You can automate the four most time-consuming parts of SEO — content creation, keyword research, link-building outreach, and website generation — entirely for free in 2026 by connecting open-source AI agents to a chat-based workspace. The key insight is that you don't need a paid SaaS platform to do this; you need an agent-agnostic chat surface (where agents participate as channel members), a persistent memory layer, and the APIs you already pay for or can use for free. This guide walks through four verified workflows, the exact tools and APIs behind each one, and where the human still needs to stay in the loop.
Last verified: 2026-08-02 · Primary keyword: AI SEO automation · Format: How-to guide
- All four workflows can run on a $0 software stack (open-source agents + free-tier APIs)
- Buzz (Block, Apache-2.0) is the chat workspace; Hermes Agent, Claude Code, Codex, and Goose are the engines
- Google Search Console + Obsidian provide the data layer for keyword research
- Hunter API (50 free credits/month) handles domain verification for outreach
- Buzz is in beta and "still in its early stages" — expect bugs and non-replies
What Is AI SEO Automation and Why Does It Matter in 2026?
AI SEO automation is the practice of delegating repetitive search optimization tasks — keyword research, content drafting, link-building prospecting, technical audits, and even site scaffolding — to AI agents that can execute multi-step workflows against live data without a human prompting every step. In 2026, the shift is from "assisted creation" (you prompt, the AI writes) to "autonomous execution" (the agent monitors, decides, and acts on your behalf). AI Overviews now appear on roughly 48% of Google queries and cut clicks to the #1 organic result by more than half, so the game is no longer just ranking — it's being cited by AI answer engines. Automation lets a one-person operation compete with a full SEO team by running 24/7 on a free stack.
How Do You Set Up a Free AI SEO Automation Stack?
The stack has three layers: a chat workspace where agents live as channel members, an agent engine that does the reasoning and tool execution, and a data/memory layer that gives the agent your real search data.
Layer 1: Chat workspace — Buzz (free, open-source)
Buzz is a free, open-source team chat app from Block (Jack Dorsey's company, also behind Square and Cash App), launched on July 21, 2026. It looks like Slack, but AI agents live in the channels as full participants — you can @-mention them, assign tasks, and they reply and execute directly in the conversation. It's built on the decentralized Nostr protocol, is model-agnostic, and the code is on GitHub under the Apache-2.0 license. You can self-host or use Block's hosting. The desktop app runs on macOS, Windows, and Linux.
Buzz connects agents through buzz-acp, a harness that translates between the Agent Client Protocol (ACP) and the Model Context Protocol (MCP). This means any ACP-compatible agent — Block's own Goose, Anthropic's Claude Code, or OpenAI's Codex — can join your channels. A buzz-cli tool also gives agents a JSON command-line interface. The result: your agent's accumulated context (channels, project decisions, files, Git events) stays in one open workspace even if you swap the underlying engine from Claude to Codex to Goose.
Layer 2: Agent engine — choose your model
Buzz is model-agnostic, so you bring your own agent. The strongest free options:
| Agent | Maker | License | Best For | Source |
|---|---|---|---|---|
| Hermes Agent | Nous Research | MIT (open-source) | General-purpose automation, SEO workflows, scripting, cron | GitHub · Official site |
| Claude Code | Anthropic | Proprietary (bring your own API key) | Deep coding, multi-file refactors, code review | Anthropic |
| Codex CLI | OpenAI | Proprietary | Code generation, app scaffolding | OpenAI |
| Goose | Block | Apache-2.0 (open-source) | Native Buzz integration, dev workflows | Block blog |
Hermes Agent is particularly cost-effective for SEO automation because it can use free model tiers (e.g., Gemini Flash or free-community APIs) and includes a built-in cron scheduler for unattended tasks. One community member ran a 24/7 agent on a Mac Mini for two months for $21/month total. Hermes also supports connecting to Buzz via its gateway system, so you can operate Buzz channels from your Hermes terminal or messaging platform.
