Shape
Shape is an AI data analyst that answers plain-English questions about a company's database, writing advanced SQL, generating charts and stating its assumptions. It runs inside Slack or embeds into products, but the service is no longer operating.
What is Shape?
Shape was an AI data analyst built around a single promise: ask a question about your database in plain English and get an answer back. Its engine, DataIQ, was presented as thinking like an analyst rather than like a query box, interpreting the question, writing the SQL, and returning something more useful than a raw number.
Four behaviours defined it. It generated advanced SQL, including complex joins and unions. It returned nuanced answers rather than bare figures. It stated its assumptions and caveats explicitly: that last month meant the last complete calendar month, that biggest cities meant those with 100 or more ticket sales, that a count covered unique events rather than every event. And it claimed to improve with every question asked.
The tool connected to the mainstream analytics stack, namely Snowflake, BigQuery, Amazon Redshift, Databricks, PostgreSQL, MySQL and dbt, and could only be pointed at databases and servers the customer owned or licensed. Answers arrived with charts generated automatically. A Slackbot brought the whole thing into the channel where ad-hoc questions actually get asked, with a Microsoft Teams bot reserved for the Enterprise plan. When an answer mattered enough, a customer could request a review from a human analyst.
Shape also sold itself as a building block. An API and a set of React components let software teams embed AI-powered analytics and dashboards into their own products. Two deployments were offered: an instance hosted by Shape at an unnamed third-party facility, or a Docker image the customer ran on its own infrastructure. Accuracy was credited to GPT-4 custom agents combined with dataset pre-analysis and reviews; security to encrypted credentials and a SOC 2 Type I certification.
The company behind it was Teamspace Technologies Ltd, a London firm built by four people largely out of Meta and backed by Y Combinator. The past tense is deliberate: the company was dissolved on 3 March 2026, the application no longer responds, and the site has published nothing since October 2023.
What it does
- Ask a question about your database in plain English and get an answer back
- Generate advanced SQL, including complex joins and unions
- Read the assumptions and caveats the engine states alongside each answer
- Get charts built automatically from the result
- Query your data from Slack through the Slackbot
- Request a review of any answer by a human analyst
- Embed analytics into your own product with the API and React components
When to use Shape / When not to
A quick filter to help you decide if Shape is the right fit.
When to use Shape
- Data and analytics teams drowning in ad-hoc questions they have no time left to answer
- Business users who need figures out of Snowflake, BigQuery or PostgreSQL without writing SQL themselves
- Product and growth teams tracking metrics that fixed dashboards go stale on too quickly
- Software companies wanting to embed analytics in their own product through an API and React components
- Security-conscious organisations that require a self-hosted Docker deployment rather than a vendor-run instance
When not to use Shape
- Anyone without an existing warehouse or database: Shape queries data, it never collects or ingests it
- Individuals and small budgets, since there is no free plan and entry costs 49 USD per month for two seats
- Buyers who want to sign up and test the same day, as every plan goes through a sales call and a signed Order Form
- European organisations needing documented GDPR compliance, a DPA or a published subprocessor list, none of which exist here
- Anyone shopping for a live product today, the publisher having been dissolved in March 2026 with the application no longer responding
How to use Shape
A typical end-to-end flow, from setup to results.
