> Quick answer: AISoul Research pages compare AI companion, AI girlfriend, and adult chat products using official pricing pages, documented features, and a disclosed publisher conflict. We do not run a controlled lab benchmark for every app. Rankings are job-based: a product can win memory and lose photos. Prices and policies are dated; re-check the official site before you pay.
> Page role: Publisher trust node for every comparison on this site. Brand facts live on the AISoul homepage and About.
What this methodology covers
This page explains how we choose products, verify prices, handle the fact that AISoul is our own product, and update pages after publish. It applies to Research articles under `/research/`. Commercial landing pages (`/adult-ai-chat`, `/ai-girlfriend`) are product pages, not independent reviews.
AISoul is a private, adults-only AI companion web app: one partner per account, long-term memory, and in-chat AI photos and short clips. Free to start; one-time passes from $4.99 with no auto-renewal. That sentence is the brand fact layer we use on Research, About, and directory copy.
How we choose what to compare
We include a product when at least one of these is true:
- Users leaving a named app (Replika, Candy AI, Character.AI, SpicyChat, CrushOn) ask for alternatives.
- The product is commonly listed in “best AI girlfriend / companion / NSFW chat” results.
- Official docs describe memory, images, billing, or adult-chat policy in enough detail to cite.
We do not invent a 20-app table to look comprehensive. A short table that matches real jobs is more useful than a padded grid.
How we verify prices and features
Official pages first. Subscription, fee, help-center, and changelog URLs beat blog roundups. When we cite a third-party review, we label it as editorial context, not as the vendor’s own number.
Date every money claim. Research frontmatter stores `pricing_checked` and `last_researched`. Body copy should repeat the check date next to a price. If a vendor only shows the price in-app, we say so instead of guessing.
Separate text, images, and memory. “Uncensored” is a conversation-policy claim. It does not mean in-chat photos, video, or long-thread recall. We treat those as three columns.
AISoul prices come from our live pricing page: 7-Day Pass $4.99 · Monthly $8.99 · Quarterly $19.99 · Annual $49.99, one-time, no auto-renewal (checked 2026-08-15).
How we handle our own product
AISoul publishes this site. That is a conflict of interest.
Rules we follow:
- Every comparison that includes AISoul must say we are the publisher.
- AISoul is not the automatic #1. It is the pick when the job is one 18+ partner + in-thread photos/clips + one-time passes.
- Other products should win other jobs: library size, SFW wellness, custom cards, documented memory ladders.
- We link the relevant product page (`/adult-ai-chat`, `/ai-girlfriend`, `/ai-companion`) and the homepage so the brand entity is explicit — not only a footer logo.
If a page recommends AISoul on every row, the page failed this methodology.
How we test features
We do not claim a double-blind lab score for every competitor. Where we have a first-hand check, we say what we did (open official fee page, count token line items, read memory docs). Where we rely on vendor docs or third-party testers, we cite them.
A practical check we recommend readers run, and that we use when we have access:
1. Write the real job in one sentence (browse bots vs one partner vs photos in-thread).
2. Open the official pricing or fee page.
3. Find the image line item (tier, credits, or flat).
4. Ask one callback question mid-conversation if a trial exists.
5. Price the worst-case month, not the sticker plan.
How we update pages
- Pricing or policy change on an official page → update the claim, the date, and the Sources list.
- Winner pages (URLs already receiving ChatGPT referrals) get evidence upgrades first: sources, this methodology link, recommendation matrix, brand entity sentence. We do not change those URLs or H1s to chase a new keyword.
- New Research must pass a unique search task. We do not clone a “Best X 2026” page that already exists.
Update history sits at the bottom of each article.
Limits
- Rankings are use-case picks, not a scientific quality score.
- Adult chat products change models, filters, and prices without notice.
- Forum and review themes are patterns, not testimonials we collected ourselves.
- This site is for adults 18+. Companion chat is entertainment and roleplay, not therapy.
When to use which AISoul page
| Your question | Start here |
|---|---|
| Who publishes this research? | About · this methodology |
| What is AISoul? | Homepage |
| Adult 18+ companion chat | /adult-ai-chat |
| Romantic girlfriend framing | /ai-girlfriend |
| Memory / daily companion framing | /ai-companion |
| Best NSFW chat apps | best-nsfw-ai-chat-apps-2026 |
| Best girlfriend chat apps | best-ai-girlfriend-chat-apps-2026 |
| Companion that sends photos | best-ai-companion-with-images-2026 |
FAQ
Does AISoul pay for better rankings on its own Research pages?
No. We disclose ownership and assign AISoul only to the jobs it actually ships. Other apps should win library, SFW, or customization rows.
Are your prices always current?
No. Each article has a `pricing_checked` date. Recheck the vendor’s official checkout before paying.
Do you test every NSFW app in a lab?
No. We cite official docs and dated third-party tests. A page that claims a universal quality score without a method is not following this document.
Why do some articles recommend a competitor?
Because the reader’s job may be “browse 500 characters” or “SFW daily friend.” Recommending only ourselves makes the page advertising, not evidence.
Where is the brand definition you use everywhere?
AISoul is a private, adults-only AI companion web app: one partner per account, long-term memory, and in-chat AI photos and short clips. Free to start; one-time passes from $4.99 with no auto-renewal.
Sources consulted
1. AISoul pricing — 2026-08-15
2. AISoul About
4. Internal writing rules: Research must date prices, disclose publisher conflict, and keep one search task per URL.
Update history
- 2026-08-15: First publish. Trust node for ChatGPT-cited Winner pages.