Dot
Dot is an AI data analyst that answers business questions in plain language: it finds the right tables, writes the SQL and returns a chart in seconds, delivered in Slack, Teams, email or the browser.
What is Dot?
Dot is an AI data analyst published by Snowboard Software GmbH, a German company based in Lorch that also operates the Sled brand. It sits on top of an existing cloud warehouse and turns business questions into answers, without asking the person in front of it to know SQL or to hunt through dashboards.
The product is built from four blocks. Chat takes a question in plain English, identifies the right tables, writes the SQL or calls the semantic layer (dbt, Looker) and returns a chart, typically in around thirty seconds. Deep Analysis handles the harder questions: it shows its analysis plan before running anything, investigates several angles in parallel, drills into anomalies, and comes back two to ten minutes later with findings and recommendations exportable to PDF or PowerPoint, covering A/B testing with statistical significance, marketing mix models, forecasting and scenario planning. Automated Reports turns that into a cadence: weekly, monthly or custom business reviews, delivered as a board-ready deck into Slack or by email, with follow-up questions answered in the same thread. Context Agent, presented as an AI data governance manager, pulls context from Looker, Tableau, Metabase, Power BI, dbt, Confluence, Notion, Jira, query history, warehouse schemas and even PDFs or spreadsheets, writes the documentation that is missing, flags inconsistencies, and routes every change through a pull-request workflow with human review and git versioning.
Traceability is the spine of the design: every answer exposes the SQL it ran, the Python code, the raw data and the documents it relied on, with a CSV download. Dot queries the warehouse in place using the database user the customer provides, so it cannot read what that user is not allowed to read; three internal roles, groups, workspaces and row-level security sit on top. Answers are consumed in Slack, Microsoft Teams, email or the browser, and also through an API, a CLI, an MCP server, or embedded in another application. More than 35 first-class connectors are complemented by standard SQL compatibility, and the models behind it come from Anthropic, Google and OpenAI, with Azure OpenAI available where an organization requires it. The vendor publishes its own DABStep benchmark results and customer outcomes ranging from 800 to more than 12,000 hours saved per year.
What it does
- Ask a business question in plain language and get a chart back in around thirty seconds
- Launch a Deep Analysis that investigates several angles at once and returns an executive report in two to ten minutes
- Schedule recurring business reviews delivered as a ready-to-present deck in Slack or by email
- Have the Context Agent build and maintain the documentation of your metrics across every data system
- Audit any answer through the SQL, the Python code, the raw rows and the source documents shown beneath it
- Run statistical work: A/B test significance, marketing mix models, forecasting and scenario planning
- Connect more than 35 data sources without code, in read-only access
When to use Dot / When not to
A quick filter to help you decide if Dot is the right fit.
When to use Dot
- Data teams buried under ad hoc requests, where a simple metric question sits two or three days in the queue
- Business stakeholders with no SQL, who today depend on static dashboards to get an answer
- Companies already running a cloud warehouse such as Snowflake, BigQuery, Redshift or Databricks
- Governance-minded organizations that need SAML single sign-on, role-based access, row-level security and a full audit trail
- Teams that live in Slack or Microsoft Teams and want answers and scheduled reviews delivered in the thread
When not to use Dot
- Solo practitioners and very small businesses, for whom the paid entry ticket of USD 180 per month is hard to justify
- Anyone who simply wants to interrogate a standalone spreadsheet: Dot is built around a connected warehouse, not around files dropped in on the fly
- Organizations with no documented or governed data, since the vendor itself ties accuracy to the governance already in place
- Small budgets that nonetheless require single sign-on, row-level security or workspaces, which only start at the Team tier at USD 720 per month
- Mobile-first users: there is no native iOS or Android application, only the web, Slack, Microsoft Teams and email
How to use Dot
A typical end-to-end flow, from setup to results.
