Safebots · For podcasters · Live hosts · Presenters

Give your show
a silent producer
with superpowers.

Safebots listens to your conversation and puts real-time fact checks, charts, profile cards, and archive clips on screen — without interrupting you. You approve. It shows. No one knows how you did it.

See what it does Watch the demo

I · The moment

Your guest says something.
A card appears.

You didn't touch anything. You didn't stop talking. Your guest didn't notice. But the audience — on Zoom, on your livestream, on their phones — just saw a sourced, animated stat card appear on screen with the verified number, the year-over-year delta, and the Bloomberg citation.

That's Safebots. A live AI intelligence layer that watches your conversation, proposes visualizations, and only puts them on screen after you approve — silently, without anyone knowing you're involved.

Your phone is already the remote

Open the URL on your phone. The shared screen follows every scroll, every tap, every voice command. No app. No Bluetooth. No cables. Just a URL and a WebSocket.

Scroll a PDF — the projected screen scrolls identically. Advance a slide — 500 participant phones update in the same frame. Pause a video — it pauses everywhere. The audience follows on their phones too.

$312B ▲ +65% AI INVESTMENT 3s CONTROL ▶ Next slide directions to Bob's ✓ Commit card Type or speak... ⟳ WebSocket live 500 phones follow simultaneously
safebots.ai/intelligence
$312B ▲ +65% GLOBAL AI INVESTMENT 2024 Bloomberg Intelligence 3s INTELLIGENCE FEED AI PROPOSAL Stat card: Global AI investment $312B in 2024, up 65% YoY source: Bloomberg · auto-commits in 3s ▶ Stat card shown — 14:32:08 QUEUED · AWAITING APPROVAL Profile: Sam Altman mentioned OpenAI CEO · @sama · confidence 0.97 PRIVATE · HOST ONLY Contradicts guest's claim from March interview (ep. 47, 23:14) HOST TYPED directions to Bob's house Type a query or voice command... SEND ✓ COMMITTED TO SHARED SCREEN · 14:32:11

II · The experience

What the audience sees.
What only you see.

<3s Claim to sourced card on screen
0 Times you have to touch anything
500 Phones update simultaneously
100% Cards you didn't approve never appear
🎙 "...autonomous vehicles and LIDAR sensors..." KEYWORD EXTRACT "autonomous vehicle lidar" PEXELS API 8 images BACKGROUND UPDATED Kenburns gallery swaps in 1.2s · No AI needed BROWSER SPEECH RECOGNITION → PEXELS / PIXABAY · ZERO AI TOKENS · IOS SAFARI ✓ With AI: "autonomous vehicle lidar sensor at night urban" — not just "vehicle"
🎙️

Someone mentions a number

AI detects it, verifies it against live web data, animates a stat card with source and delta. In the time it takes to finish the sentence.

👤

Someone mentions a name

Profile card: bio, handle, topic tags. Pulled from your graph if they're in it. Searched from public data if they're not.

📖

A technical term drops

"LIDAR." "ZK-proof." "Federated learning." Plain-English definition appears, calibrated to the conversation's evident expertise level.

📰

A paper is referenced

The AI identifies it, surfaces headline, key claim, direct link. Participants on their phones can tap through immediately.

⚖️

"How does that compare to Google?"

Comparison card. Both sides. Sourced from public data. You don't need to have memorized it.

🗺️

You type "directions to Bob's house"

If you have access to Bob's location in the graph, a map appears. If the other person tries the same query, they see a private error. Only you see the map.


III · The veto

Every card goes through you.
Silently.

Before anything appears on the shared screen — the Zoom, the livestream, the projector — it appears on your private dashboard with a countdown ring. Say "no" or "skip." Say nothing and it commits. The audience sees only what you approve.

CONVERSATION AI listens PROPOSAL Card generated VETO WINDOW Host sees it first VETOED → gone COMMITTED Shared screen AUDIENCE Sees card

"The host appears effortlessly informed. The audience sees nothing unusual. The AI did the research; the host did the judgment."

You also get private coaching the audience never sees. "This contradicts what the guest said in March." "This figure is lower than what they published last quarter." Sourced from your archive. Delivered only to your screen. The audience has no idea why you look so prepared.


IV · Access control

The system knows who is speaking.
Without analyzing the audio.

Each person in the conversation streams their own microphone over their own authenticated connection. The server knows whose voice it is from the connection — not from trying to figure it out acoustically. So permissions are cryptographic, not probabilistic.

Host — authenticated as Robert
🎙️ "Give me directions to Bob's house"
✓ Bob's home: 340 W 57th St, NYC
Map card queued → 8 min drive
🗺️ Map card committed to shared screen
Guest — authenticated as Greg
🎙️ "Give me directions to Bob's house"
✗ Access denied: Places/user/location/home
Greg Magarshak does not have read access
to this stream. Bob has not shared it.
This error is private. Shared screen unchanged.

