OCR for In House Counsel: Contracts and Incoming Paper

OCR for in house counsel does two jobs: it makes the company's signed contract scans searchable, and it turns subpoenas, regulator letters and other incoming paper into text the day they arrive. A legal department should have its own tool, or its own separated space in the company's tool, whenever privileged documents are involved.

An in-house lawyer has one client, and that client's employees scan things constantly without thinking about who can open the result. The sections below follow the documents a company legal team really handles and the decisions that come with them.

You are in-house counsel for one company, so the contracts, subpoenas, and regulator letters all belong to that client. Put searchable copies into contract lifecycle management and the matter file, and keep a legal hold in mind when someone asks to delete a scan. Your duty of confidentiality covers staff who upload, and your outside counsel guidelines should say which firms may receive the text you produce.

What does an in-house legal team receive on paper or as scans?

An in-house legal team receives scans from three directions: the company's own history of signed agreements, paper served on or mailed to the company, and documents business colleagues forward for advice. Most were created by someone outside legal, on whatever copier was nearest, with no thought for search.

What makes the in-house pile different

A firm's scans mostly belong to a matter that opens and closes. A company's scans belong to a relationship that may run for a decade. The master services agreement signed in 2014 still governs, and someone will ask what its limitation of liability says next week and again in three years. The value of making it searchable keeps paying out.

The other difference is volume by trickle. The scans arrive a few at a time, from people who expect an answer by lunch, and not as one large production.

Does a legal department need its own OCR tool or the one the company already has?

A legal department needs its own OCR tool, or a walled-off part of the company's, when the output of the shared one lands where non-lawyers can read it. If the company tool can keep legal's documents visible to legal alone, and it handles your documents well, using it is reasonable.

A company may already own some text recognition. The office copier may add a text layer. A PDF editor on the company's laptops may have a recognition command. The document management or content system may index scans by itself. Start by finding out what exists.

Four questions that decide it

1. Where does the output go? A copier that scans to a shared network folder, or to an inbox the whole floor can open, is the wrong path for an internal investigation memo, however good its recognition is.
2. Who administers it? Company IT administrators can often open anything on systems they run. That may be acceptable under your company's policies or it may not. Find out before you rely on it.
3. Does it give you the output you need? A text layer for search is one thing. Getting a scanned fee schedule into a spreadsheet, or a counterparty's paper terms into Word for markup, is another. Test both.
4. Who pays and who renews? A tool on IT's budget follows IT's priorities. A tool on legal's budget is yours to keep or drop.

A simple way to choose

If you look at Schema Law for this, the plans follow page volume, not department size. Every plan is one person per account: Solo allows 1,000 pages a month, Team allows 10,000, and Enterprise allows 100,000 and adds a security questionnaire and contract review. Three lawyers who each convert documents need three accounts. The pricing page lists each one.

Whatever you lean toward, the company's procurement and security teams will have a view. The companion post on OCR for legal operations walks through the buying process, the pilot and the vendor review, so this post does not repeat them.

Turning a drawer of signed contract scans into a searchable agreement archive

Turning a drawer of signed contract scans into a searchable agreement archive means converting each executed agreement to a searchable PDF, naming it by counterparty and date, and keeping it with its amendments. The goal is modest and worth a great deal: any lawyer can find the signed version and search inside it.

A procedure that survives staff turnover

1. Gather the executed versions. The signed PDF is the one that counts, not the last Word draft. Check the shared drive, the contract owner's email and the paper drawer.
2. Convert and keep the image. A searchable PDF keeps the signatures, initials and handwritten changes visible while adding text behind them.
3. Name by a fixed pattern. For example Brightwater Logistics - MSA - 2019-03-14 - executed.pdf. Counterparty first, because that is how people ask.
4. Attach the family. Amendments, statements of work and renewal letters go in the same folder as the parent agreement.
5. Spot-check the parts people search for. Party names, effective date, term, and any dollar figure.

A small example

Say the company acquired Brightwater Logistics, a fictional carrier, and with it 310 customer agreements as flat scans. A customer disputes a fuel surcharge. Before conversion, someone opens agreements one at a time looking for surcharge language. After conversion, a search for "surcharge" across the folder shows which agreements mention it and on what page, and the lawyer reads those pages on the image.

What search will miss

Handwritten changes initialed at signing are the classic trap. A struck-through payment term with "Net 60" written in the margin may not be recognized at all, yet it may be the term the parties agreed. When an agreement matters, page through the image for ink. The post on handwritten changes to a contract explains what conversion does with them.

Poorly recognized pages hide from search too. A search that returns nothing means "not found in the text layer", which is weaker than "not in the contract".

Going further than search, by pulling renewal dates, notice periods and governing law into a table, is a separate project with its own checks. The post on contract data extraction covers it.

Subpoenas, regulator letters and other incoming paper at a company

Subpoenas, regulator letters and other incoming paper at a company should be converted the day they arrive, because each carries a deadline and a list of demands that someone must paste into a tracker, a hold notice or an email to outside counsel. Retyping them is slow and invites a wrong date.

