Author: Onit

Where Legal Hold Processes Break Down and What a Stronger Process Looks Like 

Falling dominoes halted by a gap in the sequence, symbolizing how strong legal hold processes prevent compliance risks, preservation failures, and litigation challenges.

Legal hold problems rarely start with one dramatic failure. More often, risk builds through smaller process gaps: a custodian list that goes stale, a reminder that goes out late, a preservation issue that isn’t surfaced, or a release that isn’t clearly documented. Individually, those gaps may seem manageable. Together, they can make it harder to demonstrate the reasonable, consistent preservation process Legal intended to follow. 

In a recent Onit webinar on closing the gaps in legal hold processes, the throughline was straightforward: legal holds are rarely managed in one place. Most organizations run some mix of specialized technology and manual process across identifying custodians, tracking HR changes, monitoring preservation status, and reporting. Each handoff between those pieces is a place where something can slip. 

A Useful Frame: The Information Governance Reference Model 

Before getting into where things break down, it’s worth grounding the discussion in a framework that already exists for this. EDRM, the organization behind the Electronic Discovery Reference Model, maintains an Information Governance Reference Model that lays out a cross-functional approach to managing information, including stakeholder alignment, retention decisions, and the ongoing processes that connect them. It’s a useful reminder that legal holds don’t exist in isolation. They sit inside a broader information governance structure that touches Legal, IT, HR, and Records Management alike. 

With that frame in mind, here’s where the hold lifecycle itself tends to break down. 

Where the Process Breaks Down, Stage by Stage 

Identifying custodians. This is the first place data goes stale. People change roles, leave the company, or pick up new responsibilities, and a manual process often doesn’t catch it. Without a way to track this in real time, the custodian list on day one of a hold can already be wrong by day thirty. 

Issuing notice. Instructions vary depending on who is writing them and when. Without a defined template covering purpose, scope, suspension of routine deletion, and confidentiality, notices can end up inconsistent across groups handling similar matters. 

Acknowledgment tracking. Once a notice goes out, Legal needs visibility into whether custodians have received and acknowledged it. In a manual process, those acknowledgments often land in an inbox with no structure behind them. In larger matters where a hold notice might go to hundreds of people, identifying exactly who among them hasn’t responded becomes a real burden, not a minor annoyance. 

Follow-up and escalation. Reminder timing tends to be inconsistent without something automated managing it. Instead of a defined cadence, follow-up often depends on someone remembering to check for non-responses, which is exactly the kind of gap that shows up later if a hold gets challenged. 

Preservation. A hold ultimately has to translate into reasonable steps to preserve potentially relevant information. But preservation activity can sit outside the day-to-day hold workflow, leaving Legal and IT to reconcile status across systems. When preservation requests, status, failures, and follow-up aren’t visible alongside the hold itself, identifying exceptions becomes harder. 

Release and reporting. Releasing a custodian from a hold, and documenting that release, is easy to do inconsistently. Reconstructing who is or isn’t currently subject to a hold, at best, means pulling data from multiple systems. At worst, it’s a fully manual exercise every time someone asks. 

What a Stronger Process Looks Like 

A few foundations show up consistently in a process that holds up under scrutiny: 

  • Connection to matter context. The matter explains why a hold exists; the hold addresses what happens next. When hold activity, including notices, acknowledgments, escalations, and releases, stays tied to the matter record, Legal gets a complete view without reconciling separate systems. 
  • Clear custodian accountability. Knowing what each custodian is required to do, whether they’ve acknowledged it, and whether they’re actually following through, rather than assuming a notice sent equals a notice actioned. 
  • Consistency in notices, reminders, and escalations. The same level of detail and the same cadence applied every time, not dependent on who happens to be running a particular matter. 
  • Real visibility into status. Surfacing what’s complete, overdue, or needs attention through alerts, rather than requiring someone to run a report to find out something already went wrong. 

The point of technology in this process is to support it, not replace judgment. Automating notice delivery, acknowledgment tracking, and status alerts doesn’t remove Legal’s role in deciding scope or handling exceptions. It removes the administrative burden that makes those small gaps likely to happen in the first place.

Legal Invoice Review Has a Coverage Problem, Not Just a Savings Problem

Tall stack of documents and file folders next to dozens of sticky notes on a wall, representing invoice review workflows, administrative workloads, compliance tracking, and document management challenges.

Ask most legal operations leaders what percentage of invoice lines they actually review, and the honest answer is rarely “all of them.” It’s not for lack of diligence. Invoice volume has grown faster than headcount, and something has to give. What usually gives is coverage. 

That’s a bigger problem than it sounds like at first. Every line that doesn’t get substantive review is spend that passes through without anyone confirming it matches the billing guidelines the firm agreed to. Multiply that across a full outside counsel panel, and a meaningful share of legal spend is effectively running on trust rather than verification. 

The Coverage Gap 

The first question worth asking internally is simple: what percentage of invoice lines receive substantive review today? For most departments, the answer isn’t close to 100%. Reviewers triage by invoice size, by vendor, or by whatever catches their attention in a given cycle. That’s a reasonable response to limited capacity, but it means the guidelines a department spent real effort negotiating are only being enforced some of the time. 

The Vendor Comparison Gap 

Uneven coverage doesn’t just mean some spend goes unchecked. It also makes vendor comparisons unreliable. If Firm A’s invoices happened to get closer scrutiny this quarter than Firm B’s, any conclusion about which firm bills more cleanly is confounded by how closely each one was actually reviewed, not just how compliant each one actually is. 

That matters beyond any single invoice cycle. Legal operations teams that want to negotiate rates, rebalance a panel, or make the case for moving work to a different firm need confidence that the underlying data reflects consistent scrutiny across the board, not whichever invoices happened to get a closer look that month. Without that baseline, spend comparisons across the panel are measuring review effort as much as they’re measuring vendor behavior. 

