Customer service is being stretched to its limits. Teams are under pressure to do more — resolve more cases, meet rising customer expectations, and justify every dollar of investment — all while dealing with disconnected tech stacks and growing volumes of content.
That’s why service leaders are turning to a new generation of AI; one that doesn’t just assist humans but acts autonomously. Salesforce’s Agentforce is leading the charge, promising AI agents that can go beyond suggestions to take meaningful action.
But here’s the challenge: AI agents are only as good as the information they’re built on. Without trustworthy, relevant, and up-to-date content, even the smartest AI risks hallucinations, low adoption, and ROI that’s hard to prove — the same hurdles many teams faced with early GenAI rollouts.
That’s where Coveo for Agentforce comes in. Coveo grounds Agentforce with the most relevant, trusted knowledge, improving its ability to execute on actions like question answering and case classification, boosting self-service success and service efficiency.
With Coveo for Agentforce 2.0, these capabilities go even further. It now delivers higher relevance and personalization by passing custom context for more precise retrieval. And with support for Agentforce Flex Prompt Templates, Coveo’s Passage Retrieval API packaged as an Apex class unlocks new actions for customer service use cases.
Grounding Agentforce With the Most Secure, Relevant Knowledge
Most enterprises run on knowledge ecosystems that span far beyond Salesforce. While Salesforce’s Data Cloud offers some built-in capabilities, it struggles with large volumes of complex unstructured data. The platform requires disruptive migration, has gaps in connectivity and security that can hurt Agentforce performance, and the rising costs of indexing and querying data discourage adoption.
That’s where Coveo comes in. With the ability to index vast volumes of enterprise content — across SharePoint, Confluence, Jira, Google Drive, wikis, and more — Coveo brings the full depth and complexity of your knowledge ecosystem into reach.
Once integrated, Coveo becomes Agentforces’ foundational knowledge hub. Using advanced AI and machine learning, Coveo finds and delivers the most relevant, high-quality content to Agentforce in real-time. This means Agentforce always has access to fresh, trusted information, supercharging its ability to deliver higher case deflection and self-service resolution outcomes.
That’s why leading companies like Xero are already seeing results. With Coveo for Agentforce, they built a self-service chatbot on Facebook Messenger that deflected 75% of incoming messages, cut time spent monitoring social channels by 80%, and reduced escalations to special assists by 30%.
And while question answering and case classification are powerful, our customers wanted to customize these out-of-the-box actions and unlock powerful use cases. Well, the wait is over with Coveo for Agentforce 2.0!
Coveo for Agentforce 2.0: Expanding What’s Possible With Greater Context & Control
Deliver context-aware answers tailored to each user
You can now send extra context or case details directly to Coveo’s retrieval system. This means Agentforce can automatically include specific information about each case or user when searching for relevant content, helping it give more personalized and contextually aware responses.
This works across self-service and assisted service use cases. For example, while engaging a customer, Agentforce can tap Coveo to ensure that customer’s geographic location is used to prioritize region-specific documentation. And in assisted service, if a support case includes metadata about product classification or account type, Coveo dynamically incorporates those details when retrieving most relevant content.
Supported context types include:
- Case subject and description
- Product or issue classification
- User location or region
- Account or entitlement metadata
- Any other Salesforce field available at runtime
By enriching retrieval with custom context, Coveo augments Agentforce to surface content that’s not only relevant to the query, but tailored to the user and their unique scenario.
It also opens the door for intelligent rules and branching. For instance, if a case description contains “warranty dispute,” you could prioritize policies or escalation playbooks in the retrieval pipeline. If the customer is located in EMEA, you could prioritize region-specific documentation. It’s the same logic behind our Case Assist action — except now you can apply it anywhere.
Coveo for Agentforce 2.0 takes your Agentforce experience beyond generic answers to deliver the right knowledge for every user, in every moment.
Unlock new actions with Flex Prompt Template support
The real power of Agentforce isn’t just in its pre-built agents — it’s in how flexibly they can be tailored to fit your business. That’s where Flex Prompt Templates come in. Built into the Agentforce Prompt Builder, they allow teams to define the tone, structure, and format of LLM-generated responses.
By integrating Coveo, you can extend these capabilities with context-rich retrieval that augments actions driven by Agentforce. By combining these enhancements, clients can unlock entirely new Agentforce actions and use cases. For example:
- Automated Case Resolution: Augmented with Coveo, Agentforce can take a case from start to finish, developing potential solutions based on historical knowledge, similar cases, and other relevant content. Service reps act as the human-in-the-loop, verifying solutions and selecting the right option. Ultimatewly, this speeds case resolution, saves time and money, and emopwers human proficiency.
- Case Follow-Up: After a case is resolved, Agentforce can use Coveo-retrieved content to generate a follow-up email that’s accurate, personalized, and written in your company’s voice.
- Knowledge Article Creation: When new insights are surfaced during a customer interaction, Agentforce can leverage Coveo’s retrieval and case metadata to draft a structured knowledge article — no manual authoring required.
- Internal Communications: For weekly internal updates, Agentforce can build a formatted newsletter using the latest, most relevant content retrieved by Coveo — and shaped by your preferred prompt style.
In each scenario, Flex Prompt Templates define the structure; Coveo ensures the information is relevant and reliable. Together, they allow teams to orchestrate intelligent, business-specific workflows within Agentforce.
Use All of Your Enterprise Knowledge With Coveo for Agentforce
Coveo is deeply integrated with Agentforce, and enhancements like Flex Prompt Templates and custom context are designed to help organizations get the most out of the platform — delivering AI agents that are faster, smarter, and contextually aware.
And while our retrieval platform is compatible with a wide range of LLM ecosystems, the focus here is clear: enabling Agentforce to perform at its best with trusted, enterprise knowledge.
Want to discuss the potential for your business in person? Come meet us and hear Illumio’s story at Dreamforce in San Francisco in October!
“When we combined Agentforce with Coveo, the results were on another level. Our internal testers went from saying, ‘this is okay’ to literally saying ‘wow.’ The multiplier effect of bringing the two solutions together was incredible—consistently returning responses that truly amazed our team.”
– Joyce Leung, VP of Support Operations at Illumio
If you won’t be in California, book a personalized demo to see how Coveo + Agentforce can help your team deliver faster resolutions, higher deflection, and a foundation for agentic success.
Dig Deeper
See why hundreds of enterprises are augmenting and extending their Salesforce foundations (including Salesforces Agentforce) with Coveo’s agnostic AI search & generative capabilities.
Coveo for Agentforce 2.0 represents the second major release of Coveo’s packaged integration for Agentforce. It indicates the evolution of Coveo’s capabilities and enhancements in this integration specifically, independent of Salesforce’s internal versioning.