Before You Begin
Choose one call objective, the target caller, the information the agent must collect, the allowed outcome, and the situations that require a human.DialNexa Agent Builder Workflow
Use Agent Builder when you want a first draft faster than writing the agent from a blank page. Choose Proceed to Agent Builder from Create agent, then complete Goal & Context, Call Playbook, and Personality. Choose Manual Build when you want to start with a template or blank configuration.Agent Builder creates a draft. Review the prompt, variables, voice stack, and post-call fields before publishing or assigning the agent to live traffic.
When To Use Agent Builder
What The Builder Asks For
The builder uses three intake sections.
After the first section, DialNexa can draft success criteria, information to collect, and objections from the company, caller, and goal. After the second section, it can draft the agent tone from the objectives you have accepted or edited.
Reuse Saved Businesses In Agent Builder
When your workspace has saved business context, the Goal & Context step offers Saved Businesses beside Company & what you offer. Open the picker and select the business this agent serves. Review the populated business details before continuing; saved context is not applied automatically to every new agent.
Choose The Suggested Agent Type
On the Personality step, review the suggested type and select another card if it better fits the call. A recommendation does not create or publish an agent. Use Single Prompt for one instruction set, Conversational Flow for explicit branches, Speech to Speech for realtime conversation, and IVR for keypad routing. Manual Build also offers a description-based recommendation on the type screen. It can ask one clarifying question or suggest an IVR menu followed by a conversational agent. A combined recommendation requires configuring both agents and their handoff; it does not automatically connect two agents.Build Phases
When you click Create Agent on the Personality step, the builder streams progress through three phases.
Review The Draft Before Creating
The result screen can show quality scores, issues, suggested fixes, dynamic variables, and default values. Review these before creating the agent.
1
Check the quality summary
Read the builder’s summary and any prompt, post-call analysis, or technical fit issues.
2
Review dynamic variables
Fill safe default values for each detected variable. Leave values blank only when the value will always be injected by API, batch CSV, workflow lead data, or test call input.
3
Create the draft agent
Click Create agent only after the draft looks usable. DialNexa creates a draft using the selected agent type and supported settings. For Conversational Flow, inspect the generated graph and each node. For IVR, inspect menu keys and handoff targets. For Speech to Speech, review the realtime model and voice.
4
Test and publish
Place test calls, edit the draft where needed, then publish a version before using it in phone numbers, batch calls, workflows, web calls, or API calls.
What Agent Builder Creates
Editing An Existing Agent With AI
After an agent exists, the agent editor can show a DialNexa AI modifier for natural-language changes on Single Prompt Agents. This modifier lives inside the agent editor page. It is separate from the guided Agent Builder create flow described earlier on this page. Use it for focused prompt edits, configuration updates, and optimization suggestions such as tightening an opening line, adding a qualification question, changing tone, adjusting a supported setting, or improving a known issue from testing. When recent call evidence is available, the empty-state suggestions can include prompts based on patterns from that agent’s calls.
Config proposals can cover LLM fallback, predictive preprocessing, Prompt Caching where your workspace has access, transcriber choices, Boosted Keywords, Audio Cache, denoising, timezone, voice choices, post-call fields, and knowledge base attach or detach requests. Unsupported or unsafe changes stay unresolved for manual review.
Assistant Memory And Proposal Decisions
The assistant can retrieve confirmed business facts, earlier assistant-conversation recaps, and accepted or rejected agent proposals. Ask a focused question such as “What did we decide about the welcome message on Support Agent?” Include the agent or business name when your workspace serves more than one business. Memory describes editing history and business context, not call transcripts. Check current agent settings and call records before relying on an older decision. A saved fact for one business should not be applied to an agent for another business. After rejecting an agent-edit proposal, you can save an optional reason. Explain the constraint, such as “Keep this question because it is required for appointment booking,” so later suggestions have that context. A rejection reason does not apply the rejected change. Connected MCP clients can use Get Memory to retrieve the same categories of workspace memory.Workspace Questions And Approved Actions
Use the workspace assistant from the dashboard when you want operational context without opening a specific agent. The same workspace lane is available inside the agent editor for questions that reach beyond the current prompt or settings. It can read supported data for the active workspace and answer with current names, IDs, statuses, counts, billing limits, wallet balance, call details, retry chains, transfers, dependencies, and webhook configuration. If the requested information is unavailable, it should say so instead of filling the gap. The assistant prepares one reviewable action card at a time. Card wording and displayed argument rows come from DialNexa’s action registry, not free-form model text. Reversible actions use Run, actions that spend wallet balance use Approve & run, and destructive actions use Confirm & run with an irreversible warning where applicable.
An approved card can show a concise saved result and a follow-up button, such as Go to the agent, after successful execution. A failed card shows the operation error and does not prove that the requested state was saved.
1
Ask for the workspace fact or action
Name the campaign, batch, call, agent, phone number, workflow, or other resource as clearly as possible.
2
Review the answer and target
Confirm that the returned resource name and prefixed ID belong to the intended workspace.
3
Approve only the exact action
Check every displayed ID, name, version, phone number, schedule, amount, and consequence. Use the action button only when the card matches the request. Destructive and billable cards use stronger labels and colors.
4
Verify the result
Read the result shown on the card, follow any resource link, and confirm the saved state on the relevant dashboard page. After a failed or timed-out card, read the current resource before retrying.
Common Mistakes
Submitting vague business context
Submitting vague business context
The builder needs the company, caller, and outcome to be concrete. “Handle support calls” is weaker than “triage dental appointment questions, collect patient name and preferred slot, and book a callback when insurance details are unclear.”
Accepting generated objectives without editing
Accepting generated objectives without editing
Treat generated objectives as a draft. Remove goals the agent should not pursue and add required handoff or escalation rules.
Creating before checking variables
Creating before checking variables
Variables with empty defaults can make the agent sound incomplete if runtime data is missing. Review every detected variable before test calls.
Skipping real test calls
Skipping real test calls
Builder simulations do not replace phone or web call tests. Always test with names, numbers, objections, interruptions, and the route you plan to use.
Publishing an AI-edited agent without reviewing proposals
Publishing an AI-edited agent without reviewing proposals
Treat AI proposals as drafts. Review highlighted prompt changes, check configuration changes, confirm variables and post-call fields that depend on the prompt, then test before publishing.
Verify The Generated Agent
Review every generated field before creating the agent. After creation, publish a test version and confirm the prompt, welcome message, variables, post-call fields, provider stack, and Call History evidence.Recap
Use Agent Builder for a structured first draft, not an automatic production release. Define one objective, review every generated field, publish a test version, and validate the conversation in Call History.Related Reading
Types Of Agents
Confirm when Agent Builder is the right start path.
Single Prompt Agents
Learn when one prompt is the right structure.
Dynamic Variables
Set safe defaults for placeholders.
Testing Agents
Verify the generated draft before publishing.