AI-agent GTM stack with Claude Code
Researched by the Tentt team · draft awaiting expert review · Updated
An AI agent GTM stack uses Claude Code to build and run workflows, Clay's official plugin and API for data, n8n for scheduled glue, OpenAI, Perplexity and Exa for research, Instantly to send and HubSpot as the CRM. At published October 2026 prices, it costs about $500 to $750 a month for one operator.
Who is this stack for?
Technical GTM engineers comfortable in a terminal who want an agent to write and run their enrichment, research and routing workflows, with Clay still doing the data work.
Which tools are in the stack, and what does each one do?
01Agent and build layer
Claude Code
Writes and runs the scripts, calls tools through MCP servers and drives Clay through Clay's official agent plugin and the clay CLI.
Pro $20/month or $17/month on annual billing; Max 5x $100/month for heavier daily use
02Data and orchestration
ClayRuns searches, enrichment waterfalls and tables that the agent triggers; webhook tables receive rows from n8n and HTTP API columns call back out.
Growth from $495/month (monthly) or $446/month (annual); Growth is the first plan with webhooks and the HTTP API
n8nRuns scheduled and triggered workflows between tools, posting JSON to Clay webhook tables and receiving calls from Clay HTTP columns.
Free self-hosted Community edition (you pay for hosting), or Cloud Starter €20/month billed annually for 2,500 executions
03AI research
OpenAI
Cheap classification and copy drafting inside n8n or Clay on your own key, billed per token.
Per token: gpt-6-luna $0.10 / $0.50, gpt-6.1-sol $2.00 / $10.00 per 1M input / output tokens
PerplexityWeb search with ranked results for account research, called from n8n or scripts through the Search API.
Search API $5 per 1,000 requests (standard) or $1 per 1,000 (fast), pay as you go
ExaFinds lists of similar companies and pages from a natural-language query, from Claude Code via MCP or from Clay's Exa action.
Free tier ($10 credits a month), then search from $7 per 1,000 requests (Fast/Auto)
04Sending
InstantlySends sequences from warmed-up mailboxes; Clay's Add Lead to Campaign action or the Instantly API adds contacts.
Growth $47/month or $37.6/month on annual billing, 5,000 emails a month
05CRM
HubSpot
System of record; Claude Code can reach it through HubSpot's MCP server, and Clay and Instantly both have native HubSpot integrations.
Free tier (up to 2 users)
How does data flow through the stack?
- 1Write the ICP and the workflow as a spec in the repository, then have Claude Code scaffold the n8n workflows and Clay tables it describes.
- 2Use Exa from Claude Code to find lookalike companies, and send the list into a Clay webhook table.
- 3Clay enriches people and emails; n8n calls Perplexity and OpenAI to research each account and draft a first line.
- 4Clay's Instantly action adds approved contacts to an existing campaign.
- 5Replies and contacts sync to HubSpot through Instantly's native integration, and Claude Code reports results from HubSpot over MCP.
How much does this stack cost per month?
About $500 to $750 a month for one operator
Assumes one operator, 1,000 to 5,000 researched accounts a month and HubSpot Free. Low end: Claude Pro $17 (annual) + Clay Growth $446 (annual) + n8n self-hosted $0 (hosting excluded) + OpenAI about $2 + Perplexity Search 1,000 requests $5 + Exa 1,000 Fast searches $7, covered by the $10 free credit + Instantly Growth $37.6 (annual) = about $508. High end: Claude Max 5x $100 + Clay $495 (monthly) + n8n Starter €20 + OpenAI about $35 + Perplexity 5,000 requests $25 + Exa 5,000 searches $35 less $10 free = $25 + Instantly $47 (monthly) = $727 plus €20, about $750. The OpenAI figures are estimates from the per-token prices, assuming 5,000 calls at roughly 2,000 input and 300 output tokens (gpt-6-luna low, gpt-6.1-sol high). Prices checked on 17 October 2026.
What are the trade-offs?
- Claude Code subscription usage is shared with Claude chat and resets on a rolling five-hour window with weekly limits, so long unattended runs can hit limits; API billing avoids this but is metered.
- Every layer is an API, so this stack needs someone who can read code, debug a failed webhook and manage keys safely. It is not a no-code setup.
- Perplexity ended Sonar Chat Completions support on 27 September 2026 and is moving requests to its Agent API, so older Perplexity integrations may need rewriting.
- Clay Growth is the largest line. Without Clay webhooks and the HTTP API, n8n and Clay can only exchange data by CSV or through a sequencer.
Questions about this stack
Can Claude Code run Clay?
Yes. Clay publishes an official agent plugin for Claude Code, Codex and Cursor that installs the clay CLI, signs you in and gives the agent Clay's skills. Clay also offers a Public API authenticated with a clay-api-key header.
Do I need n8n if I use Claude Code?
n8n is useful for anything that must run on a trigger or schedule without you at the terminal, such as a nightly sync or a webhook handler. Claude Code can also run headless with claude -p and on schedules, so a small team can start without n8n and add it when workflows need to run continuously.
Which is cheaper for account research, Perplexity or Exa?
Perplexity's Search API is $5 per 1,000 standard requests, and Exa's search starts at $7 per 1,000 Fast or Auto requests after a $10 monthly free credit. They do different jobs: Perplexity suits questions about one account, Exa suits finding lists of similar companies.
Is Claude Code included in the free Claude plan?
No. The Claude Free plan does not include Claude Code; it starts with Pro at $20 a month, or $17 a month on annual billing. You can also run Claude Code on a Console account billed per API token.
What does an AI agent add to a GTM stack?
An AI agent writes and operates the workflows themselves, so changes to a table, a script or a routing rule can be made from a written instruction. The data still comes from providers such as Clay, and the agent's output needs review before it reaches prospects.
Other GTM stacks
- Clay-centred enrichment stackA Clay enrichment stack on Growth: Prospeo, LeadMagic and FullEnrich waterfall, ZeroBounce, OpenAI and Perplexity research, and CRM sync, with costs.
- Enterprise outbound stackAn enterprise outbound stack: ZoomInfo, Bombora, Clay, ZeroBounce, Outreach or Salesloft and Salesforce, with published costs and what is contact sales.
- Lean cold email stack for small teamsA lean cold email stack for small teams: Prospeo, MillionVerifier, Smartlead and a free CRM, with verification, warm-up and costs from about $70 a month.
- LinkedIn and email multichannel stackA LinkedIn and email outreach stack: HeyReach with Smartlead, or La Growth Machine, plus Prospeo, Surfe and HubSpot, with published costs per sender.
- Signal-based outbound stackA signal-based outbound stack: RB2B and PredictLeads signals, Clay, a Prospeo and Findymail waterfall, MillionVerifier, Smartlead and HubSpot, with costs.
Want a GTM engineering system built for you?
Tentt builds signal-driven outbound, enrichment and CRM systems into B2B companies' stacks. Or start free with the resources from every video.
Sources
Checked on 17 October 2026.
