Guides6 min read

Tuning Your Agent's Behavior

Fine-tune how your agents communicate and make decisions — from the job instruction and review modes to tone and language preferences.

Shaping What Your Agent Does

Everything below lives on the agent's page, which scrolls through its sections. Company admins can edit them; other members see them read-only. Every change is saved as a draft and takes effect when you publish the agent.

The Job

Every agent has a The Job section with two fields:

  • Job title — a short imperative phrase naming the role (e.g., "Handle first-line billing support")
  • Job instruction — one clear instruction that anchors every run (e.g., "Resolve billing questions from incoming support tickets, escalate anything outside billing scope")

The section ends with a Runs when line saying what starts the agent; see Scheduled Triggers.

Tip

Keep the instruction focused on what the agent should do, not how it should say things — tone and language are set separately under How it communicates.

What it does

The What it does section lists the agent's capabilities. Company admins can turn capabilities on or off and set the review mode for each action group. See Review Mode Progression below.

How it works

The How it works section shows a diagram of how a run flows and the agent's operating manual — the detailed instructions that shape its decisions. Symbi writes the first version; company admins can edit it. What you read here is exactly what the agent is given.

How the Agent Presents Itself

Who it is

The Who it is section shows the agent's persona. Like the operating manual, Symbi writes it and company admins can edit it. For most agents it is deliberately short.

How it communicates

The How it communicates section lets company admins set communication preferences that apply to every reply this agent sends:

  • Communication Tone — choose from Professional, Friendly, Formal, or Casual
  • Response Language — leave blank to let the agent auto-detect, or enter a language code (e.g. en, no, de) to always respond in that language
  • Max Response Length — an optional cap in characters (50–5000). Leave blank for no cap
  • Sign-Off Text — an optional closing phrase appended to every reply
  • Greeting — toggle a greeting on or off, and customize the text

Each field saves to the draft as you leave it. Publish the agent to make the changes live.

Classifier Categories

For agents that categorize incoming work (tickets, emails, messages), a set of categories controls how items are labeled. Categories are set when the agent is built; there is no control for them on the agent's page yet. Contact Symbi to change an agent's categories.

Typical category sets:

  • Ticket priority — low, medium, high, urgent
  • Request type — question, complaint, feature request, bug report
  • Routing — which team or workflow should handle the item

Company Profile Influence

Your company's purpose, vision, values, and brand narrative (set in Settings) are automatically injected into every agent's context. All your agents share a consistent understanding of your brand — no per-agent configuration needed.

Tip

The company profile is the single most impactful setting. A clear brand narrative improves every agent's output across the board.

Knowledge Base

The knowledge base gives your agent factual context to draw from. To get the most out of it:

  • Keep it focused — quality over quantity. A few well-maintained sources outperform a large, outdated collection
  • Prioritize FAQs — frequently asked questions and common scenarios should be front and center
  • Remove outdated content — stale docs cause bad answers
  • Update after process changes — whenever a policy or workflow changes, update the knowledge base

See Knowledge Base Management.

Review Mode Progression

One of the most powerful aspects of Symbi is gradually increasing an agent's autonomy. We recommend this three-stage progression:

  1. Start with Always Review — every action goes through human review. Builds your baseline understanding of how the agent handles different scenarios
  2. Move to Confidence-based (X%) — the AI auto-approves actions above your chosen confidence threshold; lower-confidence ones are still queued. Saves time while keeping you in the loop on edge cases
  3. Graduate to Autonomous — once you're confident, enable full auto-approve. You can still spot-check from the Operation Center any time

You can drop back to a stricter mode at any point — for example, during a promotion week when stakes are higher.

Review mode changes are saved to the draft and take effect when you publish the agent — the same as every other field on the agent's page. Until you publish, runs keep using the review modes from the live version.

Per-Tool-Group Review for Agent-Mode Agents

Agent-mode agents support a finer level of control: each of the five action groups can be set to a different review mode independently.

| Group | What it covers | Default for a new agent | | ------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------- | ----------------------- | | Communicate | Anything the agent sends to a person — customer replies and emails, Slack messages and DMs, calendar invitations, comments on issues or pull requests | Always Review | | Lifecycle | Ticket and record status changes | Always Review | | Organize | Categorization, tagging, prioritization | Autonomous | | Internal | Internal notes and comments | Autonomous | | Produce Output | Data writes to spreadsheets, registers, and external records | Always Review |

Those are the settings a newly hired agent starts with. A common first loosening is to move Lifecycle to Confidence-based once you trust the agent's status changes, while keeping Communicate on Always Review — so you approve every outgoing message but don't need to review every label, note, or ticket close.

Configure these settings from the What it does section on the agent's page. The agent-wide review mode shown there reflects the dominant setting across applicable groups.

Iterating Through Reviews

When you edit an action during review, you're not just fixing a single reply — you're teaching the agent. With Auto-extend FAQ from support tickets enabled, approved and edited tickets are distilled into FAQ entries and added to the agent's knowledge base automatically each day — without duplicating content already in your uploaded FAQ. Over time you'll see:

  • Fewer edits needed per review cycle
  • More consistent tone and formatting
  • Better handling of edge cases the agent previously struggled with

This feedback loop is the fastest way to shape your agent's behavior over time. Toggle Auto-extend FAQ from support tickets from the What it does section on the agent's page.

What's Next?