Layer 3: Data and memory — Obsidian + Google Search Console
- Obsidian (free personal knowledge base): acts as your agent's persistent memory. Notes, research, and workflow outputs are stored as markdown files the agent can read and search.
- Google Search Console (free): the only free tool that shows the actual queries Google already ranks you for, with real clicks, impressions, position, and CTR. Your agent can pull striking-distance keywords (positions 5–20) directly from the Search Console API — no guessing, no estimated volumes.
- Hunter API (free tier: 50 credits/month): for link-building outreach, the Domain Search endpoint finds professional email addresses from any company website. The free tier gives you 50 credits/month with no credit card required.
Workflow 1: How to Automate Content Creation and Publishing
Goal: Have an agent write, format, and deploy a blog post from a single chat message.
Steps:
- Create a dedicated content channel in Buzz (delete the default channels first — you want purpose-built channels per workflow).
- @-mention your agent with a task like: "Create and deploy a blog post about [topic], tagged [category], on our content channel."
- The agent uses your connected APIs — website CMS, image generation, deployment pipeline — to draft the article, generate a cover, and publish it directly to your live site.
- Review the deployed post in the channel. The agent posts the live URL and a summary.
Why this works: The agent already has access to your publishing APIs (plugged in via the agent's tool configuration), your brand voice (stored in Obsidian memory), and your existing content structure. You're not prompting from scratch each time — the agent pulls context from your memory layer and existing deployments.
Where the human stays: Final review of the published post. Agent-generated content in 2026 can carry factual errors or "AI slop" patterns. Always verify claims against primary sources before leaving a post live. Google's February 2026 core update explicitly rewards content that adds information gain — original synthesis, testing, or data — and demotes commodity rehash.
Workflow 2: How to Automate Keyword Research Using Your Own Data
Goal: Generate a prioritized list of keywords to target, based on what's already working for your site.
Steps:
- Connect Google Search Console API to your agent (this is a standard API with free access).
- Point your Obsidian vault into the agent's memory so it can read your existing notes, research files, and workflow outputs.
- Ask the agent: "Look through my Obsidian vault and Google Search Console data and find new keywords for me to rank for."
- The agent pulls fresh Search Console data (impressions, clicks, position for every query), cross-references it with your Obsidian notes to identify topics you haven't covered yet, and produces a markdown report with the top articles to publish next.
What you get: A data-driven keyword report that surfaces:
- Striking-distance keywords: queries where you rank positions 5–20 with decent impressions — optimizing these pages can push them to page 1.
- Content gaps: topics in your niche that competitors cover but you don't.
- Predictive keywords: terms emerging from your existing content workflows and saved research that haven't been targeted yet.
Why this is better than a keyword tool: Third-party keyword tools show estimated search volumes. Search Console shows you the actual queries real people typed to find your site. That's your own conversion-relevant data, not a database estimate. For more on using Search Console for keyword research, see our guide to Google Search Console platform properties.
Workflow 3: How to Automate Link-Building Outreach
Goal: Have an agent prospect leads, verify domains, and draft outreach emails — without you touching a spreadsheet.
Steps:
- Give your agent Google Workspace API access (so it can send and receive email via Gmail).
- Give it Hunter API access (free tier: 50 credits/month) for domain search and email verification.
- Ask the agent: "Prospect a list of leads for SEO outreach for an article about [topic]. Save the list to a markdown file."
- The agent uses Hunter's Domain Search endpoint to find email addresses at relevant websites, verifies each domain is live, and tiers the prospects by relevance (e.g., tier 1: direct competitors; tier 2: adjacent industry blogs; tier 3: aggregators).
- Approve the prospect list, then ask the agent to draft and send personalized outreach emails via the Gmail API.
