--- title: "Legal AI that lives in your ERP: contract obligations as data, not PDFs" canonical_url: "https://cognilium.ai/blogs/contract-obligations-as-data-not-pdfs" slug: "contract-obligations-as-data-not-pdfs" section: "blogs" date_published: "2026-10-12" date_modified: "2026-10-12" word_count: 863 reading_time_minutes: 4 author: "Mudassir Marwat" author_identifier: "0009-0008-1927-2598" person_same_as: - "https://www.linkedin.com/in/mudassir-marwat/" - "https://orcid.org/0009-0008-1927-2598" - "https://github.com/mudassirmarwat" - "https://medium.com/@mudassir-marwat" - "https://www.youtube.com/@mudassir-marwat" - "https://dev.to/mudassirmarwat" - "https://hashnode.com/@mudassirmarwat" - "https://huggingface.co/mudassirmarwat" - "https://bsky.app/profile/mudassir-marwat.bsky.social" - "https://substack.com/@mudassirmarwat" - "https://topmate.io/mudassirmarwat" - "https://fueler.io/mudassirmarwat" - "https://www.instagram.com/mudassirmarwat/" - "https://www.facebook.com/mudassir.marwat" - "https://www.reddit.com/user/mudassirmarwat/" - "https://www.f6s.com/member/mudassir-marwat" - "https://cognilium.ai/founder" entities: [] related: [] --- # Legal AI that lives in your ERP: contract obligations as data, not PDFs Your contracts are signed, filed as PDFs, and forgotten — until a renewal rolls or an obligation is missed. Here's how legal AI built into your ERP turns the paper into data that lives against the record. A contract's most dangerous moment is not when you sign it. It is six months later, when everyone has forgotten what is in it. The agreement is negotiated, signed, and dropped into a drive as a PDF. The obligations it created — the renewal notice window, the price cap, the SLA credit, the thing you promised to deliver by a date — are now buried in prose, in a file nobody opens, disconnected from the systems where the work actually happens. And they stay buried until one of them costs you: an auto-renewal that rolled because the notice window closed, a price you kept paying above the cap, a commitment you missed because it lived in a document and not on the record. The problem is not that your team is careless. It is that a contract is stored as a *document* when what the business needs is *data*. ## The gap between where contracts live and where work happens In most companies, contracts and the operational record live in two different worlds. The agreements sit as PDFs in a repository or a shared drive. The operational reality — the vendor, the purchase order, the finance record — lives in the ERP, in Dynamics 365, SAP or Business Central. Nothing connects them. So the obligation that should trigger an action in the ERP has no way to get there. It is trapped in prose, one system over. A document-first tool — a place to store and search contracts — does not close this gap. It makes the PDFs easier to find, which is useful, but the obligation is still a sentence in a file, not a field on the record. Someone still has to read it, remember it, and re-key it into the system where it matters. ## Turning the paper into data that lives against the record The fix is to treat a contract the way you treat any other source of business data: read it, extract what matters, and put it where the work happens. Legal AI built into the ERP does exactly that. It reads each agreement, pulls out the obligations and the dates and the terms, and returns them as **structured fields, not prose** — a renewal date that becomes a field on the vendor record, an obligation that sits against the party it binds, a price cap that can be checked against what you are actually being invoiced. And every extracted value carries the clause it came from, so a person can see the source and trust the figure enough to act on it. Two disciplines make this something a legal team will actually rely on: - **It reads against your own playbook, not the internet.** A general model tells you what is *standard*. An in-house team does not care about standard; it cares about *their* positions and red lines. So the review is done against the company's own rulebook, and every flag traces back to the exact rule that raised it. It never invents a position — it applies yours. (This is the pipeline behind our own product, [Paralegent AI](/products/paralegent-ai): specialist agents review a contract clause by clause against the firm's own playbook, and a review that took hours comes back in five to ten minutes. Smart routing — sending only the ambiguous clauses to the specialists that matter — cut model calls on that build by roughly 75%.) - **A person always decides.** The system does the reading, the extraction and the first-pass flagging. The lawyer makes the call on what is left. That is the design, not a limitation — AI removes the work that never needed a lawyer, so the lawyer's time goes to the work that does. ## Why it belongs *inside* the ERP, not beside it The most expensive legal AI is the one nobody opens. A tool that sits outside the systems people already use gets tried for a week and quietly abandoned, because it asks everyone to switch context all day. Building the capability into the ERP removes that switch entirely. The obligations surface where the vendor and the PO already live. The review happens where the lawyer already works. There is no second system to keep in step, and — critically for regulated or privileged work — the agreements never have to leave the customer's own environment. That is also why this complements, rather than competes with, whatever contract repository or implementation partner you already have. We do not replace your systems. We add the layer that turns the paper inside them into data you can act on. ## Where to start Pick the obligation that has cost you before — the renewal you missed, the cap you overpaid, the commitment that slipped — and start there. Prove that the paper can become a field on the record, and the rest follows. See [Legal AI Ops for Dynamics 365 and SAP](/solutions/legal-ai-ops) for the full picture, or [AI contract review](/solutions/legal-contract-review) for the review pipeline specifically. If your contracts live in your ERP and the obligations keep slipping, [tell us what you run and where it is slipping](/contact) — you will get a straight answer, including "you do not need us for this" when that is true. ## Frequently asked questions ### What does it mean to treat contract obligations as data? It means reading the signed agreement and returning what matters — renewal dates, price caps, SLAs, commitments — as structured fields on the record, each carrying the clause it came from, instead of leaving them as prose buried in a PDF. ### Can't a contract repository already do this? A repository makes PDFs easier to store and search, which helps. But the obligation is still a sentence in a file, not a field on the vendor or the PO. Someone still has to read it and re-key it into the system where it matters. This closes that gap. ### Does the AI review contracts against our own playbook? Yes. It reads against your own rulebook, not general legal knowledge, and every flag traces back to the exact rule that raised it. It applies your positions; it never invents one. ### Does legal AI replace our lawyers? No. It removes the work that never needed a lawyer — the reading and the searching — so the lawyer's time goes to the judgement calls. A person decides on every finding. ### Do our contracts have to leave our environment? No. The capability runs inside your own environment, which is what makes it workable for regulated or privileged material — the agreements never leave your tenancy. --- Canonical HTML: https://cognilium.ai/blogs/contract-obligations-as-data-not-pdfs