God, Ghosts and GPT

When I think about it, I have the impression that  AI will make more of the rational world “cheap”. More problem solving and more knowledge work will be handled by machines (just read how mathematicians are wondering if mathematics won't become just a hobby in the future). But I do not think that means people will become more rational. We will simply have more time for questions where there is no clear answer. And as AI changes how we think about work and our own value, the need for meaning may become stronger.

That is why I wanted to look at The Belief Economy. Religion, astrology, dream interpretation and the paranormal all help people make sense of things that cannot really be solved. Below is the analysis of the four categories and on the right you can find the landscape.

Astrology

What is this category about?

• This category is about consumer products that use astrology to explain what is happening in a person's life or what may happen next. Most start from birth information and planetary positions, then turn them into personalized interpretations. 

• The strongest need is usually not abstract curiosity about astrology itself. People often come with a concrete question about a relationship or a decision they are struggling with. Astrology gives them a personalized narrative around that question, which can feel more relevant than generic advice.

• There is also a recurring content need that makes astrology particularly suited to consumer apps. A user does not need to wait for a major life decision to come back. Horoscopes can change every day, while planetary movements continuously create new material to interpret. This gives astrology products a natural reason to become part of a user's routine.

What do products in this category do?

• The app type turns astrology into a personalized content app. Co-Star generates horoscopes and birth chart insights based on each user's astrological profile. CHANI uses a similar subscription model but puts more emphasis on guidance and editorial content. The Pattern focuses heavily on personality and relationship dynamics. The product behaves more like a personalized media app than a traditional consultation with an astrologer.

• The other major model is an on-demand marketplace for human astrologers. AstroTalk and InstaAstro let users pay to chat or speak directly with an astrologer when they have a question. Nebula also built a marketplace around spiritual advisors. This model monetizes moments when the user actively wants reassurance or an answer, rather than relying only on a recurring subscription.

• The separation between automated astrology and human advice is starting to become less clear. Marketplace products use free astrology tools to bring people into paid consultations, while content apps are becoming more conversational. The Pattern introduced In-Depth in 2025, which lets users discuss its astrology based insights through an AI conversation instead of simply reading them.

What do the company milestones tell us?

• AstroTalk shows how large the consultation model can become. Its operating revenue more than doubled to ₹651 crore (\~$69 million) in FY24 before reaching ₹1,176 crore (\~$125 million) in FY25. By July 2025, the company said 80 million users had connected with astrologers and 18 million paid consultations had taken place. In August 2026, its revenue run rate had crossed ₹2,500 crore (\~$265 million) and an ESOP buyback valued the company at $1 billion.

• There is meaningful scale behind other consultation platforms as well. InstaAstro grew revenue from ₹52 crore (\~$5.5 million) in FY25 to ₹111.26 crore (\~$11.8 million) in FY26. By August 2026 it had reached 12 million users and more than ₹200 crore (\~$21 million) in annualized revenue. OBRIO, the company behind Nebula, reached $50 million in ARR in 2024. Nebula later reported more than 250,000 monthly active users and 5 million expert consultations during 2025.

• The personalized content model has produced large consumer apps even without the same transaction volume. Co-Star passed 30 million organic users before being acquired by Midjourney in 2026, when it still had around 4.3 million monthly active users. The Pattern had already reached 18 million profiles by early 2022, with active users growing 50% year over year. CHANI reached around 250,000 active users while remaining bootstrapped, then passed 2 million downloads in 2025.

What are the major trends shaping this category?

• Astrology is widely used, but there is little evidence that belief itself is suddenly exploding. Pew found that 27% of US adults believed in astrology in its 2024 survey, roughly unchanged from 2017. At the same time, 30% said they consulted astrology, tarot or a fortune teller at least once a year. Most participants said they engaged mainly for fun rather than using these practices to make important decisions. This suggests digital astrology does not need to create a new belief system. It can build products around a behaviour that was already widespread.

• The audience skews heavily toward younger consumers, particularly younger women. Pew found that 43% of US women aged 18 to 49 believe in astrology, while 46% consult astrology or horoscopes at least once a year. This helps explain why astrology has adapted so well to mobile products. The behaviour already exists among consumers who are comfortable turning recurring personal interests into app usage.