The stakes of a slow start are real. Under Rule 45 of the Federal Rules of Civil Procedure, a person commanded to produce documents who wants to object must serve the written objection before the earlier of the time specified for compliance or 14 days after the subpoena is served. State subpoenas and agency demands run on their own clocks, so read the document in front of you and check the rules that govern it.

An intake routine for served paper

1. Scan the whole thing as received, including the envelope or the process server's slip if there is one. Record who received it, where and when.
2. Convert it. Export a searchable PDF for the file and a Word or plain text copy for working.
3. Pull five facts onto the first line of your tracker: who issued it, the matter or case number, the date served, the response date, and the contact for questions. Read each against the image.
4. Copy the requests verbatim into your working document, numbered as the issuer numbered them. You will reuse that text in the hold notice, the collection plan and the response.
5. Route it. Send the scan and the text to the lawyer who owns it, and to outside counsel if the matter goes out.

Documents that do not look urgent

Wage garnishments, licensing renewals, a state attorney general's consumer complaint referral and tax notices often reach legal late because the person who opened the envelope did not see them as legal documents. Ask the mailroom, reception and the registered agent to send anything from a court or government body to one legal inbox. A converted copy can then be searched by employee name, account number or agency reference when the follow-up letter arrives.

Fax and photo quality

Agencies and courts still fax, and store managers photograph letters on a counter. Recognition is weaker on both. Read dates and reference numbers against the image every time, and ask for a clean copy when the response date itself is hard to read.

How should in-house counsel keep privileged scans apart from business users?

In-house counsel should keep privileged scans apart from business users by controlling three places: where the scan is made, where the converted output is stored, and who holds accounts on the tool. A privileged document in a folder the sales team can browse is a problem no software setting fixes afterward.

In-house lawyers wear two hats, and documents do too. A scanned board presentation is business. Your notes on it for the general counsel are legal advice. Whether a given document is privileged is a legal question with its own post, on attorney-client privilege and AI tools. The handling question is simpler: treat anything that might be privileged as if it is, until a lawyer decides.

Where leaks happen in a company

• The shared copier. Scan-to-folder often writes to a location the department, or the building, can open. Scan-to-email may keep a copy in a shared sent box.
• The "helpful" forward. A business colleague scans a demand letter, adds commentary, and copies eight people before legal sees it.
• The shared drive. Converted files saved next to the business's working folders inherit the business's permissions.
• The shared login. One account used by legal and contracts administration means everyone sees everything.

Practical separation

Give legal its own accounts on whatever tool you use, held by named people. Do not share a login with a business team, even one you trust.

Store converted output in legal's own repository, under legal's permissions. Keep a separate area for investigations and personnel matters that even most of legal cannot open.

Ask the vendor how its tool divides documents and who inside your company can see what. Schema Law's published line is "Documents stay inside your own account", and an account is one person: nobody else's account can see what you uploaded. That keeps an investigation file away from colleagues. It also means nothing is shared for you, so the rest of legal sees a converted file only after you export it and save it where they can open it. Read the security commitments as written.

Delete what you no longer need in the tool once the converted file is safely in your own system. Fewer copies means fewer places to control.

When a business user needs the text

Sometimes the business needs the converted contract and none of your analysis. Export a clean copy and send that. Do not give a business user an account in legal's space so they can fetch it themselves.

Sending converted documents to outside counsel in a usable form

Sending converted documents to outside counsel in a usable form means sending the original scan together with the converted file, named so the pair is obvious, with a note saying what you checked. A firm that receives flat scans has to make them searchable before it can work with them, and that handling may reach your invoice.

What to send

• The scan exactly as the company received it. This is the evidence. Do not replace it with the converted version.
• A searchable PDF of the same pages. Outside counsel can load it into their own system and search at once.
• Tables as spreadsheets, when the document is a ledger, a price schedule or a payment history. Say which page each sheet came from.
• A short cover note: what the documents are, where they came from inside the company, how they were converted, and what you verified.

A note can be this short:

Re:        Subpoena served on Halden Foods, Inc. on 2026-10-05
Attached:  1. Subpoena as received (scan, 11 pp.)
           2. Same, searchable PDF
           3. Requests 1-23 as text (Word)
Checked:   response date, case number, all 23 requests read against the scan
Not clear: handwritten note on p. 2, appears to be from the process server

Ask what they want before a big transfer

For a handful of documents, a searchable PDF and a note is enough. For thousands of pages going into a litigation review platform, ask the firm what form it wants first. Where a subpoena does not specify a form, Rule 45 speaks of producing electronically stored information in a form in which it is ordinarily maintained or in a reasonably usable form, and what the other side receives is your outside counsel's call. Your job is to hand counsel material that does not need rework.

Keep what you sent

File a copy of the exact set you transmitted, with the date. When the firm asks six months later whether it received the 2019 amendment, you can answer from your own records.