The Visibility Gap 

This compounds both problems above. Without a clear view into what was reviewed, what was flagged, what was adjusted, and what was disputed, leadership is working from incomplete information. It’s harder to answer basic questions that should inform vendor management: which firms consistently comply with billing guidelines, which categories of violations show up most often, and whether the review policy itself is working the way it was designed to. 

Reframing the Value: Coverage and Time, Not Just Savings 

It’s tempting to pitch better invoice review purely on the dollars it can recover. That framing runs into a real limitation: guideline violations tend to get corrected over time as vendor behavior adjusts to closer scrutiny, so a savings number that looks compelling in month one will not hold at the same rate a year later. 

The more durable case is about capacity. First-pass invoice review is repetitive, rules-based work that doesn’t need to consume a reviewer’s day. When that first pass happens across every line, on every invoice, reviewers get back time to spend on the judgment calls that actually require their expertise, whether that’s a genuinely ambiguous billing question or a broader look at outside counsel strategy. Coverage expands without adding headcount, and the resulting data is consistent enough to actually trust when comparing vendors. 

Onit’s Spend Agent is built as an agentic approach to invoice review for exactly this reason: applying billing guidelines across every line of every invoice, so coverage isn’t limited by reviewer capacity and vendor comparisons aren’t distorted by how closely any one invoice happened to get reviewed. 

Coverage is the foundation. Once it’s in place, the next layer of accuracy comes from matter-specific context. For more on how engagement letters and matter-level terms refine what “compliant” actually means for a given piece of work, see Engagement-Aware Invoice Review: Why Matter Context Matters. 

AI Outperforms Experienced Lawyers in Legal Invoice Review

A legal operations or finance professional reviewing an invoice on screen next to a printed billing guidelines document, in a polished modern office setting. Avoid robots, glowing AI brains, gavels, and generic courthouse imagery.

Every legal department reviews outside counsel invoices the same way it always has: line by line, guideline by guideline, one tired reviewer at a time. It is slow, inconsistent, and expensive, and most legal teams have accepted that as the cost of doing business. 

A study by Onit’s AI Center of Excellence benchmarked large language models against early-career lawyers, experienced lawyers, and legal operations professionals on the same task: reviewing line items from anonymized client and synthetic invoices for compliance with billing guidelines. The result was not a close call. Two models achieved a 0.92 invoice-decision F-score, compared with 0.72 for experienced lawyers. One of those models completed reviews about 22x faster than the fastest human group. 

The results come from a published research paper. They give legal teams a reason to examine where AI-assisted review could fit in their legal spend process.  

What the research actually found 

The study, “Better Bill GPT: Comparing Large Language Models against Legal Invoice Reviewers” (arXiv:2504.02881), set out to answer a question legal operations teams have quietly wondered for years: is manual invoice review actually as accurate as it feels? Researchers built a ground-truth set of invoice decisions validated by expert legal professionals, then tested how closely different reviewer groups matched it, human and machine alike. 

Experienced lawyers, the highest-scoring human group in the study, achieved a 0.72 invoice-decision F-score. GPT-4o and Gemini 2.0 Flash Thinking each achieved 0.92. An F-score combines precision and recall; it is not a simple accuracy percentage. The speed difference was just as stark: Gemini averaged 8.68 seconds per invoice, about 22x faster than experienced lawyers, who averaged 194.75 seconds. Those times cover the review task measured in the study, not the full process of resolving and approving an invoice. 

For a function that has run on the same review model for decades, that is a meaningful signal, not a marginal improvement. 

Why manual review misses what it misses 

The accuracy gap is not really about human effort or diligence. It is about what a reviewer can reliably catch line by line, invoice after invoice, at volume. Traditional invoice validation tools, and human reviewers under time pressure, tend to check for the presence of required fields rather than the substance of what was billed. 

Consider a task description that reads: “General status and strategy work on the file, 6.2 hours.” It has a valid task code, a plausible hour count, and nothing to verify. No document, call, or decision is named, yet it passes format checks because format checks only confirm a field exists, not that the content behind it is defensible. Or take two entries describing what amounts to the same revision in slightly different language. Each line looks fine in isolation. Read in context against your guidelines, they describe duplicate work. 

These are illustrative examples, not drawn from a real matter, but they represent exactly the kind of pattern that AI-based review, working in natural language context rather than keyword rules, is built to catch. 

What this means for legal and finance teams 

None of this means AI should run unsupervised across every invoice a legal department receives. The research points to something more useful: a published, independent benchmark that legal and finance leaders can point to when deciding how much of the review process to automate, and how much to keep in human hands. 

That distinction matters in practice. Solutions like Onit’s Spend Agent are built around it, letting legal operations teams choose, vendor by vendor, whether findings route to a human for a final decision or whether compliant invoices move forward automatically. Every flagged line comes with a clear explanation tied to the specific guideline it violates, so legal and finance teams can make faster, defensible decisions instead of taking an AI’s word for it. Human oversight stays in the loop by design, not as an afterthought. 

For a General Counsel or CFO trying to get ahead of unpredictable legal spend, or a Legal Ops leader tired of chasing billing-guideline compliance manually, the research is a reason to look at what AI-assisted review actually does today, backed by data instead of a sales deck. 

See the research applied to real invoice patterns 

The full study is worth reading if you own legal spend. If you want to see how these findings translate into day-to-day invoice review, our ROI datasheet walks through the block-billing, vague-description, and duplicate-billing patterns above in more detail, and our upcoming whitepaper digs deeper into the methodology and what it means for legal operations teams building the case for AI-assisted review. 

