Most guides about building AI agents assume you have a Python environment, know what a vector database is, and enjoy reading API documentation at 11pm. This one doesn’t.
This is a guide for people who want an AI agent that does something useful — monitors a GitHub repo, triages support tickets, tracks revenue, posts to Slack — without hiring a developer or spending a weekend on setup.
Here’s how to build one in under 10 minutes.
What Is an AI Agent, Actually?
Before we build one, let’s be precise. An AI agent is not a chatbot. A chatbot responds when you talk to it. An agent runs in the background, takes actions on your behalf, and produces outputs that do things.
Chatbot: You ask “how are my GitHub PRs looking?” and get a text answer. You then go manually check GitHub.
Agent: Every morning at 9am, the agent checks your GitHub repos, finds open PRs, identifies which ones have been waiting more than 24 hours, and posts a structured triage summary to your Slack channel. You never asked. It just ran.
The three things that make something an agent rather than a chatbot:
- It can take actions — not just generate text, but actually call APIs, read databases, send messages
- It runs without you triggering it — scheduled or event-driven, not just reactive
- It produces outputs that persist — a dashboard, a Slack post, a created ticket — not just a response that disappears when you close the tab
What You Need to Get Started
Nothing except a Shogo account, which is free.
You don’t need:
- Python or any programming language
- API keys (Shogo handles authentication via OAuth)
- A server or cloud infrastructure
- Any technical background
You do need:
- A specific workflow you want to automate (more on choosing this in a minute)
- The tools that workflow touches (GitHub, Slack, HubSpot, Stripe — whatever’s relevant)
- 10 minutes
Step 1: Pick the Right First Workflow
The biggest mistake beginners make is starting with something too ambitious. “Automate my entire sales process” is a project. “Post a Slack message when a deal has had no activity for 7 days” is an agent.
Good first agent criteria:
- It’s something you do manually right now — not a dream workflow, a real one
- It involves at least two tools — agents are most useful when connecting things
- It’s repetitive — happens daily, weekly, or whenever a specific event occurs
- The inputs and outputs are clear — “look at X, do Y when Z” is the structure you want
Great first agents by role:
| Role | Good first agent |
|---|---|
| Engineering | ”Post a daily GitHub triage summary to #eng-standup” |
| Sales | ”Alert me in Slack when a HubSpot deal hasn’t moved in 7 days” |
| Marketing | ”Summarize 5 industry articles and drop them in Notion every Monday” |
| Support | ”Escalate Zendesk tickets tagged urgent to #support-escalations” |
| Founder | ”Pull my Stripe MRR and churn into a weekly email every Sunday” |
Pick one of these or something similar. Specific is better than broad.
Step 2: Sign Up and Open Shogo Studio
Go to studio.shogo.ai and create a free account. No credit card required.
The free tier gives you 100 credits per day — enough to run most personal and small team automations without spending anything.
When you land in Studio, you’ll see the agent builder. The core of it is a text prompt: describe what you want your agent to do.
Step 3: Describe Your Agent in Plain English
This is where most people expect it to get technical. It doesn’t.
Click “New Agent” and type what you want. Write it the same way you’d explain it to a smart intern:
“Every morning at 9am, check my GitHub repos for any pull requests that have been open for more than 24 hours without a review. List them by repo, include the PR title, author, and how many hours it’s been open. Post the list to my #engineering Slack channel.”
That’s it. That’s your agent specification.
Shogo will interpret this, identify the integrations it needs (GitHub + Slack), and scaffold the agent for you.
Tips for writing good agent descriptions:
- Include the trigger — “every morning at 9am” or “whenever a new Zendesk ticket arrives”
- Be specific about the output — “a list of PRs grouped by repo” is better than “a summary”
- Name the tools — “my HubSpot pipeline” tells it which integration to use
- Include any conditions — “only if it’s been more than 24 hours” or “only for P1 tickets”
You can also start from a template if you prefer to modify an existing agent rather than describing one from scratch. Templates are pre-configured agents you clone and customize — they’re usually faster for common workflows.