- Create a free account: no credit card is required and 300 credits are granted on sign-up
- Connect a source through the no-code integrations, in read-only access, using a database user whose rights define exactly what Dot may see
- Let the Context Agent ingest schemas, dbt models, Confluence and Notion documentation and query history, then review its proposals through the pull-request workflow
- Ask a first question, in the web chat, in Slack, in Microsoft Teams or by email
- Check the answer against the SQL, the Python code and the raw data displayed underneath, and download the CSV if you need it
- Switch to Deep Analysis for a real investigation, amending the analysis plan before it runs
- Schedule a recurring report: set its structure, its metrics and its cadence, then choose Slack or email delivery — the vendor announces under ten minutes for this step
- Refine over time by adding instructions, examples and business rules through the Context Agent, and by flagging wrong answers so Dot learns
- Set up administration: SAML or OIDC single sign-on (Entra ID, Okta, Google), the Admin, Modeler and User roles, groups and workspaces
- Export the organization audit log through the API to keep a record of who asked what
Pros & Cons
Pros
- Every answer is auditable rather than merely asserted: the SQL, the Python, the raw rows and the referenced documents are all shown
- Documented and audited security posture: SOC 2 Type II attested by MJD Advisors, annual penetration testing and continuous monitoring
- Zero data retention contracted with every model provider, and an explicit commitment never to train on customer data
- Data residency chosen between the EU (Falkenstein, Germany) and the US (Hillsboro, Oregon), with a separate database per organization
- Dot queries the warehouse with the database user you supply, so it cannot exceed the rights already granted on the warehouse side
- European publisher with a complete imprint — commercial register, HRB number, VAT number, named directors — and a public, named subprocessor list
- Broad connectivity (35+ first-class integrations plus any standard SQL database), delivery where the team already works, and unlimited users on every paid tier
Cons
- A high paid entry point: USD 180 per month on annual commitment, USD 200 per month billed monthly
- The free plan is capped at 300 one-time credits, which are not renewed each month
- Credit-based consumption with metered overage (USD 2.00 per credit on Pro, USD 1.60 on Team; USD 1.80 and USD 1.44 on annual billing) makes the real bill hard to forecast
- Single sign-on, row-level security, workspaces and embedding only start at the Team tier at USD 720 per month; self-hosting, audit logs and SLA are Enterprise-only, on quote
- A genuine technical prerequisite: an accessible warehouse or database and up-to-date documentation, the vendor stating plainly that accuracy is tied to your data governance
- No native mobile application, and only five query languages are actually named, with no exhaustive list published
- Paid contracts sit in separate PDFs (MSA, DPA), the SOC 2 report is available only under NDA, and the benchmark and ROI figures are published by the vendor with no independent audit shown
Pricing & Plans
A permanent free entry point exists: the Free plan costs nothing, requires no credit card and grants 300 one-time credits with full access to Pro features. The lowest paid tier is Pro at USD 180 per month on annual billing, or USD 200 per month when billed monthly, including 150 credits per month and unlimited users. Team follows at USD 720 per month annually (USD 800 monthly) with 800 credits, and Enterprise is quoted on request with unlimited credits and volume discounts. Consumption is measured in credits — one credit per chat answer, approximately one credit per minute of analysis task — and additional credits may be purchased on demand, overage being billed at USD 2.00 per credit on Pro and USD 1.60 on Team (USD 1.80 and USD 1.44 respectively under annual billing). No euro price is displayed, despite the publisher being German.
- 300 one-time credits
- full access to Pro features
- any data source
- team invitations
- 150 credits per month
- unlimited users
- 35+ connectors
- email and Slack reports
- organization notes and business rules
- Context Agent
- charts and visualizations
- priority email support
- 800 credits per month
- unlimited users
- everything in Pro plus workspaces
- single sign-on (Okta
- Azure
- Google)
- row-level security
- custom branding
- unlimited credits and volume discounts
- everything in Team plus self-hosted deployment
- audit logs
- SLA guarantees
- a dedicated account manager and tailored training and onboarding
Data, GDPR & hosting
A consolidated view of how Dot handles your data.
GDPR overview
GDPR compliance is claimed explicitly and backed by verifiable artefacts. Snowboard Software GmbH, established in Lorch, Germany, processes customer data as a processor under the GDPR and as a service provider under the CCPA, and publishes a DPA as a freely downloadable PDF covering both. The privacy policy, effective 1 June 2026, sets out the legal bases (consent, performance of contract, legitimate interest, legal obligation) and the full set of data subject rights: access, rectification, erasure, objection, restriction, portability, withdrawal of consent and complaint to a supervisory authority, exercised by email at hi@getdot.ai. International transfers rely on the European Commission's standard contractual clauses, provided on request, and subprocessors are named publicly. No Article 27 representative is designated and none is required, the controller being established in the EU; no data protection officer is named either. SOC 2 Type II is attested by MJD Advisors.
Who owns the data?
Ownership of Customer Content stays with the customer: the terms of service state that you retain all rights to what you upload or have Dot ingest, and Snowboard Software GmbH receives only a limited, non-exclusive licence to host, process, transmit and display it, solely as needed to run the service. The privacy policy, effective 1 June 2026, frames Dot as a processor acting on the customer's instructions, the customer remaining the controller. Little is actually held: Dot queries the warehouse in place and stores only the schema, the context documentation it needs, and the results returned into a chat. The platform itself remains Dot IP, owned by the publisher.
Reuse rights
Because the customer keeps all rights to Customer Content and the licence granted to the publisher is confined to operating the service, the outputs — charts, generated SQL, CSV exports — can be reused inside the organization without asking permission. On the vendor's side the declared uses are narrow: providing, supporting and operating the service, network and information security, and analysing usage data to develop and improve the product. Customer data is never used to train models, and every model provider (Anthropic, Google, OpenAI, with Azure OpenAI as an option) is engaged under zero data retention terms, so prompts and results are not stored on their side either. Disclosure is limited to the cases the privacy policy enumerates: vendors, legal obligations, a business transfer. Under the CCPA the publisher states that it does not sell personal information, but that it does share identifiers, contact details and network activity for cross-context behavioural advertising — a statement about website visitors, not about Customer Content. Data read through Google APIs (Google Ads) is read-only, never used for training and never sold, in line with Google's Limited Use requirements; the OAuth token is encrypted at rest and can be revoked from the Google account permissions page.