Ready to see how AI-assisted invoice review holds up against your own billing guidelines? Book a demo with Onit and find out. 

What Model Context Protocol Means for the Future of Legal AI 

Professional using a tablet surrounded by AI-powered digital interface elements, highlighting connected workflows, cloud technology, analytics, and business innovation.

In early August, Onit brought together legal operations leaders for a customer-exclusive focus group on Model Context Protocol (MCP). The discussion explored where MCP fits in contrast to APIs and what legal teams need from the next generation of AI integration. 

Collectively, customers represented billions of dollars in legal spend. One point became clear during the discussion: legal teams are not waiting for one AI platform to win. 

They’re already working across Copilot, Claude, Harvey, Glean, Wordsmith and other tools. Some are even testing the same use cases across multiple platforms to compare quality, productivity and cost.  

Add those tools to in-house solutions and an already crowded legal technology stack, and the cost of fragmentation rises. Legal teams may have to find the same information repeatedly, reconcile answers drawn from different sources and determine whether each AI tool is respecting the same permissions and underlying data. 

The challenge is not simply connecting more technology. It is connecting it without losing context, control or trust. MCP offers one possible approach. 

What Model Context Protocol means for legal teams 

Model Context Protocol (MCP) is an open-source standard for connecting AI applications to external systems. It gives AI applications a consistent way to access available data, tools and workflows. For legal teams, that could allow AI agents to retrieve permitted information from systems such as matter management or document management and, where authorized, use available tools to complete an action. 

For legal teams, the practical value may appear in requests such as: 

  • Show me my active matters. 
  • Which deadlines are coming up? 
  • Summarize this matter. 
  • Pull the key dates from this document. 

Instead of opening multiple systems, finding records and assembling context manually, users can start with the work they need to get done. 

MCP does not replace APIs. APIs define how software systems exchange information, while MCP provides a standard way for AI applications to discover and use the data and tools those systems make available. In many implementations, an MCP connection may rely on existing APIs behind the scenes. 

The focus group saw greater value in connected legal intelligence 

Participants spent less time discussing individual MCP use cases than a broader problem: how to connect the legal stack without creating another set of isolated AI experiments. 

One participant described a legal knowledge layer connecting systems such as Onit and iManage with AI tools like Copilot, Harvey and Claude. Another participant had a name for where their organization is headed: a “legal brain” capable of making sense of information across siloed SaaS applications. 

The opportunity is not another AI destination. It is a way for the tools legal teams choose to access trusted context already sitting across their technology ecosystem. 

Better connections expose bad data 

Greater connectivity also makes data quality harder to ignore. 

Years of legal technology configuration can leave behind outdated fields, inconsistent naming and backend information that was never designed to provide context to an AI agent. 

As one participant observed, what gets called an AI “hallucination” may actually be an agent pulling bad underlying data. 

That makes data hygiene an operational requirement rather than a background maintenance task. Fields, descriptions and configurations that may have been easy to ignore become much more important when AI can use them to interpret and act on legal information. 

Data quality is becoming AI infrastructure. 

Governance must travel with the data 

Participants also emphasized that greater connectivity cannot come at the expense of control. 

MCP can standardize a connection, but it does not by itself determine who should have access or which actions should be allowed. Those controls must be enforced by the applications and underlying systems involved. 

Permissions need to follow the user. Someone shouldn’t be able to access information through an AI agent that they couldn’t access in the underlying system. And as agents progress from retrieving information to creating or updating records, organizations need to know who initiated an action, what changed and how it happened.  

The focus group did not advocate connecting every available system as quickly as possible. 

Participants instead emphasized making legal AI more useful without sacrificing the permissions, approvals and audit trails required for governed execution. 

Questions legal teams should ask before adopting MCP 

Legal teams do not need to connect every system at once. Before moving forward with MCP, they should ask: 

  • Which systems contain the trusted matter, contract, spend and document data our AI tools need? 
  • Is that information accurate and consistently structured? 
  • Will existing user permissions apply when information is accessed through an AI application? 
  • Which actions should require review or approval before they are completed? 
  • Will the organization have a reliable record of what the AI accessed, changed or created? 
  • Which connections would benefit from MCP, and which are already well served by existing integrations? 

Answering those questions can help legal teams identify where MCP may add value and where data, access or governance work needs to come first. 

At Onit, we believe the long-term value of legal AI will depend less on any single model or interface and more on the trusted systems, workflows and controls surrounding it. MCP is promising because it may make those connections easier to establish. But connection alone is not the goal. The goal is to give legal teams useful, governed access to context wherever their work happens. 

As AI moves from answering questions to taking action, permissions, approvals and auditability become even more important. Read Beyond Legal AI Assistants: Why Governed Execution Matters to explore what governed execution should look like for legal teams.

Not Every Outside Counsel Decision Needs an RFP. It Still Needs Better Data.

Business team reviewing data and reports together around a laptop during a meeting.

A formal RFP can bring transparency, structure, and competition to outside counsel selection. But not every matter calls for one. 

Some engagements are significant enough to justify a competitive sourcing process. Others need counsel selected quickly. Some involve work a trusted firm already knows well. Others simply do not warrant the time and effort of a formal RFP. 

Just because you’re not running an RFP doesn’t mean you’re skipping the evaluation. 

Whether a legal team is running a formal sourcing process or deciding which firm should handle the next matter, the same fundamental question remains: Why is this the right firm for this work? 

Too often, the answer comes down to familiarity, relationships, or whoever handled something similar last time. Those factors can matter, but legal departments already have another source of evidence available to them: their own matter, spend and vendor performance data. 

The opportunity is to use that information not only during an RFP, but every time outside counsel is selected. 

When does an outside counsel RFP make sense? 

