Emma
Emma is a no-code platform for building custom AI assistants connected to your company's own data through 24 named integrations, uploaded documents or manual content, and reached from a chat page, desktop app or API.
What is Emma?
Emma is a no-code platform for building AI assistants that answer from a company's own information rather than from the open web. You create an assistant, point it at data you already hold, and then talk to it from a chat page, a desktop application, a connected tool or the Emma API.
The site frames three uses. A personalised assistant connects to business data and integrations so that people stop hunting through folders or interrupting colleagues. A help-desk chatbot is stood up quickly by linking data sources, spreadsheets or uploaded documents. A trained assistant covers the case where there is no data to connect at all: Emma interviews you about what the assistant should know, and it learns from your answers.
Data arrives three ways. Twenty-four integrations are named on the integrations page, among them Shopify, Google Sheets, Excel, Notion, Zendesk, Intercom, Coda, Mailchimp, Salesforce, Jira, Gmail, Google Calendar, Stripe, Webflow, Todoist, GitHub, Google Analytics, Trello, Google Forms, Calendly, Airtable, Typeform, Asana and Google Contacts; the homepage adds Slack, Zapier and email software. You can also upload documents or paste content in by hand.
The organising idea is one assistant per team, project or workflow, each with its own scope: access to a source can be added or removed at any time, so an accounting assistant and a marketing assistant need not see the same things. Assistants can stay private or be published publicly, and their chat page takes a custom image, cover image and colour, or can be replaced by your own front end over the API. Developers can go further with the Emma SDK, published on npm, which builds an integration against any API in JavaScript from the browser.
Six families of use case are shown: finance and accounting, marketing, sales, HR and recruiting, customer support, and engineering. The engine is described as OpenAI technology, GPT-3.5 in the homepage copy and GPT-4 in the page title. The homepage also claims a 90% efficiency gain, a figure the site never substantiates. The interface is English only.
What it does
- Answer questions about your own business data without a manual search or a message to a colleague
- Run a help-desk chatbot, kept private for your team or published for your customers
- Read from 24 connected tools, uploaded documents, spreadsheets or content you paste in by hand
- Restrict, source by source, what each assistant is allowed to see
- Run a separate assistant for every team, project or workflow
- Draft blog articles, newsletters and advertising copy from your own business data
- Answer inside your own website or application through the Emma API
When to use Emma / When not to
A quick filter to help you decide if Emma is the right fit.
When to use Emma
- Support teams that want a help-desk chatbot fed by their own Zendesk or Intercom knowledge
- Operations, finance and HR teams whose answers sit scattered across spreadsheets, Notion pages and business tools
- Sales and marketing teams already working inside Salesforce, Mailchimp, Shopify or Google Analytics
- Solo founders and freelancers who need one assistant on a small budget, on the single-seat Starter tier
- Approved educational organisations, which can apply for a free Emma account
When not to use Emma
- Organisations with a compliance review to pass: the privacy policy page is published but empty, and nothing on the site mentions the GDPR, a data processing agreement or subprocessors
- Teams that need a mobile app: only a web chat page and a desktop application are offered, with no iOS or Android release
- Heavy conversational use on the entry tiers, capped at 150 and then 650 messages in total
- Individuals looking for a personal assistant: the terms require business use and set a minimum age of 18
- Buyers who need self-hosted deployment or a data residency guarantee, neither of which is offered or documented
How to use Emma
A typical end-to-end flow, from setup to results.
- Create an account on the Emma sign-up page; the terms ask for your full legal name and a valid email address
- Set up your organisation, which requires its location details
- Create your first assistant and give it a purpose
- Connect the data: pick from the 24 catalogue integrations, upload documents or spreadsheets, or paste content in manually
- If you have nothing to connect, let Emma interview you and learn what the assistant should know
- Set the scope by adding or removing access to each source, assistant by assistant
- Customise the chat page with an image, a cover image and a colour
- Decide whether the assistant stays private to your team or is published publicly
- Talk to it from the chat page, the desktop application or a connected integration
- For a connector that is not in the catalogue, use the Emma JavaScript SDK from npm and request access through the contact form
Pros & Cons
Pros
- Set-up announced in five minutes, with no code required for the standard path
- Twenty-four named integrations across e-commerce, CRM, support, spreadsheets, project management, payments and analytics
- Data access can be scoped source by source and assistant by assistant
- Unlimited end users on every tier, including the cheapest
- Public API and a JavaScript SDK openly published on npm
- Low paid entry point at USD 12 per month, with a three-day trial on the Starter tier
- Free accounts for approved educational organisations of any type
Cons
- The privacy policy page is published and linked from every footer, but completely empty
- The terms of service stop after Account Activation: no governing law, termination, refund or security clause
- No mention anywhere of the GDPR, a data processing agreement, subprocessors or data retention
- No postal address and no email address; the only way to reach the company is a Coda form
- Message quotas are tight on the entry tiers, at 150 and then 650 messages in total
- The site contradicts itself: GPT-3.5 in the homepage copy against GPT-4 in the page title, and unlimited assistants in the FAQ against 2, 5 or 12 in the pricing grid
- No mobile application, and both the help centre and the SDK documentation live off-domain on Coda
Pricing & Plans
Emma publishes no permanent free plan open to the general public. The lowest paid entry point is USD 12.00 per month for the Starter tier, billed monthly in US dollars and preceded by a three-day free trial. Free access exists in two restricted forms only: approved educational organisations of any type may apply for a free account, and the pricing page announces a free exclusive beta program.
