Obviously AI
Obviously AI was a no-code AutoML platform turning tabular business data into classification, regression and time-series predictions in minutes. The service is now archived: the brand became Zams and the application no longer responds.
What is Obviously AI?
Obviously AI, published by Obviously AI, Inc. of San Francisco, was a no-code AutoML platform built around a single promise: take a spreadsheet of historical business data, point at the column you want predicted, and get a production-ready model in minutes rather than months, without writing a line of code.
The platform covered the classical tabular machine-learning families. Classification handled churn, lead conversion, loan repayment and fraud; regression handled sales, dynamic pricing, yield, deal size and costs; time series predicted a value at a future date, from sales in six months to tomorrow's stock price. The pricing table added clustering, custom LLMs and AI agents.
Five product blocks structured the offer: model building, one-click deployment, automatic monitoring, integration and sharing, and expert support. A trained model could be published as an instant web app shared by link, or called through a real-time REST API from another application. Prediction data could be pushed back into existing BI tooling such as PowerBI and Looker, and monitoring could be wired to conditional Zapier automations — the documented example fires an action once prediction probability passes 80%.
Twenty-three named connectors fed the platform: warehouses and databases (Snowflake, Google BigQuery, Amazon Redshift, PostgreSQL, MySQL, SQL Server, MongoDB), cloud storage (Google Drive, Dropbox, Box), CRM (Salesforce, HubSpot), BI (Tableau, Mode, Chartio, Google Analytics), plus Airtable, Zapier, Shopify, Typeform, Kaggle, CSV and JSON.
A second and distinct line sold human expertise. Data Scientist as a Service placed a master's- or PhD-level data scientist on the customer's Slack or Teams for data preparation, merging, feature engineering, visualisation, modelling, deployment and advice.
Compliance was unusually well documented for a company of this size: SOC 2 Type II audited annually, HIPAA with a business associate agreement available on qualification, CCPA, and a GDPR data processing agreement for EU customers, all on Google Cloud with encryption in transit and at rest.
The tool is no longer operating. Every page carries an Archived banner announcing that Obviously AI is now Zams, and as of 12 August 2026 the application, the API documentation, the help centre and the status page all fail to respond.
What it does
- Predict a business outcome from a historical spreadsheet without writing code
- Build classification, regression, time-series and clustering models automatically
- Publish a trained model as a shareable web app in one click
- Serve live predictions to your own product through a REST API
- Trigger conditional automations from prediction scores via Zapier
- Monitor deployed models for drift and performance decay
- Pull data straight from Snowflake, BigQuery, Salesforce, PostgreSQL and seventeen more sources
When to use Obviously AI / When not to
A quick filter to help you decide if Obviously AI is the right fit.
When to use Obviously AI
- Business analysts sitting on historical spreadsheet data with no coding skills
- Small and mid-sized teams that never hired an in-house data scientist
- Revenue, pricing and demand planners forecasting sales, inventory or dynamic prices
- Credit, lending and fraud teams scoring repayment risk or suspicious activity
- Readers researching the no-code AutoML market, since the product itself is retired
When not to use Obviously AI
- Anyone needing a working tool today: sign-up, API documentation and support are all offline
- Teams working with text, images or audio rather than rows and columns
- Data scientists who want control over model selection, tuning and code
- Buyers who require a published software price before they will evaluate a product
- Organisations that need a vendor with a live support channel and a status page
How to use Obviously AI
A typical end-to-end flow, from setup to results.