There is no reason to force every matter through the same sourcing process. 

Formal RFPs can be particularly valuable when a legal department is selecting counsel for high-stakes work, evaluating multiple firms for a significant engagement, refreshing a broader panel, or simply getting an update on firms’ evolving capabilities. A structured process gives the team a consistent way to gather information, compare approaches to managing a case, and evaluate firms on more than familiarity or hourly rates. 

But a formal RFP also requires time from the legal department and the participating firms. For many routine or lower-risk matters, the process may be more than the decision requires. 

The goal should not be to run more RFPs. It should be to make better outside counsel decisions. That starts with a clear view of when an outside counsel RFP makes sense and when a lighter-weight process will serve the matter just as well. 

What should you evaluate when you aren’t running an RFP? 

A legal team may decide not to issue an RFP, but you can still evaluate your options. 

Before assigning a matter, consider what your existing data can tell you about the firms you already work with: 

Which firms have handled comparable matters? 

Past matter records can show which firms have experience with similar types of work, jurisdictions and levels of complexity. 

That provides a more useful starting point than simply asking which firm is top of mind. 

What did similar work actually cost? 

Historical spend can help legal teams understand what previous matters cost and how different firms performed against expected budgets. 

The question is not simply which firm has the lowest rate. It is which firm has demonstrated value on comparable work. 

How did the firm staff the work? 

Two firms may approach the same matter very differently. 

Historical staffing data can help show whether previous engagements used the right mix of partners, associates and other resources for the work involved. 

Did the firm stay within budget? 

A proposed budget is useful. A history of how a firm performed against budgets provides another layer of evidence. Does one firm tend to be more accurate in their initial case assessment, and therefore expected costs, while another typically has multiple budget revisions for each engagement? 

Consistent budget performance can help a legal team assess predictability before assigning new work. 

Were there recurring billing or compliance issues? 

Invoice history may reveal patterns that are easy to overlook when matters are considered individually. 

Repeated guideline violations, staffing issues or billing adjustments can provide useful context when deciding whether a firm is the right choice for another engagement. 

What happened after the matter was assigned? 

Assigning outside counsel to a matter is just the beginning, not the end.  Having a governance process based in effective communication to manage and evaluate performance throughout the engagement is a critical metric of success. 

Where legal teams capture relevant outcome and performance information consistently, that data can help inform the next matter, the next panel review and the next RFP. In fact, matter and spend data can reveal how firms actually performed long before a formal RFP ever enters the conversation. 

Relationships still matter. Data makes them more useful. 

Legal work is not a commodity, and outside counsel selection should not become a spreadsheet exercise. 

A general counsel may know that a particular partner understands the business exceptionally well. An in-house attorney may have years of experience working successfully with a specific firm. A legal operations team may know that certain firms collaborate better with internal teams than others. 

Those are meaningful inputs. 

The problem comes when relationship knowledge is the only input or when important experience lives only in the memories of individual team members. 

Structured matter and vendor data gives legal teams a way to complement that judgment with evidence. 

Instead of asking, “Who do we usually use?” the conversation can become: 

  • Who has done this type of work before? 
  • How did they perform? 
  • What did it cost? 
  • Did they meet expectations? 
  • What have we learned from working with them? 
  • Has another firm’s capabilities evolved where they warrant consideration? 
  • Are the lawyers who typically do the work at our preferred firm still there, or have they moved on? 

That creates a more informed decision without requiring a formal sourcing event every time work needs to be assigned. 

Outside counsel selection should be a continuous cycle 

One of the biggest limitations of treating RFPs as standalone events is that the selection process can become disconnected from everything that happens afterward. 

A firm is evaluated. A decision is made. Then the actual matter, invoices, budget performance and vendor relationship move into other workflows. 

The better model is a continuous cycle: 

Select → engage → manage → measure → select again 

Each engagement should create information that improves the next decision. 

Matter history builds a record of experience. Spend data shows what the work costs. Invoice information reveals compliance and billing patterns. Performance information adds context around how the relationship actually worked. 

Over time, legal teams can build a more complete view of their outside counsel relationships instead of starting from scratch every time they need to make a sourcing decision. 

That same information becomes valuable when a formal RFP is warranted. Rather than relying only on what firms say in their proposals, the legal department enters the process with its own history and evidence: structured vendor performance data that reflects what actually happened, built on systems that connect matter, spend and vendor information instead of leaving it scattered across inboxes and spreadsheets. 

Make the process fit the decision 

The answer is not to require an RFP for every outside counsel engagement. 

It is also not to reserve structured, data-informed decision-making only for the handful of matters that receive a formal sourcing process. 

Legal teams need both. 

For significant engagements, a structured RFP can help teams compare firms, evaluate value and create a defensible record of the decision. For matters that do not need an RFP, historical matter, spend and performance information can still provide the evidence needed to make a thoughtful choice. 

The process may change depending on the matter. 

The standard for making an informed decision should not. 

Ready to rethink your outside counsel sourcing process? 

Explore The Modern Outside Counsel RFP Playbook for a practical framework to determine when an RFP earns its keep, evaluate firms on total value and build a more structured approach to outside counsel selection. 

The Questions Legal Leaders Should Be Able to Ask Their Data in Seconds

Computer monitor displaying data dashboards, charts and analytics in a modern office.

Legal leaders should be able to ask direct questions about spend, budgets, matters, vendors and overall operations, then receive a clear answer while there is still time to act. The standard is not speed alone. The answer should also be accurate – current, grounded in the legal department’s system of record, aligned with the user’s permissions and specific enough to support a decision. 

That expectation changes the role of legal data. Instead of becoming a quarterly reporting exercise, data becomes part of day-to-day leadership. The most useful questions fall into three groups: where money is going, where attention is needed and whether the answer can withstand scrutiny. 