• India is becoming a particularly important market because a large offline astrology economy is moving online. Redseer estimated India's online astrology market at around $102 million to $106 million in 2024 and expected it to grow roughly tenfold by FY30. The shift makes astrologers easier to access for younger consumers and also opens the market to Indians living abroad who may not have the same access to traditional offline services.

• Conversational AI should make astrology more interactive rather than simply generate more horoscopes. Redseer now includes astrology consultations among the consumer use cases emerging around conversational AI in India. The important change is that personalized interpretation no longer has to be delivered as a fixed piece of content. Users can ask follow-up questions and keep the interaction going, which makes the product behave more like an advisor than a horoscope page.

How does the funding environment look?

• AstroTalk is the most heavily funded company in the category and has raised at least $34 million in its two 2024 rounds, on top of an earlier ₹6 crore seed investment. InstaAstro has raised about $14.3 million across its Pre-Series A and Series A, plus an earlier ₹3.2 crore seed round. In both cases, institutional capital followed a marketplace model where consumer spending can be measured directly through paid consultations.

• The major US astrology apps raised meaningful amounts earlier in their development. Co-Star raised roughly $21 million before being acquired by Midjourney in 2026. Sanctuary has raised around $4.5 million. The Pattern has disclosed investors including Sweet Capital and DGNL Ventures, although the amounts are not public in the landscape data.

• Why did Midjourney acquire Co-Star? My read is that Midjourney was buying consumer product expertise more than astrology itself. Co-Star has shown that highly personalized generated content can become something people come back to regularly, which is exactly the type of consumer behaviour AI companies are trying to build (recurring behavior).

• I was initially surprised to see VCs backing astrology startups, but I realized that these are consumer apps borderline in the entertainment category. The belief does not need to be scientifically valid for the product to work as a business. Users can open the app for entertainment and still become frequent users or pay for additional guidance. That makes astrology closer to other consumer categories built around identity and fun than the irrational premise might initially suggest.

Religion / Digital Faith

What is this category about?

• This category lists products that bring religious belief into digital experiences. The underlying need is not new. People already pray, read religious texts or seek guidance from their faith. Software makes those practices easier to access outside a church, mosque or temple.

• A big part of the category is about helping people practice their faith more consistently. Religion often depends on repeated habits, which fits naturally with a mobile app. A product can become part of a person's daily routine by making prayer easier to follow or helping them spend more time with scripture.

• Another important need is access to religious guidance when someone has a question. Traditionally that guidance might come from a religious leader or from studying religious texts directly. Digital products reduce the friction. AI takes this much further because the user can now ask a personal question and receive an immediate answer framed through their faith.

What do products in this category do?

• The most common product model is a digital companion for religious practice. Hallow helps Catholics pray through guided audio experiences, while Muslim Pro makes daily Islamic practice easier through its mobile app. Quranly focuses more narrowly on helping Muslims build a regular Quran reading habit. These products mostly take an existing religious practice and make it easier to maintain through software.

• Another model turns religious services that normally happen in a physical place into online events. Sri Mandir lets Hindu devotees participate in pujas and make offerings through temples remotely. Utsav follows a similar model by connecting users with temples and delivering prasad afterwards. The product is not trying to create a new religious practice. It makes an existing one accessible when the devotee cannot be there physically.

• AI creates a very different product because religion becomes conversational. Bible Chat lets users ask questions and receive answers based on Christian scripture. Magisterium AI does something similar for Catholic teaching, but puts much more emphasis on citations from Church documents. Ansari is built around Islamic sources. Instead of opening an app to consume religious content, the user can now describe a personal situation and ask what their faith says about it.

• AI is also creating a software layer for religious institutions themselves. Gloo is the clearest example in the landscape. It provides technology to churches and ministries and is increasingly building around “values aligned AI”. This matters because the same shift happening on the consumer side is also happening inside religious organizations themselves.

What do the company milestones tell us?

• Traditional religious apps have already reached enormous consumer scale. Muslim Pro now reports more than 190 million downloads worldwide. Hallow had nearly 24 million installs by mid 2025, but the more interesting metric is engagement: users had already completed more than 1 billion prayers inside the app.

• Other Christian apps show that this is not limited to one breakout winner. Glorify reached more than 25 million users by May 2026. Pray.com had reached 19 million downloads by mid 2025, with users consuming more than 3 billion minutes of content. These are large consumer media products even if the content itself is religious.

