When the Thesis Is Right But the Timing Isn’t: The Shutdown of Amogha AI

Last week, Akash Gupta published a note on LinkedIn that most founders dread composing. Amogha AI — the Bengaluru-based startup behind Jamun, an AI mental health companion purpose-built for India — was winding down. Not because the idea was wrong. Not because the product had failed. But because the company could not close a seed round, and continuing to fund operations from its own resources was no longer viable. The shutdown of Amogha AI deserves more than a passing mention in the startup news cycle. It is a precise and honest account of what happens when a genuinely important idea meets the structural limitations of early-stage venture financing — and it carries specific lessons for founders, investors, and the broader ecosystem building in India’s digital health space.

What Amogha AI built

Amogha AI was founded in 2025 by Akash Gupta — an IIT Bombay alumnus, former venture capitalist, and ex-McKinsey consultant — and Ashutosh Garg, an entrepreneur and executive coach with over 25 years of global leadership experience. The third core team member was Vishal Puri. The founding team brought an unusual combination of product, investment, and clinical thinking to a category that demands all three.

Their product, Jamun, was built as a voice-first AI mental health companion specifically designed for the Indian context. The technical architecture was ambitious: persistent memory across sessions, deep personalisation that adapted to individual users over time, end-to-end conversation encryption, and a RAG-driven (Retrieval-Augmented Generation) intelligence engine built on proprietary mental health data developed in collaboration with licensed therapist partners. This was not a general-purpose chatbot with a mental health persona layered on top. It was purpose-built for the domain, with clinical inputs baked into the training data and product design from the outset.

The decision to go voice-first was deliberate. For a significant proportion of Indian users, speaking is more natural and more intimate than typing — and the experience of a voice interaction is closer to speaking with a therapist than typing into a chat window. The product was also built with genuine cultural specificity: Indian family dynamics, workplace pressures, the linguistic code-switching common among urban Indian users, and the particular stigma attached to mental health discussions in Indian households were all embedded in the product design rather than retrofitted after launch.

Jamun covered a broad range of use cases — relationship and family stress, workplace anxiety, loneliness, academic pressure, personal growth — with clear, responsible disclaimers that it was not a licensed clinical provider and could not replace human therapy. It was available on both the App Store and Google Play before the shutdown was announced.

“Really proud of what we built: a technically complex product — memory, personalisation, encryption, a RAG-driven intelligence engine on our proprietary data — all on a bootstrapped budget.” — Akash Gupta, Co-founder, Amogha AI

The problem they were solving — and why the scale of it matters

The structural problem Amogha AI set out to address is not marginal. India faces one of the most acute mental health access crises in the world, and the gap between need and supply is among the largest of any major economy.

India has approximately 9,000 psychiatrists serving a population of 1.4 billion — roughly 0.75 psychiatrists per 100,000 people, against the WHO’s recommended minimum of 1.7 per 100,000. The comparable figure for clinical psychologists is approximately 0.07 per 100,000. The global median is 1.3 psychiatrists per 100,000; high-income countries average around 10 per 100,000. The treatment gap — the proportion of people with diagnosable mental health conditions who never access care — is estimated at 80 to 90 percent in India. Approximately 197 million Indians, or one in seven, live with some form of mental disorder. Depression and anxiety alone affect more than 90 million people.

The cost of therapy compounds the access problem further. A session with a licensed therapist in an Indian metro typically costs between ₹800 and ₹2,500 — putting regular professional support beyond reach for the majority of the population. Outside major cities, access is constrained not just financially but geographically: qualified therapists are concentrated in urban centres, leaving large sections of Tier 2, Tier 3, and rural India with effectively no access to mental health care. The India online mental health market was valued at $151.4 million in 2025 and is projected to reach $464.4 million by 2034, growing at a CAGR of 12.87%. The broader India mental health market was valued at $20.82 billion in 2025 and is projected to reach $27.36 billion by 2034.

Against this backdrop, Amogha AI’s stated mission — to bring down the cost of mental health support by 10x and make it accessible at scale — was a coherent and proportionate response to a genuine crisis. The early product signals validated the thesis at the user level. The company reported approximately 10,000 cumulative users from its public beta, with around 10 percent classified as power users — a cohort with high session frequency, strong retention within that group, and demonstrated willingness to pay. User feedback described tangible emotional benefit: moments of clarity and perspective in difficult periods attributed directly to the product.

Where it broke down

The shutdown was not precipitated by a product failure. It was precipitated by the inability to close a seed funding round, and the decision — clear-eyed, if difficult — not to continue funding operations from the company’s own resources indefinitely.

Akash Gupta’s account of the fundraising process is specific: investors engaged seriously with the thesis, but the available proof points were insufficient to overcome hesitation on two particular dimensions — long-term retention and monetisation. Both concerns are legitimate and structurally difficult to address in a mental health product at the public beta stage.

