--- title: "Agent Identity (Using identities to improve accuracy and context)" description: "Overview" canonical_url: "https://help.brainfi.sh/articles/agent-identity-using-identities-to-improve-accuracy-and-context-W0pvGP8Im0" md_url: "https://help.brainfi.sh/articles/agent-identity-using-identities-to-improve-accuracy-and-context-W0pvGP8Im0.md" --- # Agent Identity (Using identities to improve accuracy and context) ## Overview **Agent Identity** lets you configure the persona, tone, and default behavior of your Brainfish AI agent for a specific use case. Instead of a one-size-fits-all assistant, you can pick a purpose-built identity that changes how the agent reasons, responds, and interacts with your users—without requiring any prompt engineering on your part. Each identity is pre-tuned for a different job, so selecting the right one is the fastest way to improve response quality and relevance for your audience. --- ## When to use Agent Identity Use Agent Identity when you need: * More **context-aware** answers aligned to a specific domain (for example, support, security, HR). * A consistent **tone and writing style** across responses. * [Combining Agent Identity with other settings](/articles/agent-identity-using-identities-to-improve-accuracy-and-context-W0pvGP8Im0#h-combining-agent-identity-with-other-settings) Outputs tailored to a particular **audience** (end users vs. engineers). * More standardized **formats** (for example, troubleshooting guides, runbooks, KB articles). --- ## Where to find it 1. Navigate to **Agents** in the left sidebar. 2. Select the agent you want to configure. 3. Open the **Agent Settings** accordion. 4. The **Agent Identity** selector is at the top of the panel. [CleanShot 2026-03-05 at 12.20.07.mp4 3022x1722](https://help.brainfi.sh/api/attachments.redirect?id=2760c4df-d2ba-4359-a1d3-f2dd65b5dd7d) --- ## Available Identities | Identity | Best for | What the agent does | |----|----|----| | **Customer Support** | External-facing support widgets | Empathetic help desk agent that escalates when stuck, offers follow-ups, and cites sources | | **Sales Enablement** | Internal sales teams | Arms sales reps with product facts, competitive positioning, objection handling, and relevant case studies | | **Internal Support Enablement** | Support agent assist tools | Drafts concise internal replies for support agents using the knowledge base | | **Website – Lead Acquisition** | Marketing / website widgets | Engages website visitors conversationally, qualifies interest, and guides toward conversion actions | | **ITSM** | IT help desks | Structured IT troubleshooting with step-by-step diagnostics, severity awareness, and ticket-ready output | | **HR Self-Service** | Employee portals | Policy-aware HR assistant that handles sensitive employee queries with confidentiality | | **Developer / API Support** | Developer portals and docs sites | Provides code examples, API references, and precise implementation guidance | | **Customer Onboarding** | Onboarding flows and in-app help | Guided setup assistant that tracks milestones, suggests next steps, and proactively helps new users | | **Compliance / Legal** | Regulated industries and internal policy tools | Cautious, regulation-aware assistant that cites specific policies and flags when professional review is needed | | **Customer Success / Retention** | CSM tools and in-app engagement | Adoption-focused assistant that identifies churn signals, suggests underused features, and drives expansion | > **Note:** The **Website – Lead Acquisition** identity is only available on widget channels, not on other platforms (e.g. Slack, Help Center). --- ## How identities improve response quality Agent identities help improve output quality by: 1. **Reducing ambiguity**: The assistant knows what role it is expected to play. 2. **Increasing relevance**: Responses can focus on the right constraints, tools, and terminology. 3. **Improving consistency**: Repeated tasks follow predictable structure and standards. --- ## How identities affect agent behavior Selecting an identity adjusts: * **Tone and empathy** — e.g. Customer Support is warm and reassuring; ITSM is precise and procedural. * **Escalation behavior** — some identities (Customer Support, ITSM) are more likely to offer an escalation path when an answer cannot be found. * **Answer structure** — Developer / API Support prioritizes code blocks and technical accuracy; HR Self-Service prioritizes plain-language policy summaries. * **Confidence thresholds** — Compliance / Legal adds disclaimers and recommends professional review on ambiguous queries. These behaviors are applied on top of any **Custom Instructions** you add, so your specific rules are always respected. --- ## Choosing the right identity **Start with the identity that matches your primary audience:** * Helping end customers resolve issues? → **Customer Support** * Embedded in your product to help new users get started? → **Customer Onboarding** * Inside your internal support tooling to help agents draft replies? → **Internal Support Enablement** * On a public website to capture leads? → **Website – Lead Acquisition** * Inside an IT portal for employees? → **ITSM** If your use case spans multiple categories, choose the one that best fits the *majority* of queries your users ask, then use **Custom Instructions** to fine-tune edge cases. --- ## Recommended practices (especially when paired with Custom Instructions) To get the best results, define: * The agent’s **role** (for example, “technical writer”, “support engineer”). * Key **constraints** (what to include/exclude, formatting rules, compliance requirements). * The expected **document type** when applicable (help article, how-to, decision record). * The **triggering question** and any essential background context. If you are generating documentation from a workplace conversation, include: * The message that triggered the request. * The relevant thread content (questions, decisions, and the final resolution). * Any technical details such as commands, errors, settings, or file names. --- ## Combining Agent Identity with other settings Agent Identity works alongside the following settings in the **Agent Settings** panel: * **Response Style** — Controls the length and detail level of answers (Concise, Normal, Detailed). This is independent of identity. * **Custom Instructions** — Free-text instructions that are always applied, regardless of the selected identity. Use this to enforce specific terminology, escalation email addresses, or formatting rules. * **Default Fallback Message** — The message shown when no relevant answer is found. Customize this per identity if needed. * **Suggestions** — Predefined question prompts shown to users. These complement the identity by surfacing the most common queries upfront. --- ## Changing an identity after go-live You can change the Agent Identity at any time without affecting your knowledge base or other configuration. Changes take effect immediately after saving. If you switch identities and notice a change in tone or behavior that you want to preserve, use **Custom Instructions** to explicitly describe the behavior before switching. --- ## Frequently asked questions **Does the identity affect what content the agent can access?** No. Agent Identity only changes how the agent responds, not which knowledge sources it searches. Knowledge source access is controlled in the **Knowledge Settings** panel. **Can I create a custom identity?** Custom identities are not currently supported. You can approximate a custom persona by combining the closest built-in identity with detailed **Custom Instructions**. **Does switching identity reset my other settings?** No. Switching the identity only changes the agent's persona and behavior model. All other settings (custom instructions, fallback message, suggestions, theme) are preserved. **Why can't I see Website – Lead Acquisition in my Slack or Help Center channel?** This identity is designed specifically for widget experiences. It is hidden on non-widget channels to avoid misconfiguration.