The AI pipeline never sees data the requester doesn't have permission to access. The access control isn't a prompt instruction the model could decide to ignore — it's enforced at the data layer before anything reaches the model. Bob never gave Robert access to his home address. The system reflects that fact structurally.

The same applies to navigation commands. Voice commands from a guest don't advance your slide. "Next slide" spoken by you advances the slide. Spoken by your guest, it's ignored — silently, with no visible awkwardness.


V · Without any AI at all

These capabilities work
before the AI even loads.

Every layer is independently useful. The AI enhances what's already there. None of it requires the layer above it.

LAYER 4 · SAFEBOX INTERNAL DATA Access-controlled graph · Speaker-identity permissions · Archive · Audit trail · Cryptographic provenance MOST POWERFUL LAYER 3 · AI + PUBLIC DATA Fact verification · Profile cards · Glossary · Article surfacing · Comparison cards · Semantic archive search LAYER 2 · VOICE NAVIGATION · NO AI "Next slide" · "Zoom in" · "Highlight the top bar" · <10ms · Pattern file · Works offline LAYER 1 · SYNCHRONIZED DISPLAY · NO AI Phone-as-clicker · PDF scroll sync · Webpage sync · Live polling · 500 phones update simultaneously ALWAYS WORKS EACH LAYER ADDS VALUE INDEPENDENTLY — NOTHING REQUIRES THE LAYER ABOVE IT
Capability No AI With AI (public data) With AI + internal data
Phone as clicker — scroll PDFs, advance slides ✓ Works now
Webpage sync — scroll a website on your phone, shared screen follows ✓ Works now
Live polling with QR code — audience answers on phones ✓ Works now
Audience follows along on their phones in real time ✓ Works now
B-roll gallery — images matching conversation keywords ✓ Browser speech, no API key ✓ More precise queries
Voice navigation — "next slide", "zoom in", "highlight the top bar" ✓ Pattern matching, <10ms
Real-time CSS theming — change colors mid-show ✓ Works now ✓ AI detects register shifts
Fact verification — numbers checked and sourced ✓ Web search
Profile cards — bio, handle, tags for named people ✓ Public data ✓ Internal graph
Private coaching — contradictions, corrections, context ~ Limited ✓ Full archive
Access-controlled data — different content per viewer ✓ Substrate enforced
Directions to "Bob's house" — resolves who Bob is and if you have access ✓ Per speaker identity
Audit trail — what appeared, when, sourced how, approved by whom ✓ Cryptographic

VI · The archive

Your back catalog.
Instantly searchable.
Surfaced mid-conversation.

A Safebox workflow has already ingested your channel — every episode, every video, every audio file. It transcribed the audio, indexed it by topic and keyword, and cut it into clips. Now the AI knows what you've covered before, and can surface exactly the right moment from your own archive while you're live.

safebots.ai/intelligence — archive results
LIVE TRANSCRIPT · KEYWORD MATCH Guest: "...the thing about transformer architecture is that attention heads scale quadratically with..." KEYWORD SEARCH · NO LLM "attention heads" → 7 clips found SEMANTIC SEARCH · WITH LLM "scaling laws for transformers" → 12 clips TOP RESULTS Ep. 184 · Karpathy on attention 23:14 · "heads scale quadratically" ▶ PLAY Ep. 171 · Scaling laws deep dive 8:42 · "the Chinchilla result showed..." ▶ PLAY CLIP CARD · QUEUED FOR VETO EP 184 · 23:14 TRANSCRIPT EXCERPT · EP 184 "...the attention mechanism means that every token attends to every other token, so you get O(n²) compute — that's why attention heads scale so... SHOW ON SCREEN SKIP PRIVATE
🎬

Every episode, already indexed

A Safebox workflow ingests your channel in the background — transcribing audio, extracting topics, cutting clips at natural break points. Done once. Available forever.

🔍

Keyword search, no AI required

"Attention heads" appears in the live transcript. The archive returns matching clips instantly — pure keyword match, no model call, no latency. Works offline.

🧠

Semantic search when it helps

With the LLM active, "scaling laws for transformers" finds the clip where you discussed Chinchilla even if neither phrase appeared verbatim. The keyword index is the fast path; the LLM is the smart path.

⏱️

Surfaced at the right moment

The AI detects when the live conversation touches something you've covered before. The clip appears in your private veto queue with the relevant timestamp and transcript excerpt. You decide whether to play it.

▶️

One tap to play on screen

Tap "Show on screen." The clip starts playing on the shared display — with the relevant section already cued. Tap "Private" to send it to a specific participant. Tap "Skip" and it's gone.

✂️

"We talked about this in episode 184"

You say it. The AI finds it. The clip appears. You play it. The audience sees continuity across your entire back catalog. No producer needed. No prep required.

"The guest says something you covered six months ago. The clip appears in your queue at the exact timestamp. You tap play. The audience thinks you have a perfect memory."