What is happening with legal spend right now? 

Leadership questions rarely arrive on the reporting calendar. Your CFO may ask how accruals are tracking against budget before a forecast meeting. Your general counsel may need to know which matters are driving variance before an executive update. Procurement may want to understand which vendors are associated with the most invoice rejections or adjustments before a review. 

Legal leaders should be able to ask: 

  • How is current legal spend tracking against budget? 
  • Which matters and vendors are driving the variance? 
  • How do accrued amounts compare with final invoice spend? 
  • Which invoices are received, on hold, rejected or approved? 
  • Which business units are represented in current matter allocations? 

These are not unusual analytics requests. They are routine management questions. When each one requires a report request, an export and spreadsheet reconciliation, the answer can arrive after the decision window has passed. Manual legal reporting also consumes time that legal operations doesn’t have or could be better spent applying to any number of other items already on their plate. 

Where should Legal investigate before a problem grows? 

Good legal reporting explains what happened. Better access to legal data helps leaders decide what to examine next. A high-level total may be useful, but the next question often carries the real insight. 

Legal leaders should be able to ask: 

  • Which vendors had the most rejected or adjusted invoices this year? 
  • Which open matters have active purchase orders? 
  • Which open matters had no invoicing activity during the period? 
  • Which invoice reviewer rules apply to a particular vendor? 
  • How are budgets, accruals and final spend comparing across open matters? 

The value comes from connecting financial and operational context. Spend without matter information can hide what is driving the number. Accruals without final spend make reconciliation harder. Vendor totals without rejection or adjustment patterns provide an incomplete view of performance. 

The goal is not to replace legal judgment with an automated conclusion. It is to make the relevant governed data easier to explore, so leaders can identify patterns, ask follow-up questions and decide where human review is needed. 

Can the answer stand up to scrutiny? 

A fast answer is useful only when people understand what it means and where it came from. Legal teams work with sensitive information, different access levels and financial definitions that can change the result. “Spend,” for example, may mean billed spend or final spend after adjustments and taxes. A system should ask for clarification when the request is ambiguous instead of guessing. 

Legal leaders should also be able to ask: 

  • Is this answer based on the current system of record? 
  • Does it reflect the permissions of the person asking? 
  • Can the team review the underlying detail? 
  • Can the result be presented clearly in a leadership update? 
  • If the question is unclear or unsupported by available data, will the system say so? 

These questions establish a practical standard for governed, conversational analytics: current data, permission-aware access, structured answers, supporting visuals and a clear path for validation. They also keep professional judgment where it belongs, with the legal team. 

Make faster answers a leadership expectation 

Legal departments already hold information about invoices, matters, vendors, timekeepers, budgets, accruals, allocations and purchase orders. The leadership opportunity is to make that information easier to question without moving it into disconnected tools or rebuilding the same analysis for every meeting. 

Ask Unity turns governed Unity ELM data into clear, actionable answers and provides guidance on product-related questions. Users can explore legal spend, matters, vendors, budgets, accruals and other operational data through natural-language questions, then review structured responses with supporting charts and tables. When a request is unclear, Ask Unity prompts the user to refine it instead of filling gaps with speculative output. 

Explore the Ask Unity datasheet to see how legal teams can move from a plain-language question to a decision-ready answer inside Unity ELM. 

Why Legal Billing Guidelines Aren’t Enough for Matter-Level Invoice Review

Business professional reviewing a document while working on a laptop at a desk.

Legal billing guidelines provide an essential baseline for invoice review, but they cannot account for every negotiated rate, fee arrangement or matter-specific exception. Accurate legal invoice review requires both the general guidelines and the engagement terms that apply to the individual matter. 

The distinction matters because billing guidelines establish the department’s policy, while engagement documents capture the commercial agreement for a specific piece of work. An invoice may comply with one and conflict with the other. Reviewers need both sources to determine what the organization actually agreed to pay. 

Without that complete context, reviewers can miss charges that violate the engagement or dispute charges that were expressly permitted. Either outcome weakens spend control and creates avoidable work for Legal Operations and outside counsel. 

What do legal billing guidelines cover and where do they fall short? 

Outside counsel guidelines establish the department-wide rules for legal billing. They define what is billable, how time should be recorded, which expenses are permitted, and what supporting documentation firms must provide. They give legal teams a repeatable standard and help firms understand expectations before invoices arrive. 

That baseline remains important. Onit’s guide to creating and enforcing legal billing guidelines explains how clear requirements improve consistency, transparency and collaboration with outside counsel. 

But a baseline is not the complete agreement for every matter. An engagement may include a negotiated rate, an approved timekeeper, a fixed fee or alternative fee arrangement, a matter budget, or a travel exception. Those terms can change how an invoice should be evaluated for that specific engagement. 

Consider a simple example. A matter-level engagement letter caps a partner’s hourly rate at $850, while the vendor’s general billing guidelines do not reflect that negotiated rate. If the firm submits time at $950 per hour, a review based only on the general guidelines may miss the variance. Spend Agent can apply the matter-level rate agreement, prioritize it over the general guidelines and identify the line item that exceeds the agreed rate. 

For invoice-review purposes, matter-specific terms should take precedence when the organization has documented and designated them to govern that engagement. Reviewing only the general guidelines leaves the process with incomplete context. 

What risks arise when invoice review lacks matter context? 

Missing engagement terms create risk in both directions. 

First, noncompliant charges may be approved. A rate can exceed the amount negotiated for the matter, an unapproved timekeeper can appear on the invoice or a charge can fall outside the agreed fee structure. If the review process sees only the general guidelines, those issues may appear compliant. 