Step 4: Connect Your Integrations
After you describe the agent, Shogo will identify which integrations it needs and ask you to authorize them.
Each connection is OAuth-based — you click “Connect,” authenticate with the service (same as signing in with Google), and Shogo gets the scoped access it needs. No copying API keys, no reading documentation.
For the GitHub + Slack example:
- Click “Connect GitHub” → authorize with your GitHub account
- Click “Connect Slack” → authorize with your Slack workspace
- Done
If you’re using a template, the integrations are already identified for you. You just click to authorize each one.
Common integrations available:
- GitHub, GitLab (code + project management)
- HubSpot, Salesforce (CRM)
- Slack, Discord (communication)
- Zendesk, Intercom (support)
- Stripe (billing)
- Google Calendar, Notion, Airtable, Linear, Jira
- Gmail, Google Sheets, Google Drive
- Brave Search, Exa (web research)
- PostgreSQL, MongoDB, SQLite (databases)
Step 5: Test Before You Deploy
Before setting your agent loose on a live schedule, run it manually once to see what it produces.
In Shogo Studio, every agent has a “Run Now” button that executes it immediately so you can see the output. This is where you catch mismatches between what you described and what you wanted.
Common things to verify on a test run:
- Does it pull from the right repos/accounts/deals?
- Is the output format readable and complete?
- Is it posting to the right Slack channel?
- Does it include or exclude the right things?
If something’s off, go back to your description and be more specific. “Only include PRs from my main repo, not forks” or “format each deal as a single line with name, stage, and days since last activity.”
Iteration is normal. Most agents need 1–2 rounds of refinement before they’re dialed in.
Step 6: Set the Schedule and Deploy
Once the test output looks right, set your trigger:
- Scheduled: “Every weekday at 8:45am,” “Every Sunday at 6pm,” “On the first of each month”
- Event-triggered: “Whenever a new Zendesk ticket arrives,” “When a GitHub PR is opened”
- On demand: Run manually via the Shogo dashboard
Click Deploy. The agent is live.
For scheduled agents, you’ll see it next to every other running agent in your dashboard with its last run time and next scheduled run. You can pause, edit, or delete it at any time.
What to Build Next
Once your first agent is running, you’ll start to notice every other manual process that’s eating your time. That’s the pattern.
The most commonly deployed second agents by teams using Shogo:
- If you built a GitHub triage agent → build a Standup Generator that reads the same data and writes per-engineer summaries
- If you built a sales pipeline monitor → add an Inbound Lead Qualifier that routes new leads in real time
- If you built a revenue tracker → add an Invoice Manager that handles billing reminders
- If you built a knowledge base bot → extend it with a Meeting Prep agent that pulls Notion context before calls
Frequently Asked Questions
How much does it cost?
Free tier is 100 credits/day — enough to run daily agent schedules without paying anything. Paid plans start at $19/month for higher volume.
Is this actually no-code or do I eventually have to write code?
You can go as deep as you want without code. Some power users write custom JavaScript inside agents for advanced transformations, but it’s entirely optional. The vast majority of workflows are configured through natural language.
What if I want to do something that isn’t covered by a pre-built integration?
Shogo supports custom webhooks and HTTP requests, so any service with an API can be connected. This does require slightly more configuration but still no traditional coding.
Can I use this for my whole team?
Yes. Team plans let multiple people share agent workspaces, collaborate on configurations, and see each other’s running agents. Starting at $149/month.
What if the agent gets something wrong?
Every agent has a run log showing exactly what it did, what it read, and what it sent. If something goes wrong, you can see why and adjust the description or rules.
Start Building
The best way to understand what AI agents can do is to run one.
Browse templates — or go to Shogo Studio and describe your first workflow right now. Free tier, no credit card, first agent in 10 minutes.