Due diligence
Kira vs Luminance: Which AI Review Tool Fits Your Due Diligence Work
Written by Elliot Crane If your deal team is choosing between Kira and Luminance, you are choosing between two tools that grew up doing the same job: reading a data room full of contracts faster than a room of associates can. Both are mature. Both are sold by quote. The real differences sit in how each one learns, what each one has become since its due diligence days, and what kind of firm each fits. This page sets those differences out so you can decide with your own matters in front of you, not a vendor deck. A short answer first. Pick Kira if your work is repeat M&A, finance and real estate diligence where you want a large library of lawyer-built extraction fields on day one and you already live in the Litera ecosystem. Pick Luminance if you want a broader contract platform that also handles negotiation and repository work for an in-house team, and you value a system that groups unfamiliar documents for you before anyone has told it what to look for. If your volume is small and mostly Indian-law contracts, neither may be the right size, which is covered near the end. ## Where each product came from, and who owns it now Kira started as Kira Systems in Toronto. Litera announced it would acquire Kira Systems in August 2021, and at the same time a separate company, Zuva, was spun off to offer Kira's underlying contract technology to corporate users. So when a buyer says "Kira" today, they mean a Litera product that is aimed squarely at law firms and transactional teams. It sits next to other Litera tools used on deals, which matters if your firm already pays for them. Luminance is a UK company that began in M&A due diligence. An independent product review by Artificial Lawyer described it as starting out as a due diligence tool "mostly focused on law firms" and then widening into property and compliance work. On its own site Luminance now presents itself as a contract intelligence platform with separate areas for a repository, negotiation and workflows. In practice that means Luminance today sells as much to corporate legal teams as to law firms. That history explains most of the comparison. Kira has stayed close to the review grid. Luminance has spread across the contract lifecycle. ## How the two engines learn your contracts This is the question that decides day-to-day experience, and the two vendors answer it differently. Kira: a large library, plus fields you train. Litera's Kira product page says it ships with more than 1,400 pre-built smart fields trained by lawyers, and that teams can train their own fields on top. In a diligence project you pick the fields that matter (change of control, assignment, exclusivity, governing law, termination for convenience and so on), run the documents, and review the extracted text in a grid. When your firm has a clause type the library does not cover, someone trains a custom field with examples. The page also describes generative AI layered on the predictive extraction, used to turn results into client-ready output. Luminance: cluster first, then learn as you tag. The Artificial Lawyer review records Luminance's description of its method as unsupervised machine learning for pattern recognition, with supervised learning as lawyers tag clauses during live review. The visible effect is that Luminance groups documents by type, clause and language before anyone configures it, which helps when a data room is messy and you do not yet know what is in it. The same review noted that firms which invested time in training the system got more from it than firms that expected instant results. Put plainly: Kira rewards teams that know what they are looking for and want consistent, named fields across many deals. Luminance rewards teams that open a data room cold and want the tool to show them the shape of it first. ## A side-by-side for a deal team | Decision point | Kira (Litera) | Luminance | |---|---|---| | Core strength | Field-based extraction in a review grid for M&A, real estate, banking and finance | Grouping and anomaly spotting across unfamiliar documents, then clause review | | Out-of-box coverage | Vendor states a library of lawyer-trained smart fields, extendable with your own | Vendor-described pattern learning plus tagging during review; language starter packs reported in independent review | | Beyond diligence | Mostly stays in the transactional review lane, alongside other Litera deal tools | Repository, negotiation in Microsoft Word, and workflow modules for in-house teams | | Typical buyer | Law firm transactional groups and deal teams | Law firms and corporate legal departments | | Data room fit | Check which data rooms your team uses against Litera's current connectors | Older independent review reported it did not connect to every data room; check yours | | Pricing | Quote-based; no public list price on the product page | Quote-based; no public list price | Treat the table as a starting map. Every row should be confirmed in a live trial on your own documents, because both products change often. ## Run the same mini data room through both A vendor demo uses documents the vendor has seen before. A fair test uses yours. Here is a test that takes about a week of elapsed time and gives you a defensible decision. 1. Build a closed set of 40 to 60 documents from a past matter where the answers are already known. Mix clean NDAs and supply agreements with ugly ones: scanned PDFs, amendments that override earlier terms, a side letter, two contracts in a second language if your practice sees them. 2. Write the answer key before either trial starts. For each document, record the change of control trigger, the assignment restriction, the term and renewal mechanics, the governing law and any exclusivity. This is your ground truth. 3. Give both vendors the same set and the same two days. Ask each to run its standard fields or models with no hand-tuning on your documents, then a second pass after one hour of tuning by your own associate. 4. Score misses, not hits. A tool that finds 95 easy clauses and misses the one buried in an amendment has failed the test that matters. Count the misses on amendments, side letters and scanned pages separately. 5. Time the human part. Record how long an associate needs to check the output and produce the diligence table your partners actually send to clients. 6. Export and inspect. Look at what leaves the system: a spreadsheet, a Word report, a memo draft. If it needs a day of reformatting, that day belongs in your cost. If the two tools score close on misses, the decision usually comes down to the human-time number in step 5 and to what else the firm will use the platform for. ## What each costs you beyond the quote Neither vendor publishes a list price, so you will negotiate. What you can control is the cost around the licence. - Training time. Kira's custom fields and Luminance's tagging both need someone senior enough to know the clause and junior enough to have the hours. Budget for that person.