Archive capability How it works Requires LLM?
Find clips by keyword from live transcript Inverted index, sub-100ms lookup No
Find clips by semantic meaning Embedding similarity search Yes
Detect when current conversation matches prior episode Keyword overlap + topic graph Optional
Surface clip with correct timestamp cued Clip boundary metadata from ingest No
Transcribe new episodes automatically Safebox ingest workflow (Groq Whisper) No
Surface contradictions between guest's claim now vs 6 months ago LLM cross-references session entity graph with archive Yes
Access control — some clips private to host only Stream-level permissions, substrate enforced No

VII · Live charts

Data visualized
the moment it's relevant.

When the conversation touches a trend, a comparison, or a claim — Safebots can pull the data and render a chart on the shared screen in seconds. From public sources, from an ingested knowledge base, or from private data the requester has access to. The chart appears. The host decides whether the audience sees it.

Public data · Web search

AI investment by year

Guest mentions a funding figure. Safebots queries public data, renders this.

200B 100B 0 2020 2021 2022 2023 2024 $312B Source: Bloomberg · pulled live during conversation
Private knowledge base · Aligned News

AI safety coverage volume

From your ingested archive of tracked sources — not on the public internet.

400 200 Jan Feb Mar Apr May Jun 390 articles Source: Aligned News archive · ingested nightly
Ingested Twitter/X data · proprietary

AI researcher tweet sentiment

From a tracked list of researchers ingested into your knowledge base — not a public API call.

50% 0% Positive Neutral Concern Safety Capabilities Regulation Source: tracked researcher list · ingested weekly via Safebots
Mixed sources · public + private

Coverage by AI topic this quarter

Blends public search data with your privately ingested source archive.

Agents 2,840 Safety 2,210 Reasoning 1,890 Multimodal 1,610 Regulation 1,260 Hardware 860 private archive public sources Article count Q1 2026 · Safebots + web search

These charts are generated live. The host says "show me AI investment by year" — or the AI detects the topic from the conversation and proposes the chart without being asked. The host approves. It appears on screen.

🌐

Public internet

Web search grabs the current figure, structures it into chart data, renders it in under three seconds. Bloomberg, Statista, research papers — wherever the data lives publicly.

📰

Your ingested sources

A Safebots workflow has already indexed the outlets you track — Aligned News, specific newsletters, RSS feeds. Trend data from your archive that isn't anywhere on the public internet.

🐦

Ingested social data

Twitter/X lists, LinkedIn signals, community channels — ingested on a schedule into your private knowledge base. Sentiment trends across a tracked researcher list. Topics your community is discussing this week versus last month.

🔐

Private databases

Internal CRM, subscriber data, proprietary research — anything your organization has ingested and access-controlled. The chart appears only if the requester's identity has read access to the underlying streams.

🔀

Blended sources

One chart, multiple sources. Public web data for the industry baseline, your private archive for your own coverage, ingested social for community signal — all normalized and combined.

🎙️

Voice or type to generate

"Show me AI funding trends" — spoken or typed quietly while your guest is talking. The chart proposal arrives in your veto queue. You approve with one tap. The audience sees a sourced chart that appeared from nowhere.

"The guest mentions a figure. You say nothing. A sourced bar chart appears on screen behind you. The audience assumes you prepared it. You didn't — Safebots did, ten seconds ago."


VIII · The control surface

You can type it
or say it.

The control panel is the same chat the AI uses to show proposals. You type a query mid-conversation — quietly, while your guest is talking — and a card proposal comes back. You approve it. It appears. No one knows you asked.

⌨️

Silent typed queries

Type "what's Anthropic's current valuation" while your guest is mid-sentence. A stat card appears in your veto queue 2 seconds later. Commit it or don't.

🎤

Voice without AI overhead

"Next slide." "Zoom in." "Highlight the top bar." These resolve in under 10ms from a vocabulary file — no model call, no latency.

🔘

Inline options

When the AI finds three angles on a comparison, it posts them as tappable buttons. You pick one. That card gets built and committed.

📱

Side-by-side on your stream

Show the presentation alongside the intelligence feed on your Zoom or livestream. Viewers see the AI proposals, the host approvals, the access errors. Governance as theatre.


IX · The demo

What this looks like
on a real show.

"Robert is the host. Greg is the guest. You say something. A sourced card appears. You try to look up something private. It says no. The audience sees all of it. No one planned any of it."

— The Safebots demo, May 25 2026

Two screens side by side. Left: the presentation — b-roll running, cards floating over it. Right: the intelligence feed — proposals arriving, Robert approving some and discarding others, private coaching appearing only for Robert, access errors appearing only for Greg.

The moment that lands: Greg says "give me directions to Bob's house." An error appears in his chat — Bob never shared his address with Greg. Robert says the same words. A map appears on the shared screen. Same words. Different identity. Different outcome. No one wrote a rule about it. The data structure decided.

Try the demo    About Safebots