When we tell firms that August works from their precedent bank, the first question is almost always about what that process looks like. Most transactional practices have years of closed deal files in iManage, NetDocuments, or another document management system. The materials are there. What they typically lack is the structure that makes those materials queryable in the way August needs to query them.
Setting up a precedent bank for August is not a one-time data migration project. It is the process of making your existing deal library useful for future drafting, which has implications for how files get organized during active matters as well as what happens at closing. This piece describes what that process looks like in practice.
What the Bank Needs to Contain
August queries the precedent bank primarily by transaction type and deal characteristics to identify comparable prior matters. For that query to produce useful results, the bank needs to contain the executed final agreements from closed matters, tagged with enough metadata to allow meaningful comparison. The metadata that matters most is transaction type, party positions, industry sector, governing law, deal date, and where available, deal size range.
What is not required, at least at the outset, is a comprehensive record of every draft and every negotiated change. The executed agreement reflects the positions both parties accepted at closing. That is the ground truth about what the firm successfully papered. Negotiation history is useful for understanding how the firm got there, and we do ingest it where it is available in structured form, but the executed agreement is the core input and a reasonable starting point for most practices.
The bank also does not need to be curated in the form-book sense. You do not need to pre-select the "best" agreements or identify which ones reflect current standard positions before ingesting. August handles the comparison at query time, which means you can ingest broadly and let the system surface what is relevant when it is needed.
The Metadata Challenge
In most document management systems, the metadata that matters for precedent retrieval is either missing, inconsistent, or structured for billing purposes rather than for knowledge retrieval. Matter type fields often reflect billing codes that have limited alignment with transaction type categories. Industry codes are frequently absent or assigned at the client level rather than the matter level, which means all matters for a diversified client get the same industry tag regardless of what the deal actually involved.
There are two approaches to this problem. The first is retrospective metadata enrichment: working through the existing deal library to apply a consistent classification scheme to past matters. This is valuable and produces a cleaner bank, but it is labor-intensive and often stalls when it runs up against the volume of historical files a mid-size firm has accumulated.
The second approach, which is what we recommend for most firms starting out, is a combination of automated classification at ingest and prospective metadata standards for new matters. At ingest, August extracts deal characteristics from the documents themselves, which allows a degree of automatic classification that reduces the need for manual tagging. For new matters, the firm establishes a standard of tagging at opening and at closing that ensures future deals enter the bank with the metadata the system needs.
The realistic outcome of this approach is a bank where recent deals are well-structured and highly queryable, and where older deals have varying levels of metadata quality. The system's comparison accuracy is therefore better for transaction types with significant recent deal volume than for types represented mainly by older materials. This is not a limitation specific to August. It reflects the state of most firms' historical records, and it improves over time as more well-structured deals accumulate in the bank.
The iManage and NetDocuments Integration
August connects directly to iManage and NetDocuments, which are the document management systems used by most of the mid-size transactional practices we work with. The connection is read-only from the DMS perspective. August ingests documents and metadata from the DMS but does not write back to it. The firm's document management system remains the authoritative record.
The ingest process identifies closed matter folders, ingests the executed agreement documents within them, extracts document-level metadata, and supplements it with system-level metadata from the DMS where available. For firms with iManage Workspace, the workspaces structure often maps reasonably well onto matters, which simplifies the ingest boundary. For NetDocuments, the folder taxonomy varies more across firms, and the ingest configuration needs to account for that variation.
We do not ingest every document in a matter folder. The ingest configuration targets executed agreement documents specifically, which in practice means filtering on naming conventions, document type flags, and folder structure within matters. Drafts, correspondence, research memos, and other matter-related documents are excluded from the precedent bank by default, though they can be ingested separately into a different corpus if the firm wants to support other use cases.
What to Prioritize When Starting
The most common mistake firms make when setting up a precedent bank is trying to ingest everything at once. A firm with fifteen years of matter history across multiple practice groups is looking at a large volume of documents with highly variable metadata quality. Attempting to process all of it before using the system delays the point at which the bank becomes useful, and the quality of the comparison results during that period is lower than what the system produces once the bank is well-populated with recent, well-structured deals.
The better approach is to prioritize by transaction type and deal volume. Identify the two or three deal types that represent the highest share of the firm's transactional volume and where the firm has the most interest in improving the starting-point quality of drafts. Start the ingest there, with a focus on deals from the past three to five years where metadata quality tends to be better. Get the system in use on live matters in those practice areas. Expand the bank incrementally as the firm builds confidence in the results.
This approach also allows the firm to refine the metadata standards and classification scheme while working with a manageable volume of documents, rather than trying to establish that scheme in the abstract before the bank is in use.
Ongoing Bank Maintenance
A precedent bank is not a static artifact. It is most useful when it reflects current practice, which means new deals need to enter the bank promptly after closing and deals that reflect positions the firm has moved away from need to be appropriately weighted or excluded.
The practical mechanism for keeping the bank current is a closing checklist item: at deal close, the final executed agreement is tagged with the relevant metadata and flagged for ingest into August. Most firms that implement this find it adds minimal overhead at close, since the document is already being filed at that point. The metadata tagging is the new step, and it is most efficient when it is done at matter opening with a standard set of fields, then confirmed or supplemented at close.
For deals where the firm's position has evolved, the date filter in August allows the review to be weighted toward recent practice. If the firm shifted its standard liability cap structure two years ago, the attorney can configure the comparable deal query to prioritize deals from the past two years, which surfaces the current standard rather than the prior one. This is a working judgment the attorney makes rather than an automatic rule, which preserves flexibility for matters where historical practice is more relevant.
The goal is a bank that a transactional associate can rely on as a current, queryable record of how the firm papers deals. Building toward that state takes some initial effort, but the compound value of each new deal that enters the bank in good order makes the investment increasingly worthwhile over time.