AI Simulator

Run an AI simulation

1. Open AI Simulator in the AI Email Marketing menu and click New AI simulation.
2. Basics: name the simulation, and start from one of your campaigns to load its emails, or write new ones.
3. Offer and market: what kind of offer it is (pick Something else to type your own), what you sell, the specific pain you solve, the outcome with a number, your proof, competitors or alternatives, what makes you different, your price, why they should act now and the month you will send.
4. Audience: the buyers' industry, the niche in one sentence, job titles, department, seniority, company sizes, regions, the tools they use, their buying stage, how well the list fits your best customers and where it comes from. These start from your AI Training.
5. Pains and objections: the problems they have and the objections you hear.
6. Email copy: how personal the emails are and what the personal line is based on, then the first email and up to four follow-ups. Turn on Test a version B to compare two sequences.
7. Sending setup: domain authentication, warmup, how many mailboxes send and how many emails each sends a day, the sending domain and tracking.
8. Size and cost: test on 10,000, 50,000, 100,000 or 1,000,000 AI agents. The cost and your balance show here. Click Run simulation.

Every field is required

• Next stays blocked until the step is complete, and each missing answer is marked under its field. Write "None" where nothing applies.
• A follow-up you add needs its email and how many days it waits. Follow-ups reply in the same thread.
• Nothing is emailed. The AI Simulator never uses your mailboxes or your campaigns.

Why the numbers are conservative

• Results are judged against your market. A generic offer in a crowded market, such as a general SEO or lead generation agency, scores very badly, however good the emails look.
• Only a specific offer can do well in a crowded market: a named niche, a named pain, an outcome with a number and a timeframe, peer proof with a number and a reason to act now.
• The best cold email gets 2.5% to 3% replies, 0.05% to 0.09% of contacted prospects interested, and meetings with 40% of the interested. Only an excellent niche campaign in a quiet market comes close.
• The page says which market your offer was treated as, and why. If your words describe a more crowded market than the one you chose, that market is used.
• Placeholder copy is not an email: a first email of a few words, "test", lorem ipsum, random letters, no sentence that says what you offer, or a one-word subject such as "Hello" gets no replies and the verdict Not a real email.
• It is a prediction, not a guarantee. Use it to compare versions and find what to fix before you send.

Read the results

• Each simulation has its own page. While it runs, the page shows its progress; you can leave and come back. Large tests take a few minutes.
• The top shows the verdict with a campaign score out of 100 (50 is the market median), your market and its competitiveness, and the size you tested on, for example Tested on 10,000 AI agents.
• Reply rate, interested, meetings and unsubscribes each show a likely range and how far they are above or below the industry median.
• Charts show the funnel from agents to meetings and where the emails landed, you against the market median, top quarter and top tenth, where your replies go from an excellent campaign to yours, your score by area, why prospects pass, who replies by title and industry (a heatmap), seniority, size and region, who the agents were, how likely each agent is to reply and replies by email.
• Weak areas and fixes ranks the changes to your pitch, offer, value proposition, targeting, copy and setup by how many replies and interested prospects each would add per 10,000.
• A vs B compares the two sequences side by side and says which is likely better. Method and confidence is folded at the bottom, and the copy tested is below it.
• The page shows no AI-written replies, only numbers, reasons and fixes.
• Download PDF, at the top right, opens the full report with every chart, ready to print: choose Save as PDF. Download CSV saves the numbers in the report for a spreadsheet.
• Previous simulations on the AI Simulator page lists every run with its date, size, reply rate, interested, meetings, status and credits used. Click a row to open it.

Improve and launch

• Generate improved copy rewrites the sequence to fix what the result found. Your current copy becomes A and the rewrite B in a new simulation, ready to run.
• Re-run repeats the simulation. Edit and run opens the wizard with the same answers.
• Create draft campaign puts the tested sequence into a draft campaign. Nothing is sent until you add recipients and launch it.

Credits

• A simulation costs 2 credits per AI agent: 10,000 agents is 20,000 credits and 1,000,000 is 2,000,000. Testing a version B does not change the price.
• The last step shows the cost and your balance before you run. When the balance is too low, Run is blocked: buy credits or upgrade your plan, or ask the workspace owner.
• Credits are taken when the run starts, also with your own AI key. A run that fails is refunded in full. Demo workspaces are free.
• Credit usage lists it as AI Simulator.