Second, permitted charges may be disputed incorrectly. Pre-approved travel can be flagged under a general restriction, or a fixed-fee invoice can be evaluated as though it were billed hourly. Legal Operations then has to reverse the dispute, explain the mistake and restart the payment process. That creates rework, delays payment and can strain an otherwise productive outside counsel relationship. 

The challenge is not simply finding more potential violations. It distinguishes a true billing issue from an approved exception. A system that flags every deviation without understanding the engagement can create noise instead of control. 

Onit’s AI Center of Excellence tested large language models and experienced legal invoice reviewers against the same billing-review tasks. The top-performing model achieved 92% invoice-level accuracy and 81% line-item accuracy, compared with 72% and 43% for experienced lawyers. The models also completed reviews 50 to 80 times faster. Importantly, the research found that discretionary human judgment improved reviewer accuracy, reinforcing the value of combining automated analysis with human oversight. Read the Better Bill GPT research. 

These findings establish the potential of AI-assisted review, but accuracy still depends on context. A review process cannot consistently apply a negotiated exception if the agreement containing that exception is unavailable to it. 

The goal is not to challenge every possible charge. It is to make the right decision using the agreement that actually governs the work. 

How does engagement-aware invoice review work? 

Engagement-aware invoice review evaluates an invoice against the department’s general legal billing guidelines and the documented terms associated with the matter and vendor. This gives the review process the context needed to recognize negotiated rates, fee arrangements and approved exceptions while maintaining a consistent department-wide baseline. 

Legal teams can strengthen that process by: 

  • Identifying where matter-specific billing terms are currently stored 
  • Confirming which documents should govern the review when terms conflict 
  • Associating engagement documents with the correct matter and vendor 
  • Establishing a consistent method for recording approved exceptions 
  • Testing review outcomes against representative matters and invoice scenarios 
  • Determining where automated action is appropriate and where a person should make the final decision 

Relevant engagement documents can be added to the matter so Spend Agent can review invoices using both matter-level terms and general vendor billing guidelines. When those sources conflict, the designated matter-level terms receive priority. 

The documents must be uploaded and associated with the applicable matter and vendor. Relevant email correspondence is not automatically pulled from a mailbox; it must be captured in a supported document, such as a PDF, and added to the matter. 

This is an important governance control. The review is based on the terms the organization intentionally provides and associates with the engagement, rather than assumptions drawn from unstructured communications elsewhere. It also makes the source behind a decision easier for reviewers to inspect. 

When Spend Agent identifies a line-item violation, Auto Adjustments can calculate a recommended financial adjustment and present the original amount, the issue, the applicable billing guidance and the recommendation together. Organizations can manage this capability by vendor, reflecting the fact that review requirements and outside counsel relationships are not identical across the panel. 

This additional context does not remove human judgment. Legal teams can use Sentry mode when a reviewer should approve or reject a recommended action and Auto mode when automated handling fits the organization’s operating model. This allows teams to expand review capacity while preserving oversight where judgment, invoice value or vendor relationships require it. 

Key takeaways 

  • General legal billing guidelines establish the default billing standard. 
  • Matter-level engagement documents capture negotiated terms and exceptions that may change how an invoice should be reviewed. 
  • Using both sources helps prevent missed violations and incorrect disputes. 
  • Engagement context strengthens invoice review without eliminating human oversight. 

Legal invoice review works best when every decision reflects the complete agreement, not just the department-wide default. See how AI, business rules and human judgment can work together in engagement-aware invoice review. Watch The Right Invoice: Rethinking Legal Invoice Review to learn more.

Business Rules, AI, or Human Review? A Decision Framework for Legal Invoices 

Two professionals reviewing a document with pens, overlaid with a digital network graphic representing connected data and technology.

The best legal invoice review process does not force every decision through the same layer. Use deterministic e-billing rules for objective checks with no contextual exception, contextual AI when the answer depends on written billing or engagement terms and requires interpretation, and human judgment when a finding is material, ambiguous, or sensitive to an outside counsel relationship. 

That division of labor creates a more practical operating model. Each layer does the work it is best suited to perform, while the legal team remains responsible for governance and the final business decision with greater accuracy and less manual effort than before. 

This framework builds on the discussion in The Right Invoice: Rethinking Legal Invoice Review about assigning business rules, AI, and human judgment to the work each handles best. 

Type of decision Best review layer Examples 
Objective, with no contextual exception Deterministic rule Duplicate invoice number, budget threshold 
Depends on written engagement context / requires interpretation Contextual AI review Appropriate level resource, fee arrangements, daily hour threshold 
Material, ambiguous, or relationship-sensitive Human judgment Review/Act on findings, worked performed in alignment with goal 

1. Use deterministic rules for fixed, objective checks 

Conventional e-billing rules remain an important first layer of legal invoice review. They are well suited to binary checks that should produce the same answer every time and do not change based on the matter, vendor, or surrounding documents. 

A duplicate invoice number is the clearest example. If two invoices carry the same number, the system does not need to interpret a narrative, compare negotiated terms, or weigh an exception. A deterministic rule can identify the condition efficiently and consistently. 

The practical test is simple: Could valid context change the answer? If the answer is no, use a business rule. This keeps stable controls stable and avoids adding complexity where interpretation offers no benefit. 

2. Use contextual AI when written terms can change the answer 

Not every billing decision is binary. General outside counsel guidelines may prohibit a charge, while a matter engagement letter permits it. An approved rate agreement may supersede a standard rate. A negotiated fee arrangement or travel exception may apply only to one engagement. 

These decisions require the reviewer to read multiple sources, understand which terms apply, and resolve conflicts between general and matter-specific guidance. That is where contextual AI review can add value. It can evaluate invoice lines against the relevant written context and surface an explanation for review and is also best positioned to interpret a line item where the coding is not necessarily accurate. 

