Your team already knows how to answer product questions, prepare a useful brief and review a draft. The difficulty is making that expertise available without asking the same colleague—or rewriting the same prompt—every time.
Custom AI agents give recurring work a reusable role. This guide explains how to create a Campaign Strategist in Covey through conversation, customize it with AI assistance and use it with your team.
What are custom AI agents?
A custom AI agent is an AI assistant configured for a particular job, with instructions, relevant knowledge and the tools it needs. Instead of starting each conversation from a blank brief, you establish a reusable role and refine it as the work changes.
For a team, customization should answer four questions: What is the job? Which facts can the agent use? How should it work? Who can use its knowledge and tools? A name and a friendly personality are not enough.
alt：Diagram of role, knowledge, Skills and Plugins within a permissions boundary, leading to useful work and human review.
The usual approach: find the files, repeat the brief
Planning a campaign with one-off prompts can mean finding the product brief, copying brand guidance, writing detailed instructions and choosing tools before any useful work begins. The next campaign may require the same setup again.
That is unnecessary friction for a beginner. Instead of making a long configuration checklist the starting point, describe the job first and let AI help build the Agent. Refine its knowledge and methods as you use it.
Common problems a longer prompt will not solve
| Problem | Why it happens | What to change |
| The answer sounds generic. | The request describes the output but not the business or audience. | Add relevant company, product and brand context. |
| Two colleagues get different facts. | They supply different revisions or conflicting sources. | Agree on the approved source set and who maintains it. |
| The same mistake returns. | A correction fixes one answer but not the reusable instructions or source material. | Update the underlying rule or fact, then retest. |
| The Agent cannot complete a request. | The required source, tool or permission is unavailable. | Check the connection and access; do not add more prompt text as a substitute. |
The practical distinction is simple: instructions describe behavior, knowledge supplies evidence, Skills capture methods, and Plugins connect capabilities. Access controls determine who can use the relevant resources. These parts work together, but none replaces the others.
What Covey adds to a team workflow
Covey brings custom Agents into a shared working environment. Its value here is not just producing an answer: it connects the business background, the people requesting the work and the review that makes the result useful.
Business context behind the request
Company Context, Brand Kit, Products and Assets bring different parts of the business together. Company Context explains the background; Brand Kit guides voice and terminology; Products provides offering-specific information; Assets supplies relevant files and previous work.
For a Product Knowledge Agent, that means preparing more than a pile of documents. It means giving the Agent an agreed basis for understanding the question and framing a useful answer.
An Agent shaped around the job you know
Describe the job in plain language, then give the Agent the knowledge, Skills and Plugins it needs. A team’s expertise becomes reusable when it includes its working method: how to check a claim, structure an answer and flag an unresolved issue. You can keep shaping that role through further instructions rather than treating the first version as final.
A shared place to direct and review the work
Set a shared goal, involve the right Agent and review the contribution together. Keeping conversation, context and results connected helps the next contributor understand not only what changed, but why.
Custom Agents in the Slack conversation
With the Slack connection configured, colleagues can mention a custom Agent in Slack and bring its business context into the team conversation. For example, they could ask the Product Knowledge Agent to check a proposed claim against approved material before the team uses it in a brief.
Keep the audience in mind: an answer suitable for a private project may not be suitable for every channel. Verify the connection, who can invoke the Agent and what may be shared before using sensitive information.
Team resources with managed access
Covey supports sharing team knowledge, methods and connected tools, with control over who can use them. That gives the team a way to reuse its expertise without assuming every member should have identical access. Sharing an Agent is not a reason to make all underlying information available to everyone.
How to build custom AI agents in Covey: four simple steps
Create a Campaign Strategist, give it the right context and use it for your next launch. You can start with a conversation rather than configuring everything from scratch.
Step 1: Describe the Agent in a conversation
Open a new conversation and tell Covey what you want your Agent to do. For example:
I want to create an agent that helps me plan marketing campaigns. It should take a product launch or promo goal and turn it into a full campaign plan.
Covey helps turn that request into an Agent you can review and customize.
Step 2: Let AI help with the configuration
Review the Agent’s configuration page. Use Edit with AI to adjust its role and instructions in plain language, or edit the settings directly. Add your product brief and brand guidance, then select the Plugins and Skills the job needs. Complete any required connection permissions and save the Agent.
Step 3: Select the Agent and give it a task
Open the Agent selector below the conversation box and choose Campaign Strategist. Then describe the campaign you need:
Plan a four-week product launch using the attached brief and brand guidance. Suggest the audience, messaging, channels and weekly activities.
Review the response and ask for changes in the same conversation. The example below shows the selected Agent responding to a launch request.
Step 4: Reuse, customize and share
Your saved Agent is available for future tasks. Return to Agents whenever you want to change its instructions, knowledge, Plugins or Skills. Share it with colleagues or other people, managing who can use it. After configuring the Slack connection, you can also mention the Agent in a team conversation.
Which job should you start with?
Choose a task that repeats and has a recognizable standard for a good result. Product questions suit an Agent grounded in approved product material. A weekly update suits a reporting Agent with a fixed structure. Onboarding questions suit a role that uses current policies and knows when to ask a person.
Use a one-off conversation for work that rarely repeats or still has an unclear goal. Use a shared Agent when the team can name the job, maintain its knowledge and agree on how to review the result.
Frequently asked questions
Q.1 Do I need to code to create an Agent?
Covey supports describing the role in natural language. You still need to define the work and supply the relevant knowledge. Technical setup depends on which connections and actions the role needs.
Q.2 Does uploading company knowledge train a new model?
Adding reference material is not the same as training a model. Storage, retrieval and provider training policies are separate questions; check the current data terms before adding sensitive information.
Q.3 What is the difference between knowledge, Skills and Plugins?
Knowledge provides reference material. Skills capture reusable ways of working. Plugins connect supported tools and capabilities. Instructions guide how the Agent should use them, within the access it has actually been given.
Q.4 Can teammates use the Agent in Slack?
Yes, Covey supports mentioning custom Agents in Slack once the relevant connection is configured. Check Agent availability, channel audience and resource permissions; a Slack mention does not grant additional access by itself.
Q.5 How do I keep answers current after sharing?
Assign an owner to review the source set when relevant facts change. Confirm whether the Agent reads an updated source or an older supplied copy, and rerun representative tests after changes. Do not assume automatic synchronization.
Build a role the team can keep improving
Custom AI agents are useful when the team’s knowledge and methods become easier to reuse—not when they simply produce more text. Start with one job, prepare the right background, check the access and let colleagues review the result.
Covey by Converge AI describes that relationship as “Your AI business partner, uniting your team with agent teams”. Build your first role around a question the team already answers well, then use real feedback to decide what it should take on next.