Hunter API cost breakdown (as of 2026-08-02): | Endpoint | Credits per call | Free tier covers | |----------|-----------------|------------------| | Domain Search | 1 credit per 1–10 emails per domain | ~50 domain searches/month | | Email Finder | 1 credit per email found | ~50 emails/month | | Email Verifier | 0.5 credits per verification | ~100 verifications/month |
Where the human stays: Approve the prospect list before any emails are sent. Cold outreach that goes to the wrong people damages your domain's sender reputation and can tank deliverability for your entire organization. Always verify the email list quality and personalize the pitch yourself — generic outreach gets ignored and flagged as spam.
Workflow 4: How to Build and Deploy a Website From Scratch With an Agent
Goal: Generate a complete, live website — design, content, and deployment — from a single prompt.
Steps:
- In your Buzz channel, @-mention the agent: "Build a beautiful website for a blog about [topic]. Give me the URL to preview it. Deploy to a test domain on Netlify."
- The agent scaffolds the site (HTML/CSS/JS or a framework), writes 4–5 blog posts with topic-relevant content, and creates a working design.
- The agent deploys the site to Netlify (free tier: 100 GB bandwidth, free SSL, automatic HTTPS, continuous deployment from Git — no credit card required) and returns the live preview URL.
- Review the live site at the Netlify URL. If it looks good, ask the agent to publish to your custom domain.
Why this works: Netlify's free tier supports drag-and-drop deployment and Git-based continuous deployment. The agent generates the site files, pushes them to a Git repo or uploads them directly, and Netlify handles the CDN, SSL, and global distribution. The result is not a generic "AI slop" template — if the agent has your Obsidian vault as context, it can generate a site with personalized content about your actual experiments and projects.
Where the human stays: Design review and brand alignment. The agent can generate a structurally sound site, but visual taste, brand voice, and user experience decisions still need a human eye. For more on building agent-driven systems, see how to build a free AI agent operating system in 2026.
How Does Buzz Compare to Other AI SEO Automation Approaches?
| Feature | Buzz + open-source agents | Paid SEO SaaS (Semrush, Ahrefs, etc.) | Slack + custom bots |
|---|---|---|---|
| Cost | $0 software (you bring LLM API costs) | $100–$500+/month per seat | $7–$15/seat/month + bot dev costs |
| Open source | Yes (Apache-2.0) | No | Slack: no. Bots: varies |
| Agents as full participants | Yes — agents live in channels, can be @-mentioned, @-tag teams | No — agents are API calls or dashboards | Bots respond to commands; not full collaborators |
| Model lock-in | None — swap Claude, Codex, Goose, Hermes anytime | Locked to vendor's model | Locked to whatever you built |
| Memory / data | Your own (Obsidian, Search Console, any API) | Vendor's database | Your own (if you built it) |
| Self-hosting | Yes — full data sovereignty | No | No |
| Maturity | Beta — "early stages," may be buggy | Production-grade | Depends on your build |
The trade-off is clear: Buzz + open-source agents gives you maximum flexibility and zero software cost, but requires technical setup and tolerates beta-level bugs. Paid SaaS gives you polish and reliability but locks you into a vendor's ecosystem and pricing. If you want to understand the broader landscape of agent operating systems before committing, see our decision framework for AI agent operating systems in 2026.
What Are the Limitations and Risks of AI SEO Automation?
Buzz is in beta. Block states the platform is "still in its early stages." Agents may not reply, workflows may need to be chased, and features may change. Early testers describe it as "very early" and "pretty technical to set up." This is a tool for builders willing to debug, not for a team that needs turnkey reliability.
Agent quality varies by task. Content generation can produce factual errors or AI-slop patterns. Keyword research is only as good as your Search Console data (which shows impressions, not search volume). Link-building outreach can hit the wrong people if you don't verify. Website generation can produce structurally sound but visually generic sites. For specific guidance on spotting and fixing AI slop in generated content, see our framework for subjective quality in AI-generated content.
Google's February 2026 core update penalizes commodity rehash. If your agent just rewrites what's already ranking, Google will demote it. Every piece of agent-generated content needs to add genuine information gain — an original synthesis, a test, a comparison, or a verdict that isn't available elsewhere. The agent is a research and drafting assistant, not a replacement for editorial judgment.
