Sirion

What it is

Sirion is the enterprise contract lifecycle management platform formerly sold as SirionLabs. The product name, the company name, and the domain all dropped the “Labs” — sirionlabs.com now redirects to sirion.ai — and the platform is organized around three phases the vendor calls Store, Create, and Manage. Since the October 2025 relaunch it sells as an agentic CLM: eight named agents (Search, Draft, Issue Detection, Redline, Extraction, Playbook, Obligations, Invoice) run on an in-house runtime called agentOS, and a conversational front end called AskSirion takes a plain-language request, picks the agents, and returns source-linked output with an explanation attached. Sirion reports 7M+ contracts under management worth close to $800B across 100+ languages, and it has held the top analyst slot in the category — Leader in the 2025 Gartner Magic Quadrant for CLM for the fourth consecutive year, and highest-ranked in every use case in the 2025 Gartner Critical Capabilities. The company it most resembles is Icertis.

Ownership changed on 23 February 2026, when Austin private-equity firm Haveli Investments completed a majority investment that bought out the earlier backers, Sequoia and Tiger Global among them. Founder-CEO Ajay Agrawal stayed and characterized the deal as a recapitalization rather than a sale — the company was already profitable, and the stated driver was collapsing an eleven-year-old cap table into a single controlling shareholder. Product continuity is not in question here the way it is after a strategic acquisition; pricing discipline under a new PE owner is the thing to watch.

Why it shows up in Legal Ops stacks

• Post-signature is still the reason to shortlist it. Most CLM platforms treat execution as the finish line. Sirion’s Obligations Agent tracks commitments and flags at-risk ones before a breach lands, and the Invoice Agent reconciles supplier invoices line-by-line against the terms that were actually agreed. That second one is the capability almost nobody else ships, and it is why Sirion wins in telecom, outsourcing, and managed-services portfolios where the money leaks after signature, not before it.
• It can be pointed at the repository you already have. The Extraction Agent turns an existing pile of executed PDFs into structured fields, which means the platform earns its keep before anyone changes how new paper gets drafted. Sirion publishes a 94.2% clause-extraction accuracy figure from its own benchmark; treat that as a vendor number and test it on your own clause types.
• Multi-model, with the sources shown. Sirion runs a mix of 10+ LLMs and 1,200+ in-house small models, routing each task to whichever fits, and every agent answer carries a direct link back to the clause it came from. For a legal team that has to defend a review, source-linking matters more than raw model quality.
• Buy-side is first-class. Procurement and supplier management are built in rather than bolted on, and the certified connectors go to the systems procurement actually lives in — SAP, Oracle, Microsoft Dynamics 365 — with Integration Studio and a Workato connector covering the long tail.

Pricing

There is no pricing page on sirion.ai and no self-serve tier; every call to action is a demo request. Two published anchors exist outside the vendor site, and they are the closest thing to a list price you will find:

• AWS Marketplace lists a 12-month contract for 100 users at $150,000 — an effective $125 per user per month — with everything else routed through a private offer. Note the unit: it is a 100-user bundle on an annual term, not a per-seat rate card, and the listing offers no smaller increment.
• UK G-Cloud 14, where the legal entity still files as SirionLabs Pte. Ltd, publishes £120 to £175 a user a month. That rate card dates to May 2024 and is the public-sector band, so read it as an upper bound on negotiated per-seat pricing rather than a current quote.

Taken together, an enterprise deployment realistically prices in the $125–220 per user per month range before negotiation, with a 100-seat rollout landing around $150K/year. Packaging follows the Store / Create / Manage split plus the agent set, and can be bought in parts — the common entry point is Store (repository plus Extraction and Search agents) with Create and Manage phased in later. Budget for an annual commit; nothing here is monthly.

Best for

Enterprises whose contract risk lives after signature — long-tail vendor obligations, telecom and outsourcing MSAs, service-level commitments that someone has to measure, invoices that have to be checked against agreed terms. The second strong fit is a procurement-led buying center where the biggest exposure sits in the supplier portfolio rather than in sales paper. Regulated and public-sector buyers get a third reason: Sirion completed an Australian IRAP assessment at OFFICIAL: Sensitive in August 2026 and holds a UK G-Cloud listing.

Do not buy Sirion if your paper is mostly sales-side at a company under a few hundred people, if you need the CLM to live inside Salesforce, or if you cannot staff a multi-month implementation. At roughly $150K/year for 100 users, the economics only work when contract performance is already something the business measures.

Versus the alternatives

• Icertis — the other enterprise leader and the head-to-head Sirion loses most often on ecosystem depth. Pick Icertis when you need a large bench of implementation partners in your region, or when the Microsoft relationship is load-bearing. Pick Sirion when the obligation and invoice-reconciliation side is the actual problem. The pairwise breakdown is at icertis-vs-sirionlabs.
• Ironclad — the default for sales-side workflow at US tech companies. Pick Ironclad when the bottleneck is getting outbound paper through legal quickly; it is the better in-house-legal workflow tool and the worse supplier-performance system.
• DocuSign IAM — the scale option, and the one procurement will raise if you already run DocuSign for signature. Pick it when consolidating agreement infrastructure matters more than depth in any one phase; Sirion goes deeper on post-signature by a wide margin.
• Luminance — the fastest-growing AI-native entrant in this space, and the right pick when the job is contract analysis and negotiation rather than end-to-end lifecycle management. It is materially cheaper and materially narrower.
• If the contract count is in the low thousands and the team is under ten people, Concord or LinkSquares solve the same problem for an order of magnitude less. Sirion at $150K/year is not a mid-market answer.

Watch-outs

• The AI numbers are the vendor’s own. 94.2% clause extraction and the “50% cycle-time reduction” from the relaunch announcement are both Sirion-run. Guard: define a pilot on 200–500 of your own contracts, score the Extraction and Issue Detection agents against a human baseline on your clause types, and write the accepted accuracy into the contract before signing.
• New PE ownership is a repricing risk. Haveli took control in February 2026 with a stated plan to expand go-to-market. Guard: negotiate the renewal escalator and support SLA now, on a multi-year term, rather than accepting a one-year deal and finding out what the second-year number looks like.
• No MCP server. The agents live inside Sirion’s own interface. If your plan is to query contracts from Claude or ChatGPT alongside the rest of your stack, that path does not exist today. Guard: scope the REST API and Integration Studio work explicitly during evaluation instead of assuming an assistant-side connector is coming.
• Thinner partner bench than Icertis or Conga. Guard: name your implementation partner and confirm they have delivered Sirion in your region and industry before signing, rather than assuming the vendor’s professional-services team has local capacity.
• Not Salesforce-native. Quote-to-contract works through a connector, not through a managed package living in your org. Guard: if sales operations owns the CLM budget, run the quote-to-contract flow end to end in the pilot — that is where the integration seams show.