# Phở Chat: AI for research & clinical medicine

> Real DOI and PubMed citations -- no fabricated sources: claims without a source are labeled "unverified". Built for physicians, pharmacists, researchers, and specialists who need answers they can trust.

- **URL**: https://pho.chat/en
- **For**: Clinicians, Researchers, Pharmacists

## Details

### A whole new generation.

v1 proved the idea. v2 makes it the tool you trust every day. Faster. Better grounded. More autonomous.

- **Cited, or labeled** (Cite-or-abstain, every claim): Every claim links to a real PubMed/DOI source. No source, and we say "unverified" instead of dressing it up.
- **Runs on Google infrastructure** (Vertex AI, Google Cloud): Gemini embeddings at 3072 dimensions, isolated Cloud Run sandboxes. Enterprise-grade, not a weekend script.
- **Autonomous agents** (Multi-agent, orchestrator): Deep Research coordinates agents to search, screen, appraise GRADE, and synthesize. Every phase visible -- no black box.
- **Fast as thought** (< 1.2s, first paint): Lighthouse 90+, 24-hour content cache. Repeat questions feel instant. Speed is a feature, too.

### Built for real work, not demos

- **Cite-or-abstain -- verifiable sources**: Every claim must be backed by a DOI/PubMed/OpenAlex source. No source -> flagged as "unverified", never invented.
- **Clinical Answer Mode**: Grounded clinical answers with inline citations, GRADE level, and cite-or-abstain per claim. Built for real clinical use, not demos.
- **BYOK -- zero markup**: Plug in OpenAI, Anthropic, Gemini, DeepSeek, xAI keys. Chat runs on your key at provider-direct pricing, zero markup. Keys encrypted AES-256-GCM, deleteable anytime.
- **Phở Brain -- living memory**: Store personal knowledge, auto-injected into every chat context. Ask Brain = chat grounded on your own saved documents (like NotebookLM, but private).
- **Agent sessions inside chat**: Ask for a literature review or an evidence table: Phở proposes the scope, you confirm, then it searches PubMed, checks for retractions, writes with citations and hands you a .docx. A run on your own key costs no credits.
- **Deep Research -- 5 phases**: Auto-searches PubMed/ArXiv/OpenAlex, screens title/abstract, appraises GRADE, synthesizes, exports PRISMA report. Each phase visible -- no black-box spinner.

### Paste your references. See which ones are made up.

Phở's free citation checker, run on a list with real problems: one retracted paper, one cited with the wrong year, one that does not exist. Every verdict links to its source.

1. Paste a reference list in any citation style, or just the DOIs.
2. Phở looks up each reference in Crossref, PubMed and OpenAlex.
3. Each reference gets a verdict with its evidence: the real DOI, the retraction date, the field that is wrong. A made-up reference has no DOI to click.

Check your own list (Free, no sign-up): https://pho.chat/tools/

### Full EBM standards in one workspace

Designed for physicians, researchers, and pharmacists to trust in real clinical settings.

- **PICO**: Structured clinical question framework
- **GRADE**: Evidence quality appraisal
- **RoB 2**: RCT risk-of-bias assessment
- **STROBE**: Observational study checklist
- **CONSORT**: RCT reporting checklist
- **PRISMA**: Systematic review flow diagram
- **SR Screen**: Automated title/abstract screening
- **PubMed**: NCBI medical literature search
- **OpenAlex**: 200M+ open academic papers
- **DOI link**: Real source URLs, no opaque redirects
- **Drug DB**: Drug interaction checker (Opus)
- **ClinicalTrials**: Active clinical trial registry

### Compared with ChatGPT and Perplexity

| Feature | Phở Chat | ChatGPT | Perplexity |
| --- | --- | --- | --- |
| Price | Free - Pro $10/month - Founder Lifetime $59 | $20/month | $20/month |
| Real DOI-backed answers | Yes | -- | Partial |
| Cite-or-abstain (no fabrication) | Yes | -- | -- |
| GRADE evidence level | Yes | -- | -- |
| PICO framework | Yes | -- | -- |
| RoB 2 / STROBE / CONSORT | Yes | -- | -- |
| SR Screening PRISMA | Yes | -- | -- |
| BYOK zero markup | Yes | -- | -- |
| Drug interaction checker | Yes | -- | -- |
| Image generation | -- | Yes | -- |
| Voice / realtime mode | -- | Yes | -- |

### One account, every field

- **Medical & Clinical**: Drug checker, Clinical Mode, SR Screening
- **Research**: PRISMA, GRADE, Deep Research
- **Marketing**: Content, SEO, Social copy
- **Business**: Market research, SWOT, competitor
- **Education**: Quiz, flashcard, study guide
- **Indie Hacker**: Startup tools, analytics, growth

### Simple pricing: bring your key, pay less

Free with your own API key (BYOK). Pro is monthly, cancel anytime. Founder: pay once, Pro for life -- capped at 200 seats.