Cost is not zero — it's "bring your own model." Buzz is free. Obsidian is free. Search Console is free. But the LLM that powers your agent costs money unless you use free-tier models. Hermes Agent can run on free community APIs (e.g., Gemini Flash), keeping costs near zero, but Claude and Codex require paid API keys. Budget accordingly. For strategies on cutting agent token costs, see 4 free open-source tools to cut Claude Code tokens by up to 80%.
What This Means for You
If you're a solo founder or small business owner: You now have access to the same SEO automation capabilities that used to require a $500/month SaaS stack and a content team. Start with Hermes Agent + Buzz + Search Console API — all free — and automate keyword research and content drafting first. Those are the highest-leverage, lowest-risk workflows.
If you're an SEO professional or agency: Buzz is a bet on open, agent-native collaboration. Your clients' context (Search Console data, keyword maps, content briefs) stays in your Obsidian vault, not a vendor's database. You can swap the underlying model as new ones launch without losing your accumulated context. But wait for Buzz to mature past beta before porting a production team.
If you're a developer or builder: The ACP + MCP architecture behind Buzz is the interesting part. You can build agents that work across any ACP-compatible harness, and your agent's context becomes portable rather than trapped in a single tool. The buzz-cli tool gives agents a JSON command-line interface, and the gateway pattern means you can operate Buzz from any Hermes Agent terminal session. For more on building reusable agent skills rather than one-off implementations, see how to build AI agent skills instead of collecting them.
Related reading
FAQ
Q: Is Buzz free to use? A: Yes. Buzz is free and open-source under the Apache-2.0 license. The desktop app is available for macOS, Windows, and Linux, and the full source code is on GitHub. You can self-host or use Block's hosting. The only costs are your own servers (if self-hosting) and whatever LLM API you point the agents at, since Buzz is model-agnostic.
Q: Do I need to know how to code to use AI SEO automation? A: Not for the core workflows. Buzz is designed to be as intuitive as Slack — you @-mention an agent in a channel and give it a task in plain English. The setup (connecting APIs, configuring the agent, pointing it at your Obsidian vault) does require some technical comfort, but the day-to-day operation is conversational.
Q: Can I use free AI models with this stack? A: Yes. Hermes Agent is open-source and can use free model tiers like Gemini Flash or free community API endpoints, keeping your per-run cost near zero. Buzz itself doesn't charge for model usage — you bring your own. If you prefer Claude or Codex, you'll need paid API keys for those providers.
Q: What is the best free keyword research tool for AI SEO automation? A: Google Search Console is the best free keyword research tool because it shows the actual queries real people used to find your site — not estimated volumes. It's free, it's directly from Google, and no third-party tool gives you the same insight into how Google indexes your pages. Pair it with your Obsidian vault as a memory layer, and your agent can cross-reference your existing research with live search data.
Q: How many outreach emails can I send with the free Hunter tier? A: Hunter's free tier includes 50 credits per month. Domain Search costs 1 credit per 1–10 emails found per domain, so you can search roughly 50 domains per month. Email Verifier costs 0.5 credits per verification, giving you about 100 verifications. This is sufficient for a small-scale link-building campaign (one or two articles per month), but you'll need a paid plan for higher volume.
Q: Can an AI agent really build a full website from a single prompt? A: Yes, with caveats. In verified demonstrations, agents have scaffolded complete websites with 4–5 blog posts, custom styling, and live deployment to Netlify from a single chat message. The output is structurally sound and deployable. However, visual design quality, brand alignment, and user experience still benefit from human review. The agent generates and deploys; you review and refine.
Q: Is Buzz safe to use for client work? A: Buzz is in beta and described by early testers as "very early" and "pretty technical to set up." Block says the platform is "still in its early stages." For client work requiring reliability, wait for Buzz to mature or use it for internal automation first. The open-source, self-hostable nature does give you full data sovereignty — your data stays on your servers, not a vendor's cloud.
Discussion
0 comments