• Digital devotion is also becoming a meaningful transaction market in India. Sri Mandir reached more than 40 million downloads and around 3.5 million monthly active users by mid 2025. More than 1 million devotees had paid for a prayer or offering during the previous year, helping the company reach a $12 million revenue run rate. Utsav is smaller but was already processing more than 250,000 orders per month in 2026 after growing revenue eightfold since its previous funding round.

• AI products are reaching scale really fast. Bible Chat reached 10 million active users and more than $33 million in annualized revenue during 2025, only a few years after the company was founded. Magisterium AI passed 1 million users across more than 190 countries in 2026. Religious AI is therefore already moving beyond small experiments.

What are the major trends shaping this category?

• AI is starting to move from being a tool for religion to becoming a source of religious guidance itself. Research from Barna and Gloo found that roughly one in three US adults considers spiritual guidance from AI as trustworthy as guidance from a pastor. Among Millennials, the figure reaches 44%. Nearly half of practicing Christians also said AI had helped them with some aspect of their spiritual life. This is a much deeper use case than reading a digital Bible or listening to a guided prayer.

• The problem is that general AI models do not necessarily share the worldview of the people asking them religious questions. The Economist tested 25 frontier models against the World Values Survey and found that Western models were much more secular than most of the world's population. None reflected the prevailing worldview of most African or Muslim countries. The article specifically pointed to Ansari as a response to this problem. This creates a reason for religious AI products to exist even when people already have access to ChatGPT: the differentiation is not intelligence alone, but which religious framework the answer follows.

• That makes trust and grounding central product problems for religious AI. Most pastors are already using AI in some form, with only 13% saying they do not use it at all. Yet 94% are concerned about AI misinterpreting scripture. This explains why products such as Magisterium AI and Ansari put so much emphasis on showing the religious sources behind an answer. In this market, a fluent answer is not enough if users cannot trust where it came from.

How does the funding environment look?

• VC investors have already put significant capital into traditional prayer and devotional apps. Hallow has raised around $102 million across its Series A, Series B and Series C. Glorify raised around $80 million across two rounds. These companies were funded primarily as consumer subscription businesses rather than as AI companies.

• Investors have also backed the transaction model around Hindu devotion. Sri Mandir has raised roughly $46 million across its disclosed rounds, including ₹175 crore in 2025, which is about $18.5 million at today's exchange rate. Utsav added a $3.8 million Series A in 2026. The monetization model here is different because users are paying for religious services rather than only subscribing to content.

• The first AI native religious companies are now attracting venture funding as well. Bible Chat raised a $14 million Series A in 2025 after showing unusually fast consumer growth. Magisterium AI has raised around $3 million so far. The amounts are still smaller than the largest prayer apps, but Bible Chat's revenue growth shows that an AI religious assistant can already support a real consumer subscription business.

• Gloo shows how far faith technology can go beyond the consumer app market. It is now a public company and has built a large software business around churches and ministries. This makes the category broader than it first appears. At one end are mobile apps designed to help an individual practice their faith. At the other is infrastructure built specifically for the organizations that shape that faith.

Paranormal

What is this category about?

• In this category you’ll find mostly mobile apps built around the idea that supernatural activity can be experienced or detected. Instead of only consuming paranormal stories, users can actively look for something unexplained themselves. The phone or device becomes part of the experience.

• A central user need is turning an ambiguous experience into something that feels observable. A strange noise or an unusual feeling is difficult to interpret on its own. Paranormal tools give the user a way to record what happened or attach a signal to it, even when the connection between that signal and a supernatural explanation is not scientifically established.

• The experience does not necessarily require complete belief. Part of the appeal is the uncertainty itself. Users can go ghost hunting or explore an unusual location because the possibility that something might happen makes the activity entertaining.

What do products in this category do?

• The most common product turns a smartphone into a ghost hunting tool. GhostTube uses sensors already available inside the phone and combines them with recording features designed for paranormal investigations. Weasel Development follows a similar model with apps such as Ghost Camera and Ghost Hunting Tools. This dramatically lowers the barrier to trying something that previously required dedicated equipment.

• Another product type is built around attempted communication with spirits. Some apps process audio while others generate words that users interpret as possible responses. Weasel Development's Ghost Voice Box and Spotted: Ghosts' Spirit Talker are examples. The product creates a structured interaction around something that would otherwise be difficult to interpret.

