MCP Connector Integration with Claude

Conga’s MCP server supports integration with Claude, allowing connection and configuration without manual tenant-level setup. The server provides CLM, CPQ and Advantage Platform tools to AI clients based on user roles and product entitlements, and is accessible via production-ready, publicly reachable endpoints in North America, Europe, and Australia over HTTPS. Endpoints authenticate via OAuth 2.0, enabling organization identification through user credentials rather than client IDs. Streamable HTTP transport is available on all endpoints. For more, see Configuring Conga Advantage MCP Server for Claude and Accessing AiMe with Claude.

MCP Connector Integration with Microsoft Copilot

Conga’s MCP server supports integration with Microsoft Copilot, allowing connection and configuration without manual tenant-level setup. The server provides select tools to external clients based on user roles and product entitlements, and is accessible via production-ready, publicly reachable endpoints in North America, Europe, and Australia over HTTPS. Endpoints authenticate via OAuth 2.0, enabling organization identification through user credentials rather than client IDs. Streamable HTTP transport is available on all endpoints. For more, see Configuring Conga Advantage MCP Server for Microsoft Copilot and Accessing AiMe with Microsoft Copilot.

AiMe Search Skill

This release introduces major enhancements to Search Skill (was Conga Search Agent), unifying keyword lookup, natural-language queries, structured metadata filtering, and deep semantic contract intelligence into a single, intuitive conversational experience.

With this release, contract and legal operations teams no longer need to switch between different search modes or navigate complex menu hierarchies. Users can ask natural-language questions, search by keyword, or run complex clause comparisons across the entire contract repository, receiving visually structured answers, actionable comparisons, and exportable reports. Benefits include:

• Unified Search Bar: Enter contract IDs, party names, metadata criteria, or natural-language questions in one place. The system automatically categorizes results across Contracts, Accounts, Clauses, and Documents.
• Semantic Clause Discovery & Comparison: Search concepts beyond literal keywords (e.g., uncapped liability, non-standard indemnity) and compare clauses across up to 30 contracts or amendment chains side by side.
• Amendment-Aware Intelligence: Automatically trace contract hierarchies to surface the operative clause from the latest active amendment, rather than using superseded terms.
• Interactive Results & Visualizations: View key agreement details on clean object cards, explore complex contract families via expandable accordions, filter/sort interactive tables, and view metric breakdowns in embedded charts.
• Streamlined Reporting & Deep Links: Export filtered search results directly to cleanly formatted CSV/Excel reports and open records directly in Salesforce or Conga Platform.

See Search Skill for more.

Conversation History Service and Short-Term Memory

Every agent conversation persists to durable storage. This enables an auditable conversation history as well as full conversational awareness and agent recall so that interactions flow naturally with users not having to repeat themselves. Real-time persistence ensures no data loss from session interruptions. All exchanged messages, context snapshots, invoked skills, and user and tenant identifiers are recorded. Conversations of any length are supported.

Persistent conversation history creates an auditable record of AI-assisted decisions. Users can soft-delete conversations via the UI, while administrators have access to hard-delete for auditability. See Conversation Logs and Reviewing AiMe Chat History for more.

Rule-Based Extraction

Discovery’s provision model (pattern)-based extraction approach is augmented with the addition of rule-based extraction. This improves accuracy at the cost of modest latency. The extraction method and decision path are explained in the user interface and the administrator configuring the custom provision can respond to the inferred rules to improve the custom provision model’s accuracy. See Adding a Custom Provision Model and Reviewing Imported Clauses and Fields for more.

Conversation History Control

Advantage Platform administrators can export Aime Assistant conversation histories in 10,000-message batches, filtered by user or date range, in CSV format. Administrators are notified by email when an export job is complete. File exports are purged after 7 days. See Conversation Logs for more.

AiMe Chat Enhancements

AiMe is enhanced with a set of experience improvements that make it more capable, consistent, and accessible.

• AiMe now provides a consistent, accessible, context-aware experience with deployment support, supporting multiple invocation modes, skill-aware entry, real-time streamed responses to the user on ongoing activities, and robust error handling.
• Users can attach up to five Excel, DOCX, PDF, or TXT files, up to 25 MB each, per prompt for specific Skills.
• The assistant offers a maximized full-page mode for complex multi-turn conversations.

See Asking AiMe a Question for more.

Risk Framework for Redline

Redline's rule authoring now classifies unfavorable findings by severity instead of treating every unfavorable condition the same way. When creating or editing a clause or document rule, the single Unfavorable section is replaced with three tiers: High Risk (dealbreakers requiring immediate escalation), Medium Risk (negotiation points needing attention), and Low Risk (minor concerns and preferences), each with its own color, icon, and helper text, and each accepting the same free-text input as before. If the same condition or a similar one is entered in more than one tier, an inline warning flags the potential duplicate so the admin can confirm it's intentional or remove it before saving.