The source hierarchy matters. Matter-specific terms should take precedence when they conflict with general billing guidelines. Relevant engagement documents must also be captured in a supported file, uploaded to the matter, and associated with the applicable vendor. Contextual AI should not be described as automatically retrieving terms from a user’s mailbox. 

3. Keep people in control of material and sensitive decisions 

Finding a potential variance is not the same as deciding what to do about it. A legal team may need to consider the amount at issue, the clarity of the supporting language, the strategic importance of the matter, and the relationship with outside counsel. 

Human judgment belongs at this decision point. Reviewers can determine whether to accept a recommended adjustment, request more information, make an exception, or address a broader pattern with the firm. This is especially important for high-value invoices, ambiguous guidance, and sensitive vendor relationships. 

Governance should reflect those differences. For example, Spend Agent gives legal teams vendor-level choices for automated action, human confirmation, or no AI review. Those choices are not maturity rankings. They are controls that should align with risk tolerance, billing guidance, and the relationship involved. 

Build a layered legal invoice review model 

The goal is not to choose business rules, AI, or people as a single answer. It is to route each decision to the right review layer. 

Keep deterministic e-billing rules for objective controls. Use contextual AI to interpret applicable written guidance and engagement terms. Reserve human attention for the findings that require materiality, discretion, or relationship judgment. 

Spend Agent is designed to complement established e-billing controls and support this layered approach. Deterministic rules handle fixed controls, contextual AI evaluates complex line-item descriptions against applicable written terms, and people make the decisions that require discretion. Together, those layers give legal teams a more governed and practical approach to invoice review.

Your Best Outside Counsel RFP Data Is Already in Your Matter Management System 

Legal professionals reviewing documents together during a meeting, representing outside counsel evaluation and firm selection.

Before you send an outside counsel RFP, your own matter records can answer six questions a law firm proposal alone cannot reliably validate: which firms have handled comparable matters, what that work actually cost, who stayed within budget, how they staffed it, whether their invoices caused problems and where consistently captured, what outcomes they delivered. Most legal departments never ask. They open the RFP process with a blank questionnaire and let the firms define the terms of the conversation. 

That’s a lot of evidence to leave sitting in a system you already pay for. 

What data should you review before sending an outside counsel RFP? 

Which firms have handled similar matters. Marketing decks describe capability in broad strokes. Your matter records show which firms have run comparable work: same subject matter, same complexity, same jurisdictions. A practice group brochure and a track record are different things. 

What comparable matters cost. Historical spend gives you a baseline you can defend. Rather than asking firms what they’ll charge in the abstract, you can weigh their pricing against what this kind of work has actually cost you. Look at the median and the range, broken out by firm. Use that history as a directional benchmark, particularly when comparing similar work across firms. 

Who stayed within budget. Budget-to-actual variance is one of the most telling metrics in outside counsel management. Firms that land inside their own estimates are firms you can plan around. Firms that don’t become a source of quarterly surprises. Where your department tracks estimates and actuals consistently, use that record to inform the questions and evaluation criteria in the next RFP. 

How they staffed it. Partner-to-associate ratios, team size, and how leverage shifted over the life of the matter. Cross-reference that against outcomes and cycle time and patterns start to surface. Some work rewards a lean senior team. Some doesn’t. At a minimum, ask firms to provide a clear proposed team and staffing model; where historical staffing data is available, use it as additional context. 

Whose invoices caused problems. Rejected line items, out-of-guideline charges, block billing, timekeepers who never got approved. A firm’s invoice discipline  can be a useful signal of the administrative burden it may create during an engagement, and guideline compliance is rarely evenly distributed across a panel. 

What outcomes they produced. Wins, settlements, closings, approvals, tied to specific firms and named lead attorneys. This is the hardest of the six to capture consistently. It’s also the closest thing to a scorecard you’ll get. Where your department captures outcomes in structured fields, tie them to the responsible firm and lead attorneys. 

Answer those six before an RFP goes out and the process changes shape. You stop asking firms to describe themselves and start asking them to explain the distance between what they claim and what your records show. 

Why do legal departments skip their own RFP data? 

For many departments, it lives in three or four places. Matter details in one system, invoices in another, budgets in a spreadsheet on someone’s desktop, outcomes in an email thread. Pulling a coherent picture together for one firm takes hours. Doing it across a panel takes weeks, and the RFP deadline rarely waits. 

There’s a cultural reason too. Matter management has historically been treated as record-keeping. The idea that the same records should inform the next sourcing decision is fairly new, and it doesn’t have an obvious owner in most departments. 

How do you bring matter data into the RFP process? 

Standardize what you capture. Practice area, matter type, jurisdiction, staffing, budget, actual spend, cycle time, outcome. If those fields aren’t populated consistently, nothing downstream will fix it. 

Connect matter data to spend data. Firms should be measurable across the full engagement, not only on what they billed. This is the same connected-data problem that shows up in vendor management, with the same root cause. 

Let what you find set your evaluation criteria. If invoice discipline is a chronic problem across your panel, weigh it explicitly in scoring. If budget predictability is the sore spot, ask firms to defend their record on it. 

Feed the results back in. Capture the same fields on the new panel. Year three of a sourcing program should look nothing like year one. 

What changes in the RFP itself 

Better inputs are only half of it. The evaluation has to be structured well enough to use them: standardized questionnaires so pricing arrives in a format you can line up side by side, blind review so evaluators score independently, weighted dimensions that reflect what your history says matters, and conflict questions asked the same way every time. 