- Create an account on the free plan, or start the 14-day trial described in the terms of use
- Upload a CSV, or connect a source such as Snowflake, Salesforce, PostgreSQL or Google Drive
- Point the platform at the column whose value you want predicted
- Let the platform select, train and test an algorithm automatically
- Read the prediction and the factors reported as driving it
- Publish the model as an instant web app and share the link with your team
- Or call the model in real time from your own application through the REST API
- Wire prediction thresholds to Zapier so that a score triggers an action downstream
- Watch automatic model monitoring for drift once the model is live
- Note that none of these steps can be completed today: sign-up is a dead link and the application is offline
Pros & Cons
Pros
- Genuinely no-code: no data science team, AI background or programming knowledge required
- Deployment was included rather than left as an exercise — instant web app and REST API, not just a trained model
- Broad, explicitly named connector list spanning databases, warehouses, storage, CRM and BI
- Serious documented compliance: SOC 2 Type II audited annually, HIPAA with BAA, CCPA, and an EU data processing agreement
- European data residency addressed head-on, with servers in the Netherlands and the United Kingdom
- Generous permanent free plan: 1,200 predictions, unlimited models and one million training rows
- Named and quantified customer case studies rather than an anonymous logo wall
Cons
- The service is shut down: the application, API documentation, help centre and status page are all unreachable
- No software pricing was ever public — the pricing page carries no figure at all, in text or in markup
- Sign-up and sign-in are dead anchors, so the product cannot be tried or evaluated
- No legal page is linked from the site; the privacy policy and terms exist only at URLs a visitor must guess
- No subprocessor list, no Article 27 EU representative and no data protection officer are named
- Nothing is stated, in either direction, about training models on customer data
- Headline performance figures come with no published methodology behind them
Pricing & Plans
A permanent free plan was offered to individuals, hobbyists and non-profit organisations, and the terms of use mention a 14-day free trial for new registrants, billed automatically at its end. Paid software pricing was never published: the Startup, SMB and Enterprise plans are described in detail but without a single figure. The only public price anywhere on the site is USD 1,000 per month for the Data Scientist as a Service offering. All payments were taken in US dollars.
- Free — 1
- 200 predictions
- 1 user seat
- unlimited models
- 1 million rows of training data
- CSV only
- classification
- regression
- Startup Plan — full software access
- agentic and predictive AI
- AI agents
- RPA automations
- recurring reports
- model dashboard
- REST APIs
- model finetuning and monitoring
- SMB Plan (marked Popular) — same capabilities as Startup
- up to 3 use cases
- up to 1 GB of files and 100 million rows
- price not published
- Enterprise Plan — up to 5 use cases
- up to 10 GB of files and 250 million rows
- plus SAML/SSO
- logging and audit trails
- data residency
- custom data sources and private cloud setup
- entry through Contact Us
- Data Scientist as a Service — USD 1
- 000 per month for a dedicated master's- or PhD-level data scientist
Data, GDPR & hosting
A consolidated view of how Obviously AI handles your data.
GDPR overview
GDPR compliance is claimed in so many words, not merely implied. The security page states that Obviously AI complies with GDPR data retention requirements and offers a data processing agreement to customers in the European Union. European data residency backs the claim: a Cloudflare geographic load balancer and servers in the Netherlands and the United Kingdom, so that intra-European traffic never leaves Europe. The privacy policy, effective 4 February 2025, reminds EEA residents of their right to complain to a local supervisory authority. Two gaps remain. No Article 27 EU representative is named anywhere, and no data protection officer is designated; the only route is a general contact address. Rights of access, rectification, portability and erasure are not set out article by article, the policy pointing to account settings instead. All of this describes a service that has since stopped operating.
Who owns the data?
The terms of use claim ownership of the site itself, its source code, databases, designs and trademarks, but say nothing about the datasets customers upload. Neither the terms nor the privacy policy assign ownership of customer data, grant Obviously AI a licence over it, or promise its return on exit. That silence is the notable point. The privacy policy does reserve the right to share or transfer personal information during a merger, sale of company assets, financing or acquisition of all or part of the business — a clause with real weight, given that the brand has since been replaced by Zams.
Reuse rights
Nothing in the terms restricts what customers do with their own datasets or with the predictions their models produce: no licence is granted to the vendor and no permission is needed to reuse the outputs. The restrictions run the other way, over the vendor's own material. Content, marks and source code may not be copied, republished, resold or systematically retrieved to compile a competing collection or directory, and automated access to the site is prohibited. On the personal-data side the privacy policy is broad. Names, contact details, credentials and payment data are collected, enriched from public databases, joint marketing partners and social media profiles, then used for account creation, marketing, administrative messages, order handling, testimonials, security and analytics. Online data partners may associate browsing activity with an email address; advertising opt-out runs through a third-party page, and Do-Not-Track signals are explicitly not honoured.
