How to Use Gobii: Build Your First AI Coworker Agent

Gobii is built around one idea: an AI agent that behaves like a real coworker with its own email address, phone number, and memory. For a complete overview of Gobii's capabilities and pricing plans, see our full Gobii review. Once you understand how to use Gobii, the fastest path to value is picking the right agent path, wiring in the tools your team already runs on, and pointing the agent at one recurring task instead of trying to automate everything at once.

Creating your first Gobii agent

Gobii gives you two paths to your first agent, and picking the right one matters more than any individual setting. The Template Path lets you select a pre-built worker — Talent Scout, Candidate Researcher, Lead Hunter, and similar templates — that ships with common tool integrations already configured, so you mainly customize instructions for your team. The Custom Path starts from a blank agent, where you define instructions from scratch and wire up integrations and data sources yourself. Both paths finish the same way: the agent gets its own email address, phone number, and persistent memory, then appears on your team dashboard ready to be triggered by email, API, webhook, or schedule.

For a first agent, start with a template close to your use case. It removes the guesswork on integrations and instruction wording, and you can always fork it into something custom once you see how it behaves in practice. Reserve the Custom Path for workflows that don't map to any existing template — an internal-only compliance check or a niche vendor-monitoring task, for example.

1. From the New Agent menu, choose a template that matches your workflow, or select "blank agent" for a custom build.
2. Write clear, specific instructions — what to check, where to look, and how to report back.
3. Connect the integrations the agent needs (covered next) before your first test run.
4. Trigger a manual test run, review the output, then set a schedule or webhook trigger.

Connecting integrations to your agent

An agent is only as useful as the systems it can reach. Gobii ships native integrations for Salesforce and other CRMs, Greenhouse and additional ATS platforms for recruiting, Google Sheets and Google Drive for data storage, Slack for team communication, Jira for project tracking, Zendesk for support, and PagerDuty for incident management. On top of the named integrations, webhooks work as a first-class connector — inbound webhooks wake an agent to start a task, and outbound webhooks let the agent push data or trigger actions in whatever system you point it at.

Under the hood, this works because Gobii runs agents in real, headed Chrome browsers that log into accounts, fill out forms, click buttons, and extract data the same way a person at a desktop would, backed by persistent browser profiles, proxy-aware routing, and schema-validated data extraction. Practically, that means an agent can log into your ATS, pull candidate data, and drop a clean summary into a Google Sheet without a custom API integration for every step.

Integration type	Example tools	Typical trigger
CRM & sales data	Salesforce, enrichment sources	Schedule or webhook
Recruiting	Greenhouse, other ATS platforms	Schedule or email
Team communication	Slack, email, SMS	Direct message to agent
Data & reporting	Google Sheets, Google Drive	Scheduled export
Ops & support	Jira, Zendesk, PagerDuty	Inbound webhook

Automating a real recurring task

The clearest way to learn how to use Gobii is to automate one concrete, recurring job end to end — a weekly competitor-monitoring report is a good first target because it has a fixed cadence and a clear output.

1. Start from a research-oriented template or a custom agent and give it explicit instructions: which competitor pages to check, what to flag as a change, and how to format the summary.
2. Connect the agent to Google Sheets or your reporting tool so results land in one place your team already checks.
3. Set a weekly schedule trigger so the agent runs without anyone remembering to kick it off.
4. Have the agent deliver the finished report by email — or text a short summary to the phone number tied to its own identity, since every Gobii agent gets its own email address and phone number so team members can contact it directly or receive its output the same way.
5. Review the first two or three runs closely, then let the schedule run unattended.

This same pattern extends beyond monitoring. Teams run comparable agents for recruiting — sourcing candidates from job boards, LinkedIn, and GitHub and organizing results into an ATS or spreadsheet — for sales prospecting and CRM enrichment, and for operations tasks like tracking project milestones, monitoring vendor pricing, and generating daily status reports. Pick the recurring task that already eats the most manual hours on your team, and automate that one first.

Team collaboration: shared agents and roles

Gobii agents aren't single-user tools once you're past the first build. On the Team plan, multiple people can monitor, trigger, and refine the same shared agent without deploying separate copies, and admins can publish private, organization-wide templates that team members launch from instead of starting blank. That matters for consistency: if one person tweaks an agent's instructions without review, the whole team inherits the change the next time it runs.

Access control follows the same logic. Team admins manage collaborator permissions, decide who can create or edit agents, and can set approval gates for sensitive actions, while credentials and API keys are stored centrally per integration instead of scattered across individual accounts. Billing is also pooled rather than per-agent — the Team plan allocates 1,000 task credits per seat per month, and teams share that capacity across however many agents they run. For a small team piloting Gobii, this means you don't need to negotiate separate logins or credential sharing just to let two people manage the same recruiting agent.

Recent Gobii feature releases

Gobii has shipped several template- and visibility-focused updates in the weeks leading up to this guide, worth knowing before you build. A dedicated Team Templates editor (July 7, 2026) lets admins build and launch templates directly from the New Agent menu, faster than editing an existing agent into a template after the fact. Shortly before that (June 30), Gobii added the ability to convert an organization's existing playbooks into reusable agent templates that get distributed to the whole team.

Custom Instructions, shipped June 23, let admins set persistent instructions once and apply them automatically across every new agent built from a template — useful for enforcing tone, compliance language, or reporting format without repeating it per agent. Browser File Capture (June 9) expanded what an agent can collect beyond text and structured data: agents can now capture webpage screenshots, save browser downloads, and analyze visual files. Reliability updates in the same window improved agent startup time, mobile responsiveness, and notification delivery — unglamorous but relevant if you plan to trigger agents from a phone.

Common setup mistakes to avoid

• Starting fully custom on day one. A template narrows the decision space and comes with integrations pre-wired — save the blank-agent path for workflows nothing else covers.
• Vague instructions. "Monitor competitors" is not a task; "check these five pricing pages weekly and flag any change over 5%" is.
• Skipping the test run. Trigger the agent manually and review output before attaching a schedule or webhook — catching a bad instruction after a week of unattended runs wastes the whole automation.
• Ignoring roles on a shared agent. Without approval gates on sensitive actions, one teammate's edit can silently change what a shared agent does for everyone.
• Not reading the changelog. Features like Team Templates and Custom Instructions change how you should set up new agents — build your first agent after checking what shipped recently.

Template path vs. custom path: which fits your first agent

Before you commit to a build, compare the two starting points directly. This is the core decision MOFU readers researching how to use Gobii need to make.