On November 5, 2026, Agiloft unveiled three specialized AI agents within its Ask AI ecosystem, shifting its contract lifecycle management platform from answering questions about contracts to carrying out multi-step contract work. The Bulk Update Agent, the Contract Request Agent and the eSignature Agent for Adobe Acrobat Sign arrive alongside a new Multi-Pane Document Viewer and a Model Context Protocol (MCP) Server. The pitch centers on enterprise value leakage: contracts that are signed but never fully enforced quietly cost companies money, and Agiloft wants its agents to close that gap.
What Agiloft announced on November 5
Agiloft, a data-first contract lifecycle management (CLM) company based in Redwood City, California, announced the suite on November 5, 2026. The three agents build on capabilities Agiloft introduced earlier, including automated AI-Assisted Contract Intake and conversational Guided Contract Updates, and on Astra, its contract review and analysis product, which became generally available on July 14, 2026. Where those earlier tools mostly helped teams find and understand contract terms, the new agents execute changes, route requests and set up signatures. Agiloft framed the launch around research it published on September 15, 2026: a Censuswide survey of 2,005 chief procurement officers and other senior leaders at U.S. and U.K. companies with at least 250 employees found that at least 15% of supplier spend flows outside contracted agreements, while 52% of organizations failed to collect a contractual remedy in the previous year because it was not tracked or triggered in time. Read Agiloft’s official announcement and the PR Newswire release for the full details.
The Bulk Update Agent: from search to scale
The Bulk Update Agent lets users change up to three data fields across multiple contract and attachment records directly from Ask AI search results. Instead of opening each record to fix a renewal date or status, a team can apply the same change across a set of results in one step. The agent previews every affected file first and requires explicit human confirmation before executing an update, so a bulk change cannot run silently. That guardrail matters: a wrong renewal date or status can cascade into reminders, approvals and reports, so the design keeps a person in the loop at the moment the change is applied.
The Contract Request Agent: intake without the form hunt
The Contract Request Agent focuses on intake. A business user describes what is needed in natural language, and Ask AI identifies the appropriate existing contract request form based on the user’s intent and permissions, then routes the request. The idea is to remove the familiar friction of contract intake — knowing a contract is needed but not knowing which form, which team or which approval path applies. Because the routing follows the user’s permissions, a request cannot land on a form the user should not be using. Managed agent runtimes are becoming the enterprise pattern, as seen with Kimi Hosted Agents, Moonshot AI’s managed agent runtime, and Snowflake’s always-on CoCo and CoWork Bots offer another example of agents that keep working in the background.
The eSignature Agent: signing workflows in plain language
For signatures, the eSignature Agent prepares and initiates Adobe Acrobat Sign workflows through Ask AI. Users describe what they need in natural language, and the agent guides them through selecting documents, identifying signers, establishing the signing order and reviewing the details before anything is sent. Signature setups still require human review before sending, so the agent does the assembly work while the person keeps the final say. For legal and procurement teams, that combination — agent speed with a human checkpoint — is the difference between automation they can adopt and automation they have to work around.
The Multi-Pane Document Viewer and the MCP Server
The release also introduces a Multi-Pane Document Viewer for working across contract documents side by side, and an MCP Server intended to give external software tools a standard interface to Agiloft’s central contract repository. Anthropic’s Claude is available as an MCP client in the beta environment, and external-agent access remains in beta while the rollout continues. The MCP move is the strategic piece: if outside agents can reach the contract repository through a standard protocol, Agiloft’s data becomes usable by whatever agent a customer already runs, rather than only by Agiloft’s own Ask AI. See Unite.AI’s report on the launch for more coverage.
Why this matters — and what to watch next
Agiloft’s launch is part of a wider shift in enterprise AI: agents are being judged on completed jobs, not single answers. The same logic is driving managed agent runtimes elsewhere in the industry, from Together AI’s Polymorphic Inference, which retunes infrastructure behind models automatically, to OpenAI’s planned acquisition of Pierre Computer, which buys the Git infrastructure under its Codex coding agent. Three things will decide whether Agiloft’s agents matter as much as the company hopes. First, accuracy at scale: bulk updates that touch dozens of records must be right the first time, because errors propagate into workflows. Second, adoption: the agents only help if business users trust them enough to describe a need in plain language instead of falling back to email and manual forms. Third, the MCP bet: opening the contract repository to external agents is powerful only if customers are comfortable letting outside agents read their agreements. If those three hold, the contract repository stops being a place where agreements sit after signing — and starts being a system that acts on them.
