The failure that sank the first lawyers to use AI was the invented citation. We build systems where every answer carries the document and clause it came from, and where a citation that does not resolve against a real document is never shown.
6–8 weeks to production · 3 years sustaining document retrieval in production · 48 hours to a proposal with fixed scope and price
Built on
Ninety seconds: who we are, how we work and what you get at the end.
Contracts, price lists, email, the CRM: what you already have becomes sourced answers and agents that do the work. Pick a case and follow the path.
Example: someone needs to know which contracts expire soon. The system finds them and leaves the reminders ready.
Connected by API or uploaded. Nothing leaves your infrastructure when the case demands it.
310 PDF → 4,120 clauses
Searches by meaning and by exact word, and filters by the tag before choosing.
answers with source · 91 of 100
Swipe to follow the flow →
01
"Which contracts have an exclusivity clause?", "which ones auto-renew?". It is the most frequent query in a legal team with several hundred live contracts, and today it gets answered by asking whoever has been there longest.
02
Comparing an incoming contract against the approved template and flagging which clauses changed, which are missing and which were added. It does not decide whether the change is acceptable: it marks where to look.
03
Expiry dates, notice periods, payment milestones, reporting duties. Extracted into a table with the source clause beside each row, and the alerts come off that.
04
Across a batch of documents, classifying and flagging what needs human reading. It reduces the volume that reaches the lawyer; it does not replace their reading.
01
The failure mode
The case that went around the world — lawyers sanctioned for filing non-existent case law — was not a faulty model: it was a system without verification. A language model produces a real reference and a plausible one with equal fluency, because the format of a citation is an easy pattern to imitate. Asking it in the prompt not to invent reduces the frequency and does not eliminate the problem, and in legal work a low rate is still unacceptable.
You close it in the architecture. Every passage in the index carries a stable identifier — document, version, clause or article — and the answer has to carry it through. After generation, each citation is checked against a real passage; one that does not resolve invalidates the answer, which is never shown. The system escalates instead of gambling. It is a cheap check and it is the difference between a tool a lawyer can use and one they cannot.
The second decision is not letting the model write about what it did not retrieve. If the question has no support in the archive, the correct answer is to say it is not there, not to produce a reasonable-sounding paragraph. That gets measured: the evaluation set includes questions whose correct answer is "this does not appear in the indexed documents".
The evaluation set is built from real queries from the team, with the correct answer validated by a lawyer at the firm. Without that validation there is no measurement, only an impression.
Zero. No answer carrying a reference that does not resolve against a document in the index.
How it gets fixed
Post-generation verification. An answer with a broken citation is discarded and escalated, never shown half-corrected.
Agreed with the team before building, measured at clause level rather than document level.
How it gets fixed
Hybrid search: legal language has exact terms that semantic similarity alone does not separate.
Against a hand-validated batch, how many expiry dates and notice periods come out correct.
How it gets fixed
Structured output with schema validation, and the source clause mandatory on every row.
Questions whose correct answer is "this does not appear in the indexed documents".
How it gets fixed
A higher confidence threshold and escalation. In this domain, silence is a valid answer.
We do not build systems that issue legal opinions, draft pleadings or recommend litigation strategy. That is not a technical limitation: it is that the output would be an opinion with nobody accountable for it, and in this profession accountability has a name on it.
We are also not the answer if the archive is fifty contracts in a shared folder. This pays from several hundred live documents upward, with versions, annexes and amendments coexisting — which is when finding the right clause already costs billable hours.
Chunking, reranking and hybrid search, evaluated with recall@k and NDCG.
The system is already in production and answers badly. We measure it and determine what to fix.
MCP servers against the systems already in operation.
Architecture, evaluation criteria and cost per query before writing code.
Key takeaways
The failure mode specific to this sector is the invented citation: a perfectly formatted reference to a document that does not exist. You prevent it with post-generation verification, not with an instruction asking the model not to do it.
