Skip to main content
Every agent, every skill, every email, every piece of content Ultron creates uses your profile. It’s the single highest-leverage configuration in the product.

​
What’s in your profile

What you build and who you sell to. One or two sentences. Used in cold emails, meeting prep briefs, and content to position correctly.Example: “We build AI-native workflow automation for B2B SaaS operations teams.”
Your Ideal Customer Profile. The more specific, the better.
  • Company size: e.g., 50–500 employees
  • Industry: e.g., B2B SaaS
  • Funding stage: e.g., Series A or B
  • Geography: e.g., US and Canada
  • Additional signals: e.g., teams with an ops lead, companies using Salesforce
Specter uses this to score every lead 1–10. Striker uses it to personalize every email.
List the products you’re most often compared against. Cortex uses this for competitive research. Striker uses it for objection handling. Pulse uses it for content angles.Example: HubSpot, Salesforce, a homegrown spreadsheet system.
3–5 words that describe how you communicate.Example: “direct, technical, founder-to-founder, no filler”Used by Pulse for content, Striker for emails, and any skill that produces written output.
Paste 2–3 real examples of writing you’re proud of — a cold email that got replies, a LinkedIn post that performed, a message that landed.This is the most impactful input in the entire profile. Ultron learns your actual writing patterns from examples, not from a description of them.
Which channels you actively use: LinkedIn, Twitter, email, blog, Instagram.Pulse uses this to know where to create content. Striker knows which channel to use for outreach.
Terms that should never appear in any Ultron output.The defaults already exclude: synergy, leverage, innovative, revolutionary, game-changing, seamlessly. Add your own specific terms.

​
How to update your profile

  1. Go to Settings → Profile
  2. Fill in each section
  3. Save
Changes take effect in your next session. There’s no retraining required — the profile is injected into context at session start.
Add voice samples first. It’s the one change that produces an immediately noticeable improvement in output quality.