Structured bidding belongs in the conversation too. For defined, price-comparable work where several qualified firms are competing; a reverse auction lets them adjust pricing against anonymized rankings and produces a record of how pricing moved. It is not the right approach for every engagement; expertise, capacity, conflicts, urgency, and relationship fit should remain part of the evaluation.  That record is often more useful in the CFO conversation than the final number, because it shows the competitive pressure was real. 

Where this leaves you 

Relationships should stay in the decision. Lawyers know things about firms that no dataset captures. But relationships alone won’t hold up when finance asks why a particular firm won a seven-figure engagement, and they won’t tell you which firm in an otherwise fine-looking panel is drifting. 

Unity RFP is where that structure lives. It’s a module inside the Unity platform, working alongside matter management and e-billing rather than as a separate sourcing tool, with reusable questionnaire templates, side-by-side proposal comparison, weighted scoring with blind review, conflict-of-interest disclosures and documented responses, reverse auctions, standardized conflict-disclosure questions and documented responses, and a complete record of how the decision was made. For matter-specific RFPs, confirm the applicable rate-card, rate-transfer, and enforcement workflow before stating that negotiated rates automatically flow into Matter Management. 

See how Unity RFP structures outside counsel selection 

How Government Purchase Cards Are Simplifying Legal Technology Procurement

Close-up of a computer screen displaying a Purchase button with a cursor selecting it, representing digital procurement, online purchasing, and government technology acquisition.

Government agencies are under increasing pressure to modernize operations while working within tight budgets, limited staff, and evolving procurement requirements. One area seeing significant growth is the use of Government Purchase Cards (GPCs) and other micro-purchase programs, which allow agencies to procure low-dollar solutions quickly and efficiently. 

For legal departments, procurement teams, and administrative offices, this shift presents an opportunity to adopt technology that improves efficiency without requiring lengthy procurement cycles. 

Why Government Purchase Cards Matter 

The Government Purchase Card program has become the preferred purchasing method for many low-dollar acquisitions across federal agencies. Similar purchasing programs also exist at the state and local level, helping organizations reduce administrative burden while accelerating access to essential technology. 

For legal and procurement professionals, this means the focus is no longer just on finding the right solution. It’s also about finding solutions that are easy to purchase, deploy quickly, and deliver measurable value. 

Technology that aligns with these procurement models gives agencies the ability to improve operations while remaining compliant with purchasing regulations. 

The Challenges Facing Government Legal Teams 

Public sector legal departments manage a growing volume of contracts, investigations, litigation, public records requests, and regulatory matters. Many agencies continue to rely on disconnected spreadsheets, shared drives, email, or paper-based processes that create unnecessary risk. 

Common challenges include: 

  • Difficulty locating contracts and case files 
  • Missed renewal dates and compliance deadlines 
  • Limited visibility into workloads and case status 
  • Manual approval processes that slow operations 
  • Increasing expectations for transparency and accountability 

At the same time, staffing levels often remain flat while workloads continue to increase. 

Modern Legal Operations Start with Better Information Management 

Cloud-based legal technology helps agencies centralize information while automating many of the manual tasks that consume valuable staff time. 

For contract management, a centralized repository gives authorized users immediate access to current agreements while automated reminders help prevent missed renewals. AI-assisted metadata extraction and integrated eSignature capabilities further reduce administrative work and eliminate the need for multiple standalone tools. 

Case and matter management platforms provide similar benefits by organizing documents, emails, investigations, and legal matters within a single secure system. Workflow automation helps ensure deadlines are met, while dashboards provide leadership with real-time visibility into caseloads and operational performance. 

Together, these capabilities allow agencies to spend less time managing paperwork and more time serving their constituents. 

Procurement Shouldn’t Be the Barrier to Modernization 

One of the biggest obstacles to adopting new technology has traditionally been the procurement process itself. 

Subscription-based software fits well within many agency purchasing models by providing predictable costs, rapid deployment, and reduced infrastructure requirements. 

For agencies using purchase cards, this can significantly shorten the timeline between identifying a need and implementing a solution. 

Security and Compliance Remain Essential 

Government organizations cannot sacrifice security for speed. 

When evaluating legal technology, agencies should look for solutions that provide: 

  • Secure cloud hosting 
  • Role-based access controls 
  • Audit trails 
  • Encryption for data at rest and in transit 
  • U.S.-based hosting 
  • Compliance with standards such as SOC 2 and other applicable government security requirements 

These capabilities help agencies protect sensitive legal information while supporting transparency and regulatory compliance. 

Supporting Government Legal Teams with Purpose-Built Technology 

Onit offers solutions designed to help government legal teams modernize operations while simplifying procurement. 

ContractWorks provides cloud-based contract lifecycle management with unlimited users, AI-assisted contract organization, integrated eSignatures, automated reminders, and reporting that helps agencies improve contract visibility and reduce administrative effort. 

ReadySign delivers secure, cloud-based electronic signature capabilities that help government legal teams execute documents faster without heavy IT involvement. With unlimited users, reusable templates, automated reminders, and a complete audit trail for every signature, ReadySign streamlines approvals while maintaining the security and compliance agencies require. Its straightforward, subscription-based pricing and rapid setup—with no lengthy implementation—make it easy to purchase and deploy within purchase-card and micro-purchase programs. 

Together, these solutions provide a modern foundation for legal operations while supporting the procurement flexibility that today’s government buyers increasingly expect. 

Looking Ahead 

Government purchasing continues to evolve toward faster, more efficient acquisition methods. As agencies look to modernize legal operations, technology that combines secure cloud delivery, rapid implementation, and procurement-friendly purchasing options will become increasingly valuable. 

Organizations that align both their technology strategy and procurement approach will be better positioned to improve service delivery, increase operational transparency, and make the most of limited public resources. 

To learn how Onit’s government solutions can help your agency streamline legal operations and simplify procurement, contact our team or explore our public sector solutions.