Data retention & training
Hosting summary
All data was hosted in Google Cloud facilities, described as physically secure with 24/7 on-site security and camera surveillance, in data centres certified SOC 2, ISO 27001 and HITRUST. Traffic to and from the platform was encrypted with TLS, and customer data at rest with AES-256. For European customers the vendor operated a Cloudflare geographic load balancer with regionally located servers in the Netherlands and the United Kingdom, stating that intra-European traffic never leaves Europe. The infrastructure was described as fault tolerant, with databases running in cluster configuration and an application tier scaled by load balancing. Data residency was offered as a governance option on the Enterprise plan. Two limits are worth noting: no United States hosting location is named explicitly, even though the publisher is a San Francisco company, and no subprocessor list is published anywhere — Google Cloud and Cloudflare are named as infrastructure providers, which is not the same thing.
Things to keep in mind
Risks and trade-offs to weigh before adopting Obviously AI.
- The tool is shut down: the application returns an error and the API documentation, help centre and status page do not respond at all
- Sign-up and sign-in are dead anchors, so nothing presented on the site can actually be tried
- No legal page is linked anywhere; the privacy policy and terms exist only at URLs a visitor has to guess
- The privacy policy permits transferring personal data during a merger or acquisition — exactly the kind of event that has since occurred
- Nothing is stated, in either direction, about whether customer data was used to train models
- No Article 27 EU representative and no data protection officer are named, despite claimed European hosting and an EU data processing agreement
- Performance claims of 9x faster model building, 0.08 seconds per model and 200,000 hours saved monthly carry no published methodology
Setup & Integrations
Technical difficulty
On paper, minimal. The site stated repeatedly that no data science team, no AI background and no programming knowledge were required, and described the path from upload to trained model as a matter of clicks. Developers had an optional REST API, and paid plans came with a dedicated AI strategist and data scientist reachable on Slack or Teams. In practice the difficulty is now absolute rather than low: no account can be created, because the sign-up links are dead and the application does not respond.
Deployment
Integrations
Behind Obviously AI
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Is Obviously AI still available?
What happened to the product?
What could Obviously AI predict?
What did it cost?
Did you need to know how to code?
Which tools did it connect to?
What compliance did it claim?
How long was customer data kept?
Was there a minimum age?
Should you pick Obviously AI?
Obviously AI was a credible product, and the record deserves to say so plainly. It addressed a real problem — putting predictive modelling in the hands of analysts who would never write Python — and it went further down the chain than most of its no-code peers, shipping deployment, a REST API, monitoring and twenty-three named connectors rather than stopping at a trained model. Its compliance posture was unusually thorough for a company of its size: SOC 2 Type II audited annually, HIPAA with a business associate agreement, a GDPR data processing agreement, and European servers so that intra-European traffic stayed in Europe. The customer stories were named and quantified rather than anonymous.
None of that is available any more. The publisher itself declares the site an archive of how the product operated through 2025, and the brand has been replaced by Zams. Checked on 12 August 2026, the application, the API documentation, the help centre and the status page all fail to answer, and the sign-up buttons are dead anchors. There is no path from this page to a working account.
Two cautions outlive the shutdown and matter to anyone who was a customer. The privacy policy permits personal information to be transferred during a merger, sale of assets or acquisition, which is precisely the kind of event that has occurred; and neither the policy nor the terms ever addressed whether customer data was used to train models. Anyone whose data may still sit inside the 90-day retention ceiling should raise it with the published contact address.
Read this record as documentation of what the tool was, not as a recommendation to adopt it. Obviously AI is no longer an option for a buyer today.
- Choosing a selection results in a full page refresh.
- Opens in a new window.