• Randonautica takes a different approach because the paranormal experience happens through exploration rather than detection. The app generates random coordinates and asks users to set an intention before visiting them. The belief is that the combination of randomness and intention can lead to meaningful coincidences. It turns the physical world itself into the product experience.

• Some companies are also moving beyond standalone phone apps. Spotted: Ghosts sells dedicated paranormal devices and has built ParaLink Mesh to connect its hardware through one control layer. GhostTube has also expanded from individual apps into a broader family of investigation tools. The product direction is therefore starting to look more like a paranormal toolset than a single novelty app.

What do the company milestones tell us?

• Randonautica shows how quickly a paranormal product can become a mainstream consumer phenomenon. The app reached around 10.8 million users within roughly five months of launching in 2020. What is striking is that this happened without institutional funding. Virality rather than paid distribution created the scale.

• GhostTube has built the strongest portfolio of paranormal investigation apps in the landscape. Its SLS Camera crossed 5 million Android installs in 2024. GhostTube Paranormal Videos reached the same threshold in 2026. SEER, its AI powered paranormal app, crossed 1 million Android installs that year as well. This suggests that demand is not limited to one novelty product.

• Small independent developers can also reach meaningful scale in this market. Weasel Development's Ghost Hunting Tools crossed 1 million Android installs as early as 2018, while newer products such as Ghost Camera and Ghost Voice Box now also show more than 1 million downloads. Spotted: Ghosts is much smaller, but Spirit Talker passed 100,000 Android installs despite being a paid app.

What are the major trends shaping this category?

• Belief in the paranormal remains widespread, but it does not appear to be increasing much. Gallup found in 2025 that 39% of US adults believe in ghosts, while 34% are broadly open to several paranormal ideas. Those levels are mostly similar to Gallup's findings from more than twenty years ago. This looks more like a durable consumer behaviour than a new belief trend.

• Paranormal ideas remain powerful entertainment even for people who are not fully convinced by them. The BBC's Uncanny grew from a podcast into television and live events, while paranormal series The Telepathy Tapes became Apple's most popular podcast series of 2025. The attraction is therefore broader than a hard core group of believers. Mystery itself has become a form of entertainment and community.

• Smartphones have made paranormal investigation much easier to productize. Modern phones contain sensors that can measure movement or magnetic fields, while cameras and microphones provide additional inputs. These components are available inside hardware that users already own. That allows developers to package a paranormal investigation experience entirely as software instead of asking the consumer to buy specialist equipment first.

• Generative AI creates a new problem for a market built partly around evidence. Modern video models can generate increasingly realistic footage, while researchers have warned that synthetic video challenges the idea that a recording can be treated as evidence by itself. For paranormal products, this should make provenance more important. A ghost video becomes less convincing when anyone can manufacture one, which could increase the value of products that preserve the context around how something was captured.

How does the funding environment look?

• The funding environment is completely different from Astrology or Religion. None of the four companies in the landscape has a disclosed institutional funding round. Randonautica, GhostTube and Weasel Development are all bootstrapped, while Spotted: Ghosts also operates like a very small independent company.

• That lack of VC funding does not mean the products lack users. Randonautica reached more than 10 million users and several GhostTube apps have passed 5 million installs. Weasel Development has also produced multiple million download apps. The category shows that consumer reach and venture scale are not the same thing.

• The economics make the category particularly suited to indie developers. A small team can build an app on top of hardware already present in the user's phone, then monetize through ads or in-app purchases. There is little need for large upfront capital, while the novelty of the product can create strong organic distribution if it goes viral.

• Overall, I would not describe paranormal apps as a VC category today. There is real consumer demand and clear evidence that individual products can reach millions of people, but there is little sign that this scale translates into the type of company that needs large amounts of venture capital. There is also no public company in the landscape. The category currently looks much more like a profitable indie software market than an emerging venture market.

Dream Interpretation

What is this category about?

• This category covers products that help people assign meaning to their dreams. People have interpreted dreams for centuries, but the basic need has barely changed. A strange or emotionally intense dream can feel significant, and the user wants help understanding whether it says something about their waking life.

• The difficulty is that dreams are extremely personal. The same image can mean something completely different depending on the person who dreamed it. This makes dream interpretation a natural fit for software that knows more about the user and can connect a new dream with what they have shared before.

• There is also a strong journaling component to the category. Dreams disappear from memory quickly after waking, so recording them is part of the product itself. Once someone builds a long history of dreams, the value can shift from interpreting one strange dream toward finding recurring patterns over time.