Retention in consumer mental health products is one of the hardest metrics to prove at early stage. The pattern across the category globally is well-documented: users engage intensively during a period of acute stress, then disengage as that stress resolves. This produces high short-term engagement and lower long-term retention figures — a profile that reads poorly on standard VC due diligence frameworks, even when the behaviour is entirely clinically appropriate. A user who is not in distress should not be engaging with a mental health companion daily. The meaningful metric is re-engagement rate: when users face a new difficult period, do they return to the platform? This is a harder thing to prove at beta stage with a limited user base and a short operating history — and it is precisely the proof that was unavailable.

Monetisation in digital mental health carries ethical complexity that investors find difficult to underwrite at seed. A subscription model raises uncomfortable questions — what happens to a user in acute distress who cannot afford their subscription? How do you price a service that users may genuinely depend on for emotional stability? How do you grow revenue without compromising access? These are answerable questions, but they require demonstrated thinking and ideally demonstrated evidence — which is difficult to produce before a monetisation model has been live long enough to generate meaningful data.

The regulatory environment adds a layer of uncertainty. AI-assisted therapy is among the most actively scrutinised categories in digital health globally. In the United States, several states have moved to restrict AI tools from acting in a therapeutic capacity. In India, the regulatory framework for AI mental health applications is still evolving. A regulatory shift requiring clinical validation or licensing for AI companions would materially alter the economics and operating timeline for any company in this space.

“Contrarian ideas are hard to fundraise for. Many investors believed in the thesis, but our proof points weren’t enough to overcome hesitation around long-term retention and monetisation. We couldn’t close our seed round.” — Akash Gupta, Co-founder, Amogha AI

The competitive landscape

Amogha AI was not operating in a vacant market. India’s AI-assisted mental health space has attracted meaningful capital over the past three years, and the competitive dynamics provide useful context for understanding where Jamun was positioned — and what structural challenges it faced relative to better-capitalised peers.

The most instructive comparison is with Wysa, which has raised over $20 million and serves more than 6 million users globally. Wysa’s commercial model differs from Jamun’s in one critical respect: it operates B2B2C, selling to employers, healthcare systems, and insurers who deploy the app as a benefit. This model sidesteps the consumer retention and monetisation problem entirely — the employer underwrites the cost, and unit economics are governed by enterprise sales cycles rather than consumer subscription renewals. Amogha AI was building a direct-to-consumer product in a market where B2B2C has proven structurally easier to finance at early stage. The direct-to-consumer model, if it works, ultimately reaches a far larger population than enterprise wellness programmes ever will — but it demands a longer proof-of-retention period before institutional investors will engage.

What the ecosystem should take from this

The closure of Amogha AI surfaces structural issues that extend beyond one company’s experience and are worth examining clearly.

For founders building in AI mental health: The self-funded path through product-market fit in this category is exceptionally difficult. Building a credible mental health AI requires proprietary clinical data, therapist partnerships, regulatory navigation, and a user base large enough to demonstrate retention at statistical significance — all before institutional capital is available to fund the next phase. Founders in this space should explore non-dilutive options: government health innovation grants, academic medical centre partnerships, NGO collaboration, or corporate pilot programmes that provide both early revenue and clinical credibility. The proof points investors require before writing seed cheques can only be generated with the capital investors are withholding pending those proof points — breaking that loop requires finding alternative funding pathways.

For investors: The retention concern that prevented Amogha AI from closing its seed round reflects a misapplication of standard consumer product metrics to a clinically specific category. Mental health engagement is episodic by nature. A user who is not in distress should not be engaging with a mental health app daily. The right metric is re-engagement rate: when users face a new difficult period, do they return? Applying D7 or D30 retention benchmarks from consumer social or entertainment apps to mental health products is measuring the wrong thing. Investors with genuine conviction in the category should be developing a purpose-built evaluation framework rather than defaulting to metrics designed for fundamentally different products.

The India-specific competitive advantage is real. The global mental health AI market is being built primarily by US and UK companies for Western cultural and clinical contexts. An AI companion that genuinely understands Indian family dynamics, linguistic code-switching, the pressures of competitive Indian education, and the specific form of stigma surrounding mental health in Indian households represents a defensible and durable product position. That advantage cannot be easily replicated by internationally developed platforms. The first well-capitalised Indian company to prove retention and monetisation in this space will be very difficult to displace.

“Our journey didn’t reach its destination. But it was worth taking. We leave with full conviction that this product should exist, and one day, hopefully it will.” — Akash Gupta, Co-founder, Amogha AI

That conviction is well-founded. India has 197 million people living with mental health conditions and a treatment gap of 80 to 90 percent. Closing that gap at any meaningful scale requires AI — the arithmetic of therapist supply and population need makes human-only delivery mathematically impossible. A voice-first, culturally grounded, privacy-first AI companion built specifically for India is not a niche product opportunity. It is a piece of necessary public health infrastructure. Amogha AI built a rigorous early version of it. The company that eventually reaches scale in this space will owe something to the design thinking, clinical partnerships, and honest failure analysis that Akash Gupta, Ashutosh Garg, and Vishal Puri have placed on the record.

Sources: Akash Gupta (LinkedIn), amogha.ai, Google Play Store, Business Standard, IMARC Group, DataM Intelligence, Medical Buyer, PIB India, Confederation of Indian Industry / MediBuddy Corporate Wellness Index 2025. Data verified as of August 2026.