- Book a call through the site, since there is no self-serve sign-up
- Agree the scope and sign an Order Form, which fixes the services start date and the number of seats
- Choose a deployment: an instance hosted by Shape, or a Docker image you run yourself
- Connect the databases and servers you own or license, among Snowflake, BigQuery, Amazon Redshift, Databricks, PostgreSQL, MySQL and dbt
- Supply the connection credentials, which are stored encrypted
- Ask your first question in plain English from the web application
- Check the assumptions and caveats returned with the answer before acting on it
- Install the Slackbot so the team can ask questions where the questions actually arise
- Request an analyst review whenever an answer needs a second pair of eyes
- Note that this path is no longer practicable: app.shape.xyz resolves but never answers
Pros & Cons
Pros
- States its working assumptions in writing, which is what makes an answer auditable rather than merely convincing
- Generates advanced SQL with complex joins and unions, not just simple aggregates
- A human analyst can be asked to review any answer on demand
- Covers seven mainstream warehouse and database sources out of the box
- Works inside Slack, where ad-hoc questions are actually asked
- Self-hosting through a Docker image keeps data inside the customer's own infrastructure
- Pricing is published in full, with a 30-day money-back guarantee on every plan
Cons
- The product is no longer operating: the publisher was dissolved on 3 March 2026 and app.shape.xyz does not respond
- No free plan and no free trial, with the entry point set at 49 USD per month
- No self-serve sign-up, since every plan requires a sales call and a signed Order Form
- No GDPR commitment of any kind, the privacy policy still resting on the invalidated Privacy Shield
- No data processing agreement, no subprocessor list and no disclosed hosting country
- The API is a selling point on the homepage, yet no public documentation exists for it
- A British company imposing California law and arbitration in Santa Clara County on its customers
Pricing & Plans
There is no free plan and no free trial. The lowest paid entry point is the Starter plan at 49.00 USD per month, which includes two users; each additional user is billed at 24.95 USD per month. Every plan carries a 30-day money-back guarantee.
- AI data analyst
- ad-hoc queries answered
- AI analyses your data
- 2 users included
- additional users at 24.95 USD per month
- Slackbot
- SOC 2 Type I
- dedicated Slack support
- the same scope with 20 users included
- additional users at 24.95 USD per month
- Slackbot
- SOC 2 Type I
- dedicated Slack support
- unlimited users
- Slackbot
- Microsoft Teams bot
- white glove onboarding
- SOC 2 Type I
- dedicated Slack support
Data, GDPR & hosting
A consolidated view of how Shape handles your data.
GDPR overview
There is no GDPR implementation to report, and that deserves to be said plainly: the words GDPR and General Data Protection Regulation appear nowhere on the site. The privacy policy instead commits Shape to the EU-US and Swiss-US Privacy Shield, a framework the Court of Justice of the European Union invalidated in 2020 and which was never replaced here, and declares the company subject to the enforcement powers of the US Federal Trade Commission. EU and Swiss residents are directed to info@shape.xyz or to the London postal address, then to the Privacy Shield website. No Article 27 representative is named, no data protection officer, no data processing agreement, no legal basis, no response deadline and no list of data subject rights. The publisher is established in the United Kingdom, so UK GDPR would apply; it is not mentioned either.
Who owns the data?
The terms are explicit: the customer keeps every intellectual property right in its Customer Data and in any Customer Application built on the service, while Shape keeps its rights in the Services and Software. Customer Data covers both the data held in the customer's own databases and the results returned by queries run against it. The privacy policy adds that personal information is never sold or rented, and is not disclosed outside Shape.xyz, its parent, subsidiaries or related entities. One asymmetry deserves attention: any feedback or suggestion the customer volunteers may be used by Shape without restriction and without obligation.
Reuse rights
Because the customer owns its Customer Data outright, it may reuse that data and the query results freely, for internal business purposes or, where the Order Form authorises it, for external ones, including embedding Shape inside an application served to its own clients. No permission from Shape is required. The constraints run the other way: Shape commits to access Customer Data only to deliver the ordered services and technical support, and never for other products, services or advertising. The customer must nonetheless respect the use restrictions, which forbid reselling or sublicensing the service itself, high-risk activities, ITAR-controlled material, health data covered by HIPAA without a signed business associate agreement, and benchmarking published to third parties. Access to Customer Data through the service ends when the Order Form terminates.