Data retention & training
Hosting summary
Dot runs on Hetzner Cloud, in the Falkenstein data centre in Germany for the EU region and in Hillsboro, Oregon for the US region. Each customer is served from a single region and their data stays there, with a separate database per organization; the website's own servers are described as primarily located in Germany, while the privacy policy reserves the possibility of storing and processing personal information in the United States and other countries. The warehouse itself is never copied: Dot queries it in place. Data is encrypted at rest, on volumes and backups alike, and in transit over TLS only. Backups run hourly with restic, encrypted and deduplicated before leaving the server, then shipped to off-site object storage in a different cloud (Cloudflare). No SSH or administration interface is publicly reachable, admin access going through a private mesh VPN. Named subprocessors are Hetzner Online GmbH (Germany, United States), Cloudflare Inc. (global), Anthropic, Google and OpenAI (United States, under zero data retention), Microsoft (EU, notification email) and PostHog Inc. (EU, product analytics). Observability is self-hosted, so prompts and query results are not sent to a third-party monitoring provider.
Things to keep in mind
Risks and trade-offs to weigh before adopting Dot.
- The credit model drives the real bill: overage is charged per unit and depends on how many chats and analysis minutes the team consumes, so a quiet month and a busy month cost very differently
- Prices are shown on annual billing by default; paying monthly is about 11% more expensive (USD 200 instead of 180, USD 800 instead of 720)
- The 300 free credits are a one-time allowance rather than a monthly refill, and single sign-on and row-level security are absent from Free and Pro, so any real governance requirement forces the Team tier
- Dot queries the warehouse with the database user you provide: the scope of rights granted to that connection determines everything the tool can read, and that is your security decision, not the vendor's
- The privacy policy states that identifiers and network activity are shared with advertising platforms for cross-context targeting — this concerns visitors to the website rather than Customer Content, but it deserves to be known
- Some claims cannot be verified from outside: the SOC 2 Type II report and the penetration test summary are available only under NDA, the public terms cover the website and the free tier while paid use falls under a separate MSA, the benchmark and ROI figures are the vendor's own, and any dispute is governed by German law before the exclusive jurisdiction of the Stuttgart courts
- Convenience has a cognitive cost: when a chart appears in thirty seconds, the reflex to challenge a number quietly fades. The traceability panel only protects a team that actually opens it, and delegating every question to an agent can erode the analytical judgement that made the question worth asking
Setup & Integrations
Technical difficulty
Low to moderate, with one real prerequisite. Sign-up is immediate and needs no card, integrations are no-code and read-only, and the vendor announces under ten minutes for a first automated report and days rather than weeks for the Context Agent. The effort sits upstream: you need an accessible warehouse or database and documentation or dbt models worth ingesting, the vendor conceding that accuracy is tied to your data governance. Someone data-literate should review the Context Agent's proposals through its pull-request workflow, and an identity administrator is needed to wire SAML or OIDC single sign-on on the Team tier.
Deployment
Integrations
Supported languages
Behind Dot
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
What does Dot actually do?
What is a credit, and how quickly do credits burn?
Is there a free plan, and what does it include?
Where is my data hosted?
Is my data used to train AI models?
Can I trust the answers?
How is this different from using ChatGPT on my data?
How long does setup take?
Is a DPA available, and who publishes Dot?
Is there a minimum age?
Should you pick Dot?
Dot belongs to the small group of AI analytics products that can be inspected rather than simply believed. The publisher is identified down to its register entry, the DPA is a public PDF, the subprocessor list is named with locations, and a SOC 2 Type II attestation backs the security claims. That matters in a category where a confidently wrong answer costs more than no answer — and the product agrees, since every response ships with the SQL, the Python, the raw rows and the documents behind it.
Its most distinctive choice is the Context Agent. Rather than papering over unreliable answers with better prompting, it goes after the root cause: missing documentation, contradictory metric definitions, undocumented entity relationships. The vendor is candid that accuracy is tied to your data governance, which is also the honest precondition for buying: you need a reachable warehouse and documentation worth ingesting. Without that, the tool has little to work from.
The commercial terms are transparent but not cheap. The free tier's 300 credits are a one-time allowance, and sustained use starts at USD 180 per month on annual commitment, USD 200 billed monthly. Governance features many buyers consider non-negotiable — single sign-on, row-level security, workspaces — only begin at USD 720 per month, and the credit model ties the bill to how much the team actually asks: fair, but harder to forecast.
For a European organization already running a cloud warehouse, an EU-established publisher, EU data residency and a documented no-training stance form a strong argument. For a solo user or a very small team, the entry ticket will be hard to justify. The performance and ROI figures on display are the vendor's own, published with no independent audit shown: read them as claims, not measurements.
- Choosing a selection results in a full page refresh.
- Opens in a new window.