- Matter volume. Both make most sense when diligence work is steady. A firm that runs four deals a year pays for idle capacity.
- Ecosystem pull. If your firm already uses other Litera products, Kira slots in with less change. If the in-house client wants one platform for review, negotiation and storage, Luminance's wider scope is the draw.
- Security review. Both vendors serve large firms and publish security material; ask each for its current attestations and where your documents are processed. Do not rely on a sales summary. ## Six questions for the two sales calls Ask both vendors the same questions, in writing, and keep the answers in the evaluation file. 1. Which of our data rooms and document stores do you connect to today, and which require a manual download and upload?
- If our associate trains a custom field or tags clauses on one matter, who owns that learning, and does it carry over to the next client's matter?
- Are our documents used to improve models that other customers use? Show us the contract wording, not a slide.
- What happens to the uploaded documents when the matter closes: how long are they kept, and can we delete them on demand?
- How are scanned and low-quality PDFs handled, and how will a reviewer know when text recognition was poor?
- What does a typical first-year price look like for a firm of our size, and what changes at renewal? The answer to the second question often surprises firms. Training that stays inside one matter has to be repeated; training that carries across matters may raise client confidentiality questions. Neither is wrong, but you should know which one you are buying. ## When neither is the right size: a smaller contract workspace Kira and Luminance are built for scale: thousands of documents, repeated across deals. A small firm or in-house team that reviews a few dozen agreements per transaction, mostly under Indian law, often finds both are more machine than the work needs, and the quote reflects that. That is the gap Adira aims at. It is a contract workspace built on Anthropic's Claude, with a Document Intelligence Dashboard that marks each position as favourable, neutral or unfavourable to your side and lists red flags, key dates, obligations, missing clauses and Indian compliance points such as stamp duty, FEMA and Companies Act items. On Adira's own measurement across the 18 law firms that use it, contract review and due diligence work that used to take 10 to 12 days came down to under 2 hours, with only 2 to 3 rechecks needed on its output. That is Adira's figure from its own deployments, not an independent study, and it covers Adira's users, not a head-to-head against Kira or Luminance. The honest limits matter more here than the headline. Adira does not offer a pre-built library of thousands of trained extraction fields, does not advertise data room connectors, and is not designed to bulk-review a 5,000-document data room in a cross-border deal. If that is your work, Adira is not the right fit; look elsewhere, and run the Kira and Luminance test above. Adira's pricing is public (Practice at USD 89 per seat per month billed annually, minimum 3 seats; Firm at USD 179 per seat per month billed annually, minimum 5 seats), it does not train on your contract corpus, and there is a 7-day free trial, so a small team can run the same mini data room test on it in a week without a sales cycle. For other options in the same space, see our notes on Kira alternatives for US teams and on Luminance alternatives. If you are still building the diligence request list itself, start with the business acquisition due diligence checklist. ## A decision you can defend to the partners Write the decision as three sentences in the matter-tech file: the volume you expect over the next 12 months, the misses each tool made on your closed set, and the associate hours each needed to produce the client deliverable. A partner can argue with a preference. It is much harder to argue with a scored test on the firm's own documents. If Kira wins, it will usually be on consistency of named fields across repeat deals. If Luminance wins, it will usually be on messy data rooms and on a client who wants the same platform after closing.
Frequently asked questions
- Is Kira still a separate company?
- No. Litera announced its acquisition of Kira Systems in August 2021, and Kira is now sold as a Litera product. A separate company, Zuva, was spun off at the time to offer Kira's underlying technology to corporate users.
- Which is better for a messy data room I have never seen before?
- Luminance's approach of grouping documents by type, clause and language before configuration tends to help when you do not yet know what a data room holds. Kira tends to suit teams that already know which named fields they want to extract. Test both on your own closed set of documents.
- Do Kira or Luminance publish their prices?
- Neither shows a public list price on its product pages; both are sold by quote. Budget separately for the associate time needed to train custom fields or tag clauses, which is a real cost in both.
- How many documents should a fair trial use?
- Forty to sixty documents from a past matter where you already know the answers is enough to see differences. Include amendments, side letters and scanned pages, because that is where misses happen.
- When is a smaller tool a better choice than either?
- When your transactions involve a few dozen contracts rather than thousands, and mostly Indian-law agreements, a smaller contract workspace with public pricing can be a better match. For large cross-border data rooms, Kira or Luminance remain the more suitable tools.
Sources
See how Adira drafts in your voice and reads contracts from your side.
Explore the showroomWorking through a contract like this? Weave is Adira’s free tool to read, mark up, and connect any contract in your browser — no account needed.
Try Weave — free