The architectural rule is simple and not negotiable: a citation that does not resolve against a real document in the index is never shown. The answer is discarded and the query escalates.
The cases that hold up are about document volume, not judgement: search across the contract archive, review against a template, extraction of dates and obligations, and a first pass on due diligence.
It does not replace a lawyer’s judgement and we do not pitch it that way. It replaces the hours spent finding which contract a clause ended up in — work that today falls to the most junior lawyer in the firm.
An AI system is worth what its sources are worth. These are the standard connectors; anything with an API or a database connects the same way, and what has no API is handled by file.
SAP
Enterprise ERP
Oracle
ERP and database
NetSuite
Cloud ERP
Salesforce
CRM and service
HubSpot
CRM and marketing
PostgreSQL
Database and pgvector
Microsoft SQL
Database
Snowflake
Data warehouse
BigQuery
Google data warehouse
Databricks
Data platform
Redshift
AWS data warehouse
Synapse
Azure data warehouse
Supabase
Managed Postgres
Workday
Payroll and HR
QuickBooks
Accounting
Sage
Accounting and ERP
Xero
Cloud accounting
Shopify
Catalogue and orders
WooCommerce
Catalogue and orders
Magento
Catalogue and orders
Stripe
Payments and subscriptions
Google Drive
Documents and folders
CSV y Excel
Flat files
Nothing in this category
We built and operated the assistant for a loyalty platform serving more than two million active users across ten countries.
We built the full pipeline: document ingestion and normalization, chunking, embedding generation, vector store on Azure AI Search and Pinecone, and retrieval with grounded generation on LangChain.
We held it above 99% availability for three years.
RAG systems →With post-generation verification, not with an instruction. Every passage in the index carries a stable identifier — document, version, clause — and the answer has to carry it through. After generation we check that each citation resolves against a real passage; if one does not, the answer is discarded and the system escalates rather than showing it. On top of that, the system can only answer from what is in the index: if the firm does not index case law, there is nothing to invent it from.
No, and anyone selling it that way is selling you a problem. It replaces one concrete task the junior does today: finding which document a clause ended up in, in minutes rather than hours. What it does not replace is the judgement of whether that clause serves you, how it gets negotiated or what risk it carries — which is where the lawyer’s value sits and where the system has nothing to say.
Only if you accept that, and in this sector the answer is usually no. We deploy on Azure OpenAI or Amazon Bedrock inside the firm’s own subscription, so documents and queries never leave your infrastructure. In any configuration we use enterprise plans where data sent through the API is not used for training, and we can filter names and identifiable data before inference where the engagement requires it.
It does, but the cost has to be stated. A scan needs optical recognition, and the quality of the extracted text caps everything downstream: if OCR swallows a negation, the system answers the opposite with complete confidence. So we measure extraction quality before indexing and report what share of the archive fell below the threshold. If much of the archive is poor scans, the right project starts with digitising properly, not with the assistant.
Yes, and it is among the highest-return cases because the correct answer is verifiable. The system aligns the incoming document against the approved template and flags which clauses changed, which are missing and which were added, each with the text of both versions side by side. It does not rule on whether the change is acceptable: it marks where to look, and that decision stays with the lawyer.
A first scope — search across the contract archive, for instance — usually takes six to eight weeks to production, with something running from week one. It is quoted with fixed scope, price and date in a proposal 48 hours after the first call. The price is driven by document volume, the state of the archive — scans cost more — and what integrations are needed.
It is a known pattern and it is almost never the model. The frequent causes are that the tool is not connected to the firm’s real archive but to a parallel repository nobody feeds, or that it gave a badly cited answer early on and lost the team’s trust, which is very hard to win back. The diagnostic takes a week, hands back the failures ranked by impact with each fix and its effort, and if it does not reach three actionable findings, it is not billed.
It gets fixed, it gets built, or it is not worth it. And if the diagnostic does not reach three actionable findings, it is not charged.
We use cookies to improve the user experience. Privacy
