512 and Texas Ethics Opinion 705 apply current duties of competence, confidentiality, communication, supervision, candor, and reasonable fees, and both expect lawyers to verify everything a tool produces.
Do we need client consent before using AI on a matter?
ABA Formal Opinion 512 requires informed consent before client information goes into a self-learning tool, and says boilerplate engagement-letter language is not enough. ITECS designs agents that run under no-training terms and inherit matter permissions, while consent decisions stay with the firm and its counsel.
We already run Aderant or Intapp. Do we still need a pre-bill agent?
Maybe not. Both vendors sell billing-compliance tools inside their own systems, so ITECS compares the native module first and builds only the cross-system, firm-specific layer the firm still needs.
How do you prevent fabricated citations?
Agents answer from the firm's own records with a citation to the source, and a lawyer verifies anything bound for a court. Mata v. Avianca drew court sanctions in 2023, and N.D. Tex. Local Civil Rule 7.2(f) now requires AI disclosure on briefs.
Where should a 50 to 300 person firm start?
Start with an AI use policy and a readiness assessment, then pilot one enterprise agent tied to a number the firm already tracks, such as realization or e-billing rejections. Expand to intake, conflicts, and desktop agents once governance is proven.