During contract review, Redline evaluates each rule against these tiers and returns a High, Medium, or Low risk category using a waterfall: any High condition makes the rule high-risk, regardless of what else was found; otherwise, Medium wins over Low; a rule with nothing triggered is No Risk. The same waterfall rolls up to the contract level, where the Overall Risk category is the highest category found across all rules. Alongside the category, the response lists the triggering conditions and which tier each belongs to, including decomposing compound conditions (for example, a single sentence covering a liability cap, mutuality, and multiple carve-outs) into individual, separately classified items. See Risk Framework for more.

Rule authoring also adds a Scan Clause Text action that uses AI to propose an editable list of favorable conditions taken directly from a rule's standard or sample clause text. Redline uses favorable conditions to align clauses more closely with your company's preferred contractual position, even when no risk is identified. This helps generate recommendations by eliminating identified risks while maintaining your preferred position. See Creating a Clause-level Rule for more.

Pre-Production Worksheet Testing

Administrators can upload test documents, run Discovery AI extractions, review and analyze results with confidence scores, refine configurations, and re-test to achieve acceptable accuracy without having to push a worksheet into production. See Testing Worksheets for more.

Question, Insight, and Risk Detection Improvements

During review, Discovery Agent now automatically provides risk recommendations and mitigation guidance. Risk criteria configuration and level-setting are simplified with a visual rule builder. Discovery Agent now offers common questions in addition to preconfigured questions. You can also test risk insights against sample documents, previewing likely LLM answer responses. See Configuring Discovery Automatically for more.

Automatic Data Extraction Configuration

Discovery uses intelligent provision-to-field mappings and smart suggestions to automate data extraction configuration. Automating the complex mapping process reduces errors, ensuring extracted data flows to the correct CLM fields, with system matches by informed by name similarity, data type, and historical patterns. Discovery pre-validates before saving (preventing type mismatches and duplicates), warns for common mistakes, and offers previews of mappings and test extraction before committing. Discovery maps data extraction setups, creating a default worksheet if none is detected, maps fields and selects provision models from the clause library, and mapping tables and obligations. See Automated Configuration Assistance for more.

Custom Provision Creation and Training

Users can create custom provisions with comprehensive training either by automatically extracting 20–30 sample documents for all custom provision opportunities or by manually selecting 2–5 examples per provision. AiMe validates example quality and predicts extraction accuracy. See Training Discovery on New Custom Provision Models for more.

Sample Contract Analysis and Recommendation

Discovery Agent can automatically classify contract types and recommend extraction templates based on samples, discovering, counting, and grouping provisions across uploaded contracts. The AI agent calculates the provision frequency and grouping of similar provisions and provides descriptive group names for both out-of-the-box and custom provision models. Discovery AI also suggests confidence scores for identified provisions based on provision frequency and the total number of uploaded documents, supporting consistent, transparent provision selection for contract analyses.

This enables administrators and implementation teams to validate and configure contract provisions based on real contract data, reducing post-implementation rework and increasing first-time extraction accuracy. See Automated Configuration Assistance for more.

Aime Actions Usage Dashboard

The Aime Actions Usage dashboard adds reporting for system administrators, enabling AI system usage monitoring by product feature and user.

The dashboard displays the top five features and top ten users, with all others grouped into "Other" categories. A stacked area chart presents Aime Action consumption trends over time, with toggles for weekly or monthly views. Administrators can access dynamic usage insights, including average consumption, total usage, and growth rate, all of which update in real time according to selected filters.

The dashboard supports multi-select filters for both features and users, as well as a date range filter. See AiMe Actions Dashboard for more.

Clause Library Sample Document Workflow

Discovery can accept sample Word documents for extraction outside the standard CLM agreement workflow. You can specify the worksheet to control which provisions are extracted from the sample document. If you do not select a worksheet, Discovery uses a default worksheet for records of the given record type. If no default worksheet is available for the record type, Discovery creates one. When you are testing sample documents, the clause library extraction pipeline does not create CLM agreement records or dispatch Ready for Review notifications. See Automated Configuration Assistance for more.

Salesforce Integration Support for Redline

Salesforce for Conga CLM users can now access Redline, which integrates Redline solutions with Salesforce environments. This update facilitates access and interoperability for organizations using both platforms. See How Redline Works for more.

Historical Clause Recommendation (HCR) for Redline

Redline now surfaces your organization's own negotiation history as a recommendation source during contract review. For each clause under review, HCR identifies previously signed contracts with the same or a similar profile, matching first on counterparty then falling back through industry, contract type, and region, building a pool of up to five most-relevant matching agreements. It resolves each agreement's full amendment history to the clause’s most recent version, using them to synthesize a recommended clause.

During review, this appears in a Historical Precedents card alongside any existing rule-based recommendations for that clause. The card shows the synthesized language, an attribution line noting how many contracts it's based on and what matched them, and a risk flag if there is a defined rule for that clause type. Reviewers can apply the recommendation with one click, replacing the clause as a tracked change, or leave it and continue reviewing.

Historical Clause Recommendation can be independently enabled or disabled for each playbook.

See Historical Clause Recommendation for more.