Answer from approved facts
Teach an agent from your pages and documents. Recognized personal details are removed from uploaded knowledge.
Trust, in plain words
Know what your AI employee does—and stays away from.
NetShow gives your business a live AI host while keeping people in control. Every promise below is true today; each one links to the place we explain it.
You’re talking to an AI agent: Mrs. NetShow, in the corner of this page, says so first.
Useful, with boundaries
Three ways NetShow earns trust
The host shows what it can do, asks before sensitive actions and leaves a clear next step.
Ask before it sees or changes
Camera and microphone access need browser permission. Form filling and other changes wait for consent.
Let a visitor leave cleanly
A visitor can say “Forget me” and start the removal of their visitor history, conversations and leads.
The real product
See the controls people actually use
These are current NetShow screens, not concept art. The live host on this page wears the same Alive Kit used across the product.
One platform
How trust connects to NetShow
The same consent, disclosure and removal promises follow the host from your website to its public agent page. Meet Mrs. NetShow here, explore the Alive Kit, or read the privacy policy.
Honest monthly starting prices
Choose the size of the job
AI Front Desk
from $699 a monthAny business that wants every visitor and caller answered.
AI Team
from $4,999 a monthGrowing businesses with more than one job to hand over.
Enterprise AI Workforce
from $9,999 a monthCompanies that need their AI employees on their own infrastructure.
Promises you can check
What NetShow protects today
Our agents say they are AI
Our agents are AI, not people. They are told to say so whenever you ask, and never to claim to be a person.
The microphone and camera start only when you tap
Your browser asks before our agent can use your microphone or camera, and tapping is your OK for that conversation only. We don't keep the recording: what you say is kept as text, like a typed message, and a camera picture is used to answer and not stored.
Ask an agent to forget you
Where an agent chat shows Forget me, one tap queues removal of this browser's visitor records and lists what it covers; removal finishes afterward. You can also email [email protected] to have what we keep deleted.
Global Privacy Control and Do Not Track are honoured
If your browser sends Global Privacy Control, we treat it as your request to opt out, and we also honour Do Not Track: our avatars stop remembering you between visits.
We do not sell your personal information
We do not sell your personal information, and we do not share it for cross-context advertising.
Every page travels encrypted
Every page is served over HTTPS, and browsers are told to use nothing else (HSTS), so what you type travels encrypted.
Personal details are removed from the knowledge you give an agent
When a site agent is built from a knowledge file you supply, recognized personal details in that file are replaced before its text joins the agent's knowledge.
A copied voice can always be taken back
Voice copies made in NetShow's voice studio need the voice owner's agreement first, and the owner is given a take-back code. With it they can switch the voice off at any time, without an account: every agent goes back to the NetShow voice and the copy is deleted at Google.
A real person's face needs their OK
A face uploaded to make an agent is bound to consent. If it shows a living person other than the owner, that person approves or refuses through a one-time link, and until they approve the face is not shown in public. Either side can take it back.
Agents never keep your secrets
When an agent keeps notes from a conversation, it refuses to remember passwords, one-time codes, card numbers or recovery phrases, and one owner's memories are not read by another owner's agent unless the owner shares them.
Agents never post to social media on their own
Real social posting is one of our locked switches. While it is locked, a post an agent writes can only become a draft for a person to review.
Skills are scanned before they are listed
Every skill submitted to our marketplace is scanned for dangerous instructions before anyone can add it to an agent, and a skill with a serious finding is refused.
A second check reads what an agent does
A small checking model reads each reply and each step an agent wants to take, with personal details removed first; a reply that breaks our rules is replaced with a safe answer, and a risky step needs a higher level of permission.
Agents only call public web addresses
Before NetShow calls an address given to an agent (its own endpoint, an MCP server or a harness gateway), it checks that the address is public, so an agent cannot be pointed at private networks.
Understand the safeguards
Explore how the safeguards work. Optional protections depend on the site’s settings. Ask the host to explain any topic.
Share an agent more safely
A pass for an embedded agent works only for that agent and expires. It does not give a visitor your signed-in owner session.
