ChurnZero launched Kit, an AI assistant built into its customer success platform that synthesizes customer health data, team history, and business context to answer questions and suggest actions without leaving the app.

• Knows the full customer record: Kit draws on ChurnZero's own data layer, so answers are grounded in actual account history, health scores, and open tasks rather than generic retrieval.
• Designed for CS teams, not admins: Kit is accessible to front-line customer success managers, not just operators building prompt workflows.

SmartBear released two new capabilities inside BearQ, its agentic quality assurance system: Sources, which scans connected repositories and documentation to find untested code paths, and Workflows, which chains test-run steps into automated sequences.

• Sources targets the coverage gap that AI-generated code creates by ingesting codebase context and flagging areas where no test exists.
• Workflows removes manual orchestration by letting QA teams define multi-step test execution plans that BearQ runs without human intervention.

Sierra announced a partnership with Stripe that lets its customer-facing AI agents collect card payments, issue receipts, and close transactions inside a single conversation without transferring the customer to another system.

• Cedar Health is the first live customer, using Sierra's Stripe integration to let patients pay bills during an AI support call.
• No handoff required: the agent handles the full payment flow, confirmation, and receipt inside the same session, removing the payment-link step that previously broke the experience.

Grafana Labs announced a Knowledge Graph available in every Grafana Cloud stack, alongside Assistant Watchers that monitor AI workloads and autonomous Grafana Cloud Agents that can investigate and remediate incidents without human initiation.

• Knowledge Graph connects infrastructure dependencies to user experience and business impact, giving both human engineers and agents the shared context needed to triage a production problem.
• Assistant Watchers run continuously against AI agent workloads, alerting or acting when behavior deviates from expected patterns.

Dovetail made Channels 2.0 generally available to all customers, replacing the beta with a system that automatically groups incoming feedback into ranked ideas sorted by frequency, severity, account count, and the ARR sitting behind each problem.

• ARR-weighted ranking means product teams see the highest-revenue problems first, not just the loudest voices.
• Replaces the previous Channels experience after a closed beta in July and an open beta in September, making it the new default for all plans.

Databricks introduced Agent Memory inside its Agent Bricks developer platform, letting agents persist facts, user preferences, and learned workflows across sessions using Lakebase as the low-latency storage layer.

• Runs on Lakebase for reliability and speed, avoiding the cold-start problem where agents rediscover context on every new session.
• Built on Databricks' own memory research, the feature is available in beta today for teams already on the Agent Bricks platform.

Salesforce published details on its Agent Context Engine, a new layer that compiles trusted, shared context from CRM and external sources so multiple Agentforce agents working on the same deal or customer see consistent, governed information rather than each pulling independently from the CRM.

• Solves the multi-agent context problem where one agent's view of an account can contradict another's, leading to conflicting actions on high-value opportunities.
• Works across Agentforce workflows and is designed to be extensible to external data, giving revenue teams a single source of truth agents can act on.

Writer announced that its Snowflake data connector now enables continuous agentic execution, letting marketing agents monitor live data and act on it without a human prompt, replacing on-demand queries with always-on intelligence pipelines.

• Five agentic use cases are live, including campaign performance monitoring, content refresh triggers, and audience signal detection that runs against Snowflake tables in real time.
• No data tickets required: non-technical marketers can now get answers from CRM and data warehouse tables through the Writer agent rather than waiting on engineering.