Legal Technology FAQ: Software, AI and Legal Research

Legal technology covers the software firms use to run matters, bill clients, manage documents, and research the law. These answers explain what each tool category does, how vendors price it, and where AI fits in.

Choosing Legal Software

What is LEDES billing?

LEDES is a family of standard electronic invoice formats that many corporate legal departments and insurers require from outside counsel. Instead of a PDF, the firm sends a structured file that the client's e-billing system can check line by line.

LEDES files usually carry UTBMS task and activity codes, which label what each time entry was for. Common versions include LEDES 1998B and 1998BI. TimeSolv, for example, says it supports LEDES 1998B, 1998BI, and UTBMS task codes.

Not every billing tool supports every format, so test your client's exact file before you buy. Our guide to legal practice management software compares which platforms include it.

Source: TimeSolv legal invoicing (opens in a new tab)

What is three-way trust reconciliation?

Three-way trust reconciliation compares three numbers for a client trust account: the bank statement balance, the trust account ledger, and the total of every individual client ledger. All three should match. Any difference needs to be found, fixed, and documented.

  1. Reconcile the bank statement to the trust ledger, allowing for outstanding checks and deposits.
  2. Add up every client ledger balance.
  3. Confirm the client total equals the trust ledger and the adjusted bank balance.

Legal accounting tools such as CosmoLex say they run three-way reconciliations in a few clicks, but someone still has to review the report. How often you must reconcile depends on your state. See how firms use legal practice management software for trust accounting.

Source: CosmoLex trust accounting (opens in a new tab)

How is eDiscovery software priced?

eDiscovery software is usually priced in one of three ways: by the gigabytes of data you host each month, by user seat, or by a set data tier per case. Most enterprise vendors give quotes rather than public prices.

ModelExampleWhat drives cost
Per GB hostedDISCO, LogikcullData size and hosting time
Per userNextpointNumber of reviewers
Per case tierGoldFynch, 10 GB for $76 a monthData tier and months hosted

Culling data early cuts per-GB bills, while per-user plans favor steady teams with big data sets. For adoption context, see our legal technology statistics.

Source: DISCO platform pricing (opens in a new tab); Nextpoint pricing (opens in a new tab); GoldFynch pricing (opens in a new tab)

AI in Legal Work

Can AI replace lawyers?

Not today. AI tools can speed up drafting, summarizing, and research, but they still get things wrong, and a lawyer has to catch those errors. The lawyer also keeps the judgment calls: advising the client, weighing risk, and deciding what to file.

The error rate is real. A 2024 study by Magesh and colleagues tested AI research tools from LexisNexis and Thomson Reuters. It found that each one produced false information, or "hallucinated," between 17% and 33% of the time.

What firms report about AI use is in AI in law firms statistics, and what no-cost tools can and can't do is in free legal AI tools.

Source: Magesh et al., Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools (arXiv, 2024) (opens in a new tab)

Legal Research

Software Comparisons

Each answer links to a longer guide. If you're shortlisting tools, start with these dated comparisons:

More topics are on the main Legal FAQ page.

About These Answers

Every answer opens with the direct response, then adds a list, a table, or steps. Sources sit under the claim they support, and we checked each one against the live page. Prices change often, so treat any figure as a snapshot from the date we checked it.

These answers explain how legal technology generally works. They aren't legal, accounting, or buying advice for your firm. Written and reviewed by Darlene Aberin. Last reviewed November 6, 2026. Spot an error? Let us know and we'll fix it.