Keep business sites safely separate
Before the factory works on a site, it checks the list of sites it must leave alone. Listed sites and sites beneath those domains are refused.
Keep outside text as information
When this protection is enabled, visitor messages and text from elsewhere are kept separate from the agent’s instructions.
Keep your records your own
When a record is looked up for an owner, another owner’s record is refused the same way a missing record is.
Check links before fetching them
When this link check is used, an agent cannot fetch a link that points to a local or private network.
Limit what visitors can request
When visitor action limits are enabled, actions beyond reading and drafting need the owner’s confirmation. Limited contact requests are allowed separately.
Give agents clear ground rules
When the shared ground rules are enabled, agents receive instructions to be honest about being AI, respect consent and follow the right person’s instructions.
Find nearby context more privately
The local place option uses a location database on the server. It does not send a visitor’s network address to an outside location service, and leaves place details empty when it is unavailable.
Protect the owner’s agent pages
The listed agent pages check that the signed-in person owns the agent. Signing in does not give you another owner’s agent.
Set boundaries for your agent
Agent guardrails check what people send and what the agent says. The enabled checks and their failure settings determine whether a turn can continue.
Keep agent access keys private
Access keys let approved software reach an agent. Only the owner’s protected controls manage them; this guide never shows or creates a key.
Keep permission tied to context
A conversation can keep a record of permission and what it covers, so permission can be checked for a particular activity.
See a record of checks
Receipts record checks that ran. When the receipt page is enabled, signed-in owners can review their own records.
Understand records of privacy actions
Privacy records track requests to erase information and changes to what is public. Reading this guide does not erase anything or make an agent public.
Report an unwanted character likeness
You can report a character that uses your work without permission. The report goes for review; only the platform owner can decide to hide the character.
Protection behind the conversation
These protections are built into NetShow. Availability depends on each feature's settings; this guide does not report which safeguards are enabled for your agent.
Safer messages
Keep automated signups in check
Configured signup and public forms check a Turnstile or reCAPTCHA response before accepting it.
Keep the strongest safety finding
A second input check can strengthen the first safety finding, using text with personal details removed, without weakening an existing refusal.
Pause public turns when checks fail
Public conversations pause when their safety checks cannot complete.
Check content before a reply
Messages can be checked by a moderation provider, with a local safety check when that service cannot answer.
Screen lessons before they are reused
Learned lessons are checked for unsafe instructions, personal details and secret values before an agent uses them again.
Keep chat use within its limits
Chat use follows the configured free-demo and internal-use limits.
Private details
Hide details between cooperating agents
Recognized personal details are removed from task messages before agents share them.
Hide private details in activity records
Sensitive tool inputs and visitor details are hidden in activity records when this privacy check is enabled.
Leave private material out of exports
Agent-look exports are refused if they contain private instructions, memories, conversations or secret fields.
Remove a person's records with proof
An operator can remove covered records for one person, record what was removed and block those records from being added again.
Permission to act
Keep agent creation under ownership
Creation safety checks keep a business and its new agents tied to their owner and keep files and secret values out of public view.
Keep delegated work within permission
The delegated authority check keeps team tasks within the tools their permission allows.
Professional boundaries
Know the limits before hiring
Legal and care hire pages explain that the AI is not a licensed professional.
Keep professional advice within bounds
The regulated-field guard adds clear limits to agents working in law, medicine, mental health, finance or licensed trades.
Suggestions you control
Review suggestions and action receipts
Owners can review proposed check-ins and records of agent actions from their existing pages.
Suggest check-ins only with consent
Check-in suggestions need the current owner's permission and respect quiet hours and other limits.
Checks behind the scenes
Keep a record of removals
Completed agent removals leave a record when its storage is available; memory removal and publishing have their own records.
Spot repeated failures behind the scenes
The operator's error check spots repeated failures and signs of attacks for the existing error report.
Check what record removal covers
Erasure tests check that covered records stay removed and show the known gaps.
Track where records need removal
A coverage list records where each kind of data is held and what removal can and cannot reach.
Questions? [email protected]
Start free