- 1 team member
- unlimited users
- unlimited integrations
- 150 messages in total
- 2 assistants
- three-day free trial
- 3 team members
- unlimited users
- unlimited integrations
- 650 messages in total
- 5 assistants
- 5 team members
- unlimited users
- unlimited integrations
- 1
- 650 messages in total
- 12 assistants
- unlimited team members
- users
- integrations
- messages and assistants
- Education
- free
- for approved educational organisations of any type
- granted on application
Data, GDPR & hosting
A consolidated view of how Emma handles your data.
GDPR overview
There is no GDPR implementation to describe here, because there is no statement to read. Across every page collected, the words GDPR, CCPA, SOC 2, ISO 27001 and HIPAA never appear once. The privacy policy is the clearest illustration: the page exists, it is linked from the footer of every page, and its body holds nothing but two headings. No data protection officer is named, no privacy contact address is published, no Article 27 representative is designated for the European Union, no data processing agreement is offered and no list of subprocessors exists. The terms of service themselves stop after the Account Activation section, so they add nothing on confidentiality or security. For a European buyer this is an absence of information, not a claim of compliance.
Who owns the data?
Emma's terms define your Materials broadly: trademarks, copyrighted content, product descriptions and prices, images, video, audio, code and any other data you or your affiliates supply. What they never say is who owns them. There is no assignment clause, no licence granted to Emma, and no statement that you keep ownership. They are equally silent on whether your content may be used to train models, and the privacy policy page, although published and linked in every footer, contains no text at all. The homepage promises per-assistant control over which sources an assistant may read, but that is a product setting, not a contractual commitment. The terms do warn that your Materials may travel unencrypted.
Reuse rights
The terms say what you may not do with Emma, never what you may do with what comes out of it. You are forbidden to reproduce, duplicate, copy, sell, resell or otherwise exploit any part of the Service without Emma's explicit written permission, to work around its technical limitations, to decompile or reverse-engineer it, and to access or monitor it with robots, spiders or scrapers. No clause grants or restricts a licence on the answers your assistant produces, so nothing published on the site tells you whether that output can be reused or resold. Treat the silence as an open question to raise before you build a commercial workflow on it, not as permission.
Hosting summary
Emma publishes nothing about where customer data is hosted. No jurisdiction is stated, no hosting region is offered, there is no European data residency option and no security or trust page exists. The privacy policy page is empty and the terms of service say nothing about hosting. The only location signal comes from a network probe run for this review, and it describes the website rather than the data store: the domain resolves to 44.217.184.119, an Amazon Web Services address geolocated in Ashburn, United States. The company calls itself Emma Inc. in its terms without giving an address, and the domain is registered through GoDaddy.com, LLC in the United States. Taken together these point to a United States footprint, but none of it is a vendor commitment. Anyone with a data residency requirement should treat hosting as undocumented and ask directly.
Things to keep in mind
Risks and trade-offs to weigh before adopting Emma.
- Connecting Shopify orders, Salesforce records, Stripe payments, Gmail or Zendesk tickets hands customer and financial data to a vendor whose privacy policy page is empty
- Nothing is published about how long your data is kept, whether it is anonymised, or how to have it deleted
- No documented way to exclude your content from model training, and no stated vendor position on the question
- The terms warn that your Materials may be transmitted without encryption
- A public assistant with a badly drawn scope can expose internal information to anyone who opens its page
- Leaning on an assistant instead of colleagues can quietly erode a team's own grasp of where its facts come from, and the answers come from a language model with no published accuracy guarantee or service level
- With no postal address, no email address and truncated terms, exercising your rights or raising a dispute has no obvious channel
Setup & Integrations
Technical difficulty
Low on the standard path. Emma advertises five minutes and three steps: create an account, set up the assistant, connect the data. Catalogue integrations, document uploads and manual content need no technical skill, and the chat page is customised without code. Two things raise the bar: embedding the assistant in your own site or application requires the Emma API, and a connector outside the catalogue means writing JavaScript against the Emma SDK, distributed on npm, with access granted on request. Nothing has to be installed or self-hosted.
Deployment
Integrations
Behind Emma
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What does Emma actually do?
How many assistants can I create?
Can I control what an assistant is allowed to read?
What does it cost?
Which tools does it connect to?
Is there an API?
How long does set-up take?
What does the privacy policy say?
Is there a mobile app?
Is there a minimum age?
Should you pick Emma?
Emma answers a narrow, real question well: how to put a chatbot on top of the data a company already holds without writing code. The integration catalogue is serious for a product of this size, with twenty-four named connectors covering e-commerce, CRM, support, spreadsheets, project management, payments and analytics. Pricing is public and readable, from USD 12 a month to USD 249, with unlimited end users on every tier and free accounts for approved educational organisations. The API and the JavaScript SDK are open, and the SDK is published on npm for anyone to inspect.
The governance side is another matter, and it is the reason to hesitate. The privacy policy page is published, linked from every footer, and empty. The terms of service break off after the Account Activation section, leaving no governing law, no termination clause, no refund policy and nothing on security. The GDPR is never mentioned, no data processing agreement is offered, no subprocessor list exists, and no retention rule is stated. There is no postal address and no email address anywhere on the site; the only channel is a form hosted on Coda. The site also contradicts itself in places, on the model version, on the number of assistants and on whether the product is a free beta or four paid tiers.
The practical reading: Emma is worth a trial for low-sensitivity internal knowledge, on the cheap Starter tier, where the three-day trial costs nothing to run. Anyone planning to feed it customer records, payment data or HR files should first get in writing what the site does not publish.
- Choosing a selection results in a full page refresh.
- Opens in a new window.