Data retention & training
Hosting summary
Hosting is described only in contractual terms, never in geographic ones. The terms set out two deployments. In a Shape Hosted Deployment, the software is installed by Shape on a web-connected platform run in a third-party hosting facility that Shape designates; that facility is never named and no country or region is given. In a Customer Self-Hosted Deployment, the customer receives a Docker image and runs it on its own premises or in a data centre it selects and manages, which keeps the data inside its own jurisdiction by construction. The privacy policy adds that personal information is stored in secure operating environments not open to the public, that care is taken over transmission, and that no transfer over a public network can be guaranteed fully secure. Connection credentials are encrypted to industry standards and the company claims SOC 2 Type I certification. No hosting country, no region and no subprocessor list is published, so a customer with data residency obligations would have to obtain that information privately.
Things to keep in mind
Risks and trade-offs to weigh before adopting Shape.
- The publisher, Teamspace Technologies Ltd, was dissolved on 3 March 2026; app.shape.xyz resolves but times out, and the status page named in the footer has been deleted, its address now returning a success code while serving Atlassian's marketing site
- Nothing has been published on the site since 24 October 2023, so every claim it still makes describes a product that stopped moving three years ago
- Three different legal names coexist across the site, namely Teamspace Technologies Ltd in the terms, Teamspace Inc in every footer and Shape.xyz, Inc in the privacy policy, alongside two different London addresses
- An AI that writes SQL and states its own assumptions is persuasive precisely because it sounds like an analyst; a wrong assumption, confidently stated, quietly becomes a business decision. Those stated caveats are an invitation to check, not a substitute for checking
- Handing a natural-language tool broad database credentials moves access control away from the people who normally govern it; the terms confirm credentials are encrypted but say nothing about permissions at query level
- Leaning on an AI analyst can erode a team's own SQL literacy without anyone noticing, and the day the service stops, as it has here, that skill has to be rebuilt from scratch
- Billing disputes are directed in the terms to an address on a domain that is not the site's, and a British company imposes California law with arbitration in Santa Clara County
Setup & Integrations
Technical difficulty
Setup difficulty depends entirely on the route chosen. The lightest path, an instance hosted by Shape plus the Slackbot, asks little of the customer beyond connecting a warehouse and supplying credentials. Self-hosting the Docker image demands real infrastructure skills. Embedding the API and React components is developer work, made harder by the absence of any public documentation. In every case there is no self-serve start: a sales call and a signed Order Form come first. Enterprise buyers get white glove onboarding, and dedicated Slack support is included on all plans.
Deployment
Integrations
Supported languages
Behind Shape
Fundraising
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What does Shape actually do?
Which data sources does it connect to?
How much does it cost?
Is there a free plan or a free trial?
Can it be used from Slack?
Can I host it myself?
Is there an API?
Is my data used to train AI models?
What is the minimum age to use it?
Is Shape still available?
Should you pick Shape?
Shape got one thing right that most natural-language query tools get wrong. Asking a database a question in English is easy to demonstrate and hard to trust, because the machine must decide what your words meant before it can count anything. Shape's answer was to show its work: every response carried the assumptions it had made, in plain sentences a business user could challenge. Add advanced SQL generation, automatic charts, a Slackbot and an on-demand human analyst review, and the design reads as genuinely thought through.
The gaps sat on the other side of the product. Governance never caught up: no GDPR commitment, a privacy policy still resting on the Privacy Shield five years after it was struck down, no data processing agreement, no subprocessor list, no disclosed hosting country, and complete silence on whether customer data trained any model. Commercially it was closed too, with no free plan, no trial, 49 USD per month to start and a sales call before anything else. The API it advertised had no public documentation.
None of that is the reason to be careful today. Teamspace Technologies Ltd was dissolved on 3 March 2026. The application at app.shape.xyz resolves in DNS but never answers. The status page the footer points to has been deleted, and that address now quietly serves Atlassian's own marketing site instead, returning a success code that means nothing. Nothing has been published since October 2023.
The marketing site is still standing, and it still reads as a live product. It is not one. This entry is best treated as a record of what Shape was and what it attempted, useful to anyone evaluating the same idea elsewhere, rather than as a shortlist candidate.
- Choosing a selection results in a full page refresh.
- Opens in a new window.