TorqueX.ai torqueX.ai
// BLOG PLAYBOOKS

Playbooks from the agents.

The exact best practices our agents follow — starting with SEO & GEO. More going up as we ship.

Engineering 2026-08-09 · 5 min read

Cutting AI-agent token costs — without a proxy you don’t need yet

How to actually cut LLM token costs for a tool-calling agent: trim what you resend, context compression (e.g. caveman.so), provider prompt caching, and model routing — plus why we meter Ask TorqueX at pass-through Cerebras pricing ($1 = 1 credit, no markup).

Read the playbook
Engineering 2026-08-08 · 5 min read

Why we didn’t self-host our LLM on RunPod or Vast.ai

Our AI agent was replying in ~18s and the advice was to self-host the model on rented GPUs. We ran the math and didn’t — a latency and cost breakdown of self-hosting vs fast hosted models (Gemini Flash, Claude Haiku, Groq, Cerebras) for an interactive agent product.

Read the playbook
Concept 2026-08-07 · 6 min read

What is an Agent Orchestrator?

An Agent Orchestrator runs a team of specialised AI agents — outreach, social, content, support — with shared memory and human approval, so one person gets the output of twenty. What it is, how it differs from a chatbot or a stack of tools, and how torqueX orchestrates.

Read the playbook
Workspaces 2026-08-07 · 5 min read

Personal vs Company workspaces: shared agent memory

How torqueX workspaces work: a Personal workspace keeps your agents’ memory private; a Company workspace shares memory across your whole team, so every agent and teammate works from the same context — and the agents get smarter as it grows.

Read the playbook
Outreach Agent 2026-08-07 · 7 min read

How our Outreach Agent books meetings — the end-to-end pipeline

From buying domains and warming mailboxes to sourcing 275M+ contacts, personalised email + LinkedIn + phone sequences, reading every reply, and booking the meeting on your calendar — the full Outreach Agent pipeline.

Read the playbook
Outreach Agent 2026-08-07 · 6 min read

Cold email that doesn’t land in spam — a deliverability playbook

Domains, SPF/DKIM/DMARC, mailbox warm-up, sending limits, plain-text personalisation, list hygiene and monitoring — the deliverability best practices that keep cold email in the inbox.

Read the playbook
SEO Agent 2026-08-07 · 8 min read

The SEO & GEO Playbook — get found in Google and recommended by AI

The best practices behind our SEO Agent: Google Search Console, getting recommended by ChatGPT / Claude / Perplexity / Gemini, allowing AI bots (GPTBot, ClaudeBot, PerplexityBot, CCBot, Google-Extended), and presence on directories, Product Hunt, Reddit, listicles and Crunchbase.

Read the playbook