### Free: $0

Bring your own API key

- Your own API key, any provider (BYOK): no markup
- Research engine + Citation artifacts
- Keys encrypted with AES-256-GCM, delete anytime

### Pro: $10/mo

Monthly - cancel anytime

- Claude Opus 5, Claude Opus 4.8, GPT-5.6 Sol, Gemini 3.1 Pro
- Multi-agent Deep Research
- Clinical Answer Mode with citations
- Background research runs + Telegram bot
- Unlimited Brain, 200 MB files, 500 credits / month

### Founder: $59

One-time - Pro for life - 200 seats

- Everything in Pro, no renewals
- Capped at 200 seats, with a deadline
- Pay by card (or VietQR in Vietnam)

*Full refund within 14 days of your first paid charge, no questions asked. No refunds after 14 days.

Details: https://pho.chat/pricing

## Use cases

- **Clinicians**: Grounded answers with GRADE + real DOI sources for fast point-of-care lookup.
- **Researchers**: PICO, SR screening, PRISMA, multi-agent Deep Research -- faster systematic reviews.
- **Pharmacists**: Drug-interaction checks (Opus) and protocols with verifiable sources.


## FAQ

### What is the v2 launch offer?

Phở Chat v1 shipped and earned trust. v2 is a full rebuild: cite-or-abstain on every claim, Google Cloud infrastructure (Vertex AI), autonomous agents, and far more speed. The launch offer is a chance to join v2 early at a fixed price -- seats are limited. As Phở Chat grows, pricing will increase, but your rate stays locked at renewal. Roadmap is public at pho.chat/roadmap.

### I used Phở Chat v1 -- what is new in v2?

v2 upgrades what matters most: answers you can check against their sources, cite-or-abstain (claims without a source are labeled unverified), Clinical Answer Mode for medicine, multi-agent Deep Research, Phở Brain (living memory), BYOK zero-markup, and Google Cloud infrastructure. This is not a patch -- it is a new generation.

### Pro monthly or Founder lifetime?

Pro is billed monthly, cancel anytime. Founder is a one-time payment for Pro for life, no renewals -- only 200 seats and a deadline, see pho.chat/founder. We give at least 60 days notice before any price change.

### What is BYOK and is it safe?

Bring Your Own Key: plug in your OpenAI, Anthropic, Google... API key. Keys are encrypted AES-256-GCM, never appear in logs, and can be deleted anytime. Phở does not mark up -- you pay the provider directly.

### Is my data secure?

Yes. API keys are encrypted AES-256-GCM, never appear in logs, and can be deleted anytime. You stay in full control of your data -- with BYOK, your chat goes straight to the provider you choose. See our Privacy page for details.

### What are platform credits for?

Runs on your own API key cost no credits, Deep Research included. Background runs (Deep Research, systematic-review screening, agent sessions) need Pro or Founder.

### What if I don't have an API key?

Yes, for trying it out. Gemini 3.5 Flash-Lite and GPT-6 Luna are free within a monthly allowance, shown as a percentage. Every other model (Claude, GPT, Gemini Pro...) runs on your own API key, with no markup from Phở.

### What payment methods are accepted?

In Vietnam: VietQR bank transfer (SePay) or international card (Polar/Stripe). Internationally: credit/debit card.

### Can I cancel?

Yes, anytime. No penalties or questions asked. Note: outside the 14-day full refund from your first paid charge, there are no refunds for unused time. After cancelling, you keep full access until the end of your paid period, then automatically move to the Free plan.

### Do you offer a refund?

Yes. Full refund within 14 days of your first paid charge, no questions asked. After 14 days, your plan runs to the end of its cycle and will not renew if you cancel. Exception: if the service has a major incident lasting more than 24 hours, we proactively issue credit or a pro-rata refund. See Terms of Service § 4.

### Does Phở Chat fabricate sources?

No. We only cite sources we actually find and verify -- real DOIs, PubMed IDs, or valid URLs. If no source is found for a claim, we label it "unverified" rather than inventing one. This is the central promise of Phở Chat.

### What is Phở Chat's long-term commitment?

We are building Phở Chat to last. For full transparency: if there is ever a major change, we commit to at least 90 days notice and pro-rated refunds on any unused annual plan. BYOK means your AI traffic goes straight to the provider -- no dependency on Phở Chat's infrastructure. Conversation history can be exported at any time.


## Related
- [Pricing](https://pho.chat/pricing)
- [Founder Lifetime](https://pho.chat/founder)
- [Roadmap](https://pho.chat/roadmap)
- [Tiếng Việt](https://pho.chat/release)

---
## About Phở Chat

Phở Chat is an AI workspace for clinicians, pharmacists and researchers: answers cite real sources (DOI, PubMed), and claims without a source are labeled "unverified".

- App: https://pho.chat/
- Pricing: https://pho.chat/pricing
- FAQ: https://pho.chat/faq
- Phở Labs (research, benchmark): https://labs.pho.chat/
- Support: support@pho.chat