AI Search
AI Search with Openable Sources — Free to Start, No Card
Ask a question and get a synthesized answer built from a live web search — not a page of blue links, not training memory. Runs in Telegram via @vustSearchBot: no foreign card, no extra apps. Four depth tiers, from a quick lookup to agent-driven research.
Cited Sources
How do I find answers with cited sources using AI search?
Open @vustSearchBot in Telegram and pick the Research mode. It runs several search and page-read steps behind one question, then returns a synthesized answer carrying numbered [N] markers (the excerpt below shows [1] and [2]) plus a Sources list of the real pages behind each claim, so you can open them before you reuse anything. Quick and Standard drop the inline [N] markers for speed and a lower price — you still get the source list; only Research ties each claim to a numbered source.
Checklist for a source-backed answer
| Step | What to do |
|---|---|
| 1 | Open @vustSearchBot and choose the Research mode |
| 2 | Ask one clear question, naming the entity and year |
| 3 | Read the [N] markers, then open the Sources list |
| 4 | Open 2–3 linked pages and check they support the claim |
Excerpt from a real Research-mode answer (condensed; sources renumbered)
Query: “How does the EU AI Act classify high-risk AI systems?” — Research mode, captured September 4, 2026
EU AI Act classification is primarily use-case based, not technology based. An AI system is high-risk if it is either embedded in, or itself constitutes, a regulated product requiring third-party safety assessment, or if it is used in one of the high-impact areas listed in Annex III. According to Article 6 of Regulation (EU) 2024/1689 [1], both conditions — use as a safety component and a required third-party conformity assessment — must be satisfied for that route. Annex III systems may escape the classification in limited circumstances, but profiling of natural persons always makes an Annex III system high-risk. The European Commission also publishes practical guidance on its high-risk AI systems page [2].
Sources
In the chat these arrive when you tap Src under the answer. This answer’s full source list ran to 15 links — 3 are shown here.
- 1. Article 6 of Regulation (EU) 2024/1689 — EUR-Lex (cited as [1] above)
- 2. European Commission — guidance for high-risk AI systems (cited as [2] above)
- 3. Annex III: High-Risk AI Systems Referred to in Article 6(2) (also in this answer's full source list)
Ask your own question the same way — open Research mode in @vustSearchBot.
Search Tiers
Four depth levels — pick the one that fits your question
Quick and Standard send a single query and return a synthesized answer without inline [N] markers — the source list is still there. Research runs several sub-queries and returns numbered markers with openable links. Deep Report is a Pro-subscription tier under live cost and safety guardrails; it deploys an agent that browses pages and compiles a research report.
See the difference
Real questions, real answers. Each tier goes deeper.
What people actually use it for.
- Researcher
- Fact-check a claim
- Research tier answers with numbered markers and openable sources.
- Support engineer
- Trace an API change
- Quick or Research tier depending on how much context you need.
- Shopper
- Compare product categories
- A synthesized comparison; pick Research when you want the links behind it.
Three steps in Telegram.
- 01
Open SearchBot
Start a chat with @vustSearchBot — search runs in Telegram, not on this page.
- 02
Pick a tier
Quick, Standard, Research or Deep Report — depth and cost scale with the question.
- 03
Read & refine
Open the sources on Research, then ask follow-ups in the same thread.
SearchBot is where search runs — cited answers in chat.
Four depth tiers, follow-ups, shared VUST Balance.
Built for clarity, not hype.
Where the answer comes from
Every tier searches the live web at query time rather than answering from training memory.
Tiers
Quick and Standard return a synthesized answer without inline [N] markers, though the source list is still one tap away. Research returns numbered [N] markers with openable links. Tier prices are fixed and known before you send; Deep Report is included with the Pro subscription under live cost and safety guardrails, not sold as a one-off.
Privacy
Queries are summarized in logs — raw text stays out of admin surfaces.
Frequently asked questions
What are the search tiers?
Four, trading speed against depth. Quick and Standard send a single query and return a synthesized answer without inline [N] markers. Research runs several sub-queries and synthesizes them into one answer with numbered [N] markers and openable source links. Deep Report is an agent-driven investigation with page fetching; it is included with the Pro subscription under live cost and safety guardrails rather than sold as a one-off.
Which tiers actually show sources?
All four tiers can show a Sources button when the answer has web results behind it. Research and Deep Report go further and tie each claim to a numbered [N] marker in the text; Quick and Standard drop those inline markers — they are the fast, cheap lookups. We would rather say that plainly than advertise inline citations on a tier that does not return them.
How is this different from Google or ChatGPT?
Unlike Google, you get a synthesized answer rather than a page of links. Unlike a chat model answering from training memory, the answer is built from a live web search — and on Research you can open the sources behind it and check them yourself.
Is it free?
There is a free way in: Quick and Standard each carry their own allowance of 5 free searches a day — ten a day in total, on separate counters — once you join the channel and start @vustbot, and new users get a welcome bonus on Balance. Research can be tried free once, ever. Beyond that, Quick and Standard cost $0.05 a query and Research $0.10. The tier price is fixed and known before you send — units are taken from Balance when the query runs, and a failed search is refunded.
What is Deep Report mode?
An agent-driven search that browses the web autonomously — fetching pages, following leads, and compiling findings into a structured report with sources. Best for questions that need multi-step investigation. It comes with the Pro subscription under live cost and safety guardrails and a daily cap, not as a pay-per-query tier.
Do I need any extra apps or a foreign card?
No. It runs inside Telegram via @vustSearchBot — no extra apps, no foreign card, no separate signup. Balance is topped up inside Telegram after the exact amount is shown. This is the main reason people in card-restricted regions use it.
Ready when you are
Ask with sources — not just blue links.
Open SearchBot in Telegram to run your first search.