AutoSEO Research — User Guide
NordicPulse AI's autonomous AI research agent for visibility strategies — inspired by Karpathy's autoresearch.
What is AutoSEO Research?
AutoSEO Research is an autonomous research agent that iterates on SEO/AEO strategies based on objective visibility metrics. You set a goal (brand + keyword), and the agent runs a baseline scan across 4 AI platforms. Based on the results, Gemini proposes a new strategy. The agent re-scans to measure the effect — strategies that improve the score are kept, the rest are discarded.
After 3–5 iterations, you get a research log with the best strategies and AI-generated insights.
How to use the tool
1. Fill in the form
- Brand name — e.g. "NordicPulse"
- Primary keyword — e.g. "SEO tools for businesses"
- Industry — e.g. "SaaS / Marketing"
- Geography — e.g. "Norway" (optional)
- Secondary keywords — up to 2 additional keywords for broader coverage (optional)
- Number of iterations — 3–5 (default: 5)
- Language — Norwegian or English
2. Start the research
Click Start Research. The agent:
- Runs a baseline scan of your brand in ChatGPT, Gemini, Perplexity and Claude (real API calls)
- Calculates a visibility score (0–100)
- Proposes a new strategy via Gemini based on current visibility data
- Re-scans to measure the strategy's effect
- Keeps strategies that improve the score by ≥2 points, discards the rest
- Repeats steps 3–5 for the selected number of iterations
- Generates summarized insights
Research typically takes 2–5 minutes depending on the number of iterations.
3. Follow the live log
During research, a real-time event log is displayed:
- Baseline started/complete — Initial scan with score
- Iteration started — New round begins
- Strategy proposed — Gemini proposes a new strategy
- Strategy kept (green) — Score improved ≥2 points
- Strategy discarded (red) — Insufficient improvement
- Research complete — All iterations finished
Understanding the results
Experiment Log
Table with all iterations:
| Column | Description |
|---|---|
| # | Iteration number (0 = baseline) |
| Keyword | Keyword used in this scan |
| Score | Visibility score 0–100 |
| Delta | Change from previous best score |
| Strategy | Name of the proposed strategy |
| Status | Kept or Discarded |
Kept strategies
List of strategies that actually improved visibility:
- Title — Short strategy name
- Description — 2–3 sentences about what the strategy involves
- Tactics — Concrete actions to implement
- Expected impact — High, Medium or Low
- Category — Content, Technical, Authority, Entity or Local
Platform results
For each iteration, per-platform visibility is shown:
- Visible/Not visible — Whether the brand was mentioned
- Position — Placement in the AI response
- Mentioned brands — Competitors that were mentioned
- Confidence — AI confidence level
All platform checks are marked with a Live badge (green pulse) — real API calls.
Insights
AI-generated summaries with:
- What worked best
- Patterns in AI platform preferences
- Actionable recommendations for further work
Estimation model
AutoSEO Research uses a "keep or discard" method inspired by Karpathy's autoresearch:
- Baseline — Starting point with real visibility data from 4 platforms
- Propose — Gemini proposes a strategy based on current data
- Measure — Re-scan to capture natural variation and change
- Evaluate — Strategies with ≥2 points improvement are kept
- Iterate — The process repeats with cumulative kept strategies
The visibility score is calculated as a weighted average of platform visibility, position data and confidence level.
Cost
150 credits per research session. Covers:
- Baseline scan (4 AI platforms)
- 3–5 iterations × 4 platform calls per iteration
- Strategy generation via Gemini per iteration
- Insight generation
Data integrity
- Live badge (green pulse): Platform checks are real API calls to ChatGPT, Gemini, Perplexity and Claude
- AI Analysis badge (cyan): Strategy proposals and insights are generated by Gemini
- No simulated data for platform checks
- Research log is stored in the browser's localStorage
Session history
Previous research sessions are automatically saved in the browser's localStorage:
- View complete results from previous sessions
- Compare strategies across sessions
- Delete individual sessions as needed
Tips
- Start with one strong primary keyword and add 1–2 secondary ones for broader coverage
- Run the research multiple times over time to see trends
- Implement the kept strategies and re-run after 2–4 weeks
- Use the insights as a basis for content strategy and technical SEO work
- Combine with AEO Watchtower for deeper analysis of individual platforms
Related tools
- Dark Traffic Estimator — Estimate hidden AI traffic in direct visits
- Brand Stickiness Score — Measure consistency in AI mentions
- Sentiment Trend Tracker — Track sentiment trends across AI platforms
- AEO Watchtower — Full AI visibility analysis with forensic evidence collection
Last updated 2026-03-18