What do products in this category do?

• The main product model combines a dream journal with personalized AI interpretation. Oniri lets users record a dream and receive an interpretation that can continue through follow-up questions. Wakefully follows a similar model but puts more emphasis on patterns across the user's dream history. AI makes the experience much more personal than searching for individual symbols in a traditional dream dictionary.

• Some products connect dream interpretation more explicitly with mental wellness. DreamApp interprets dreams but also offers access to one-to-one coaching sessions. The idea is that a dream can become the starting point for a broader conversation about what the user is experiencing rather than an isolated prediction about what a symbol means.

• Dream apps can also extend into lucid dreaming. Oniri includes tools designed to help users become aware that they are dreaming and potentially influence what happens next. This changes the product from simply explaining a dream afterwards into helping the user engage more actively with dreaming itself.

What do the company milestones tell us?

• DreamApp is the clearest example of meaningful consumer scale in the category. It had already reached more than 100,000 monthly active users and collected 3 million dreams by February 2024. The company now reports more than 1.6 million downloads and 3.9 million dreams analyzed. Users have also completed more than 170,000 one-to-one coaching sessions through the product.

• Oniri shows that dream journaling can create surprisingly deep engagement even inside a small company. Users had recorded their one millionth dream by April 2023. The company itself remains very small, which makes the milestone interesting because a large corpus of user generated dream data does not necessarily require a large organization to build.

• Wakefully is earlier, but its data shows a similar pattern. The company currently reports more than 20,000 users and a corpus of over 80,000 annotated dreams. It has turned those dreams into more than 5 million structured datapoints. The company is still small, but the amount of data grows much faster than the organization behind it.

What are the major trends shaping this category?

• Dream interpretation is moving away from the idea that every symbol has one fixed meaning. Recent research reinforces how strongly dream content depends on the individual. A 2026 study found that personal characteristics and waking experiences systematically shaped what people dreamed about. This makes personalized interpretation more defensible than a generic dream dictionary, even if there is still no scientifically established method for discovering a hidden objective meaning behind every dream.

• Generative AI is unusually well suited to the format of dream interpretation. Dreams arrive as messy pieces of personal narrative rather than structured data. A 2026 review argues that LLMs can analyze large volumes of dream reports while producing interpretations that take more context into account. This removes much of the manual work that previously made personalized dream analysis difficult to offer at consumer scale.

• Early research suggests that people can actually find AI dream interpretations insightful. In a 2026 study, participants gave ChatGPT their dream along with information about their waking life. They generally rated the resulting interpretations as accurate and useful for personal insight. ChatGPT scored lower than previous therapist or group studies on the broader insight measure, so this is not evidence that AI replaces a therapist. But it does show why a consumer dream app can feel valuable even when the interpretation itself is subjective.

• The longer-term product opportunity may be less about interpreting one dream and more about analyzing hundreds of them. Researchers are already using LLMs to study large collections of dream reports and identify how dream content changes with the dreamer's circumstances. A consumer app can apply the same logic at the individual level. Once it has months of dream history, it can surface changes that would be almost impossible for the user to notice manually.

How does the funding environment look?

• There is almost no meaningful institutional funding in the current landscape. DreamApp has one disclosed angel investment, but the amount is unknown. Wakefully has received funding from ALIAVIA Ventures, also for an undisclosed amount. Oniri received a CHF 100,000 entrepreneurial project award from EPFL Startup Launchpad and the Wyss Center.

• This looks much more like an indie consumer app category than a venture category today. DreamApp has reached more than one million downloads and Oniri has accumulated more than one million recorded dreams without raising large rounds. Software costs are relatively low, so a small team can build a meaningful consumer product without needing much outside capital.

• The stronger venture case may emerge when dream interpretation becomes part of a broader wellness product rather than remaining the entire product. DreamApp already connects interpretation with human coaching, while Wakefully is expanding toward a wider view of personal patterns. Dream interpretation can be the hook that gets users to share highly personal information, while the larger business could eventually be built around what the product does with that context.

• Overall, I would describe Dream Interpretation as not yet VC compatible as a standalone category, but AI makes it more interesting than the older generation of dream apps. The market is still small and there is no heavily funded company or public company in the landscape. What has changed is the product: generative AI makes personalized interpretation cheap enough to turn a very old human behaviour into a much better consumer experience.