The AI runtime for business
Sigma is a business intelligence and AI application platform for building analytics, apps, and agents on cloud data warehouses.
AI Panel Score
6 AI reviews
Reviewed
AI Editor ApprovedApproved and published by our AI Editor-in-Chief after full panel analysis.Sigma works as a spreadsheet-style interface layered on top of a cloud data warehouse. Filters, pivots, group-bys, and formulas that users build in the UI are compiled into SQL in the warehouse's native dialect and executed where the data lives, with results returned to the browser. This means users can analyze large datasets using familiar spreadsheet mechanics without exporting files or waiting on pre-built extracts, and can drill into individual records from dashboards or reports.
The platform includes an AI Toolkit for natural language queries and LLM functions embedded in spreadsheet cells, along with Sigma Agents for governed, scaled interactions with data. Sigma Reveal provides ad hoc data discovery through pivoting and grouping without requiring pre-built dashboards or SQL knowledge. For technical users, Sigma supports writing Python and SQL directly inside workbooks alongside the spreadsheet editor. Additional capabilities include pixel-perfect paginated reporting with scheduled exports and bursting, row-level security, embedded analytics with white-label theming and multitenant controls, and a set of prebuilt AI Applications for use cases like budget variance analysis, demand planning, pipeline forecasting, and territory management.
Lets teams build AI-driven apps on cloud data for use cases like budget variance, demand planning, and territory management, without writing code.
Enables users to ask questions in natural language and use LLM functions inside spreadsheet cells to analyze warehouse data.
Provides governed agents that can act on data at scale to automate analysis and workflows.
Provides governed metrics and reusable business definitions so teams get consistent, trusted data across analytics and AI.
Builds board-ready, paginated reports with layout control, scheduled exports, and bursting for distribution.
An interactive data discovery tool that lets users pivot and group data to uncover insights instantly without pre-built dashboards or SQL.
Allows users to combine live warehouse data with manual entries for planning, forecasting, and closing the loop on analysis.
Lets users analyze massive datasets using familiar columns, rows, formulas, and pivots without writing code.
Allows users to write Python and SQL code directly inside Sigma Workbooks alongside the spreadsheet-style editor for deeper analysis.
Embeds white-label analytics and AI-powered apps into other products using secure iFrames, themes, and multitenant controls.
Connects directly to cloud data warehouses like Snowflake, Databricks, BigQuery, and Redshift so queries run on live data without extraction.
Governs data distribution and access with row-level security controls across reports and applications.
For small teams getting started with self-service analytics; viewer/consumer seats are free while creators pay per user.
For growing teams that need expanded collaboration and analytics capabilities beyond the Essential tier.
For mid-size to larger teams needing broader governance, embedding, and advanced BI functionality.
For large organizations with complex data needs, requiring custom contracts, premium support, and advanced security/compliance features; pricing requires contacting Sigma's sales team.
Warehouse-native BI with real AI muscle, priced like a serious enterprise buy.
“Sigma competes with Looker and Tableau by skipping data extraction entirely. Strong feature set, but $75/seat Business tier and contact-sales Enterprise means this is a real budget line, not a quick trial.”
No data movement. Queries run in Snowflake, Databricks, BigQuery, or Redshift directly, and results land in a spreadsheet interface. That's the pitch, and it's a real architectural advantage over tools that extract and cache.
Input Tables for write-back is the differentiator worth watching. Most BI tools are read-only; combining live warehouse data with manual entry for planning work closes a loop that usually needs a second tool like Anaplan.
Three tiers before you hit $75/month Business, then Enterprise disappears behind a sales call. SOC 2, HIPAA, GDPR, SSO/SCIM are all present, which derisks procurement. Competes directly with Looker and Tableau — the question is whether your analysts want spreadsheet mechanics or dashboard mechanics first.
Direct warehouse connectivity to Snowflake and Databricks positions it ahead of extract-based legacy BI tools.
Named alongside Looker and Tableau as a credible BI choice, low risk of looking like an odd pick.
Spreadsheet interface lowers ramp time but Enterprise tier requiring sales contact slows procurement.
Write-back via Input Tables and warehouse-native execution advance planning workflows, not just cheaper dashboards.
Well-known Series D-stage company in market for years, but no public funding data in this evidence.
Teams already on Snowflake or Databricks who want planning write-back without a second tool.
Skip if your team just needs simple dashboards and doesn't need spreadsheet-native write-back.
Warehouse-native architecture means Sigma extends my governance model instead of fighting it.
“Sigma pushes compute into Snowflake, Databricks, BigQuery, and Redshift rather than pulling data into a proprietary layer. That single decision drives almost everything I like about it.”
No extraction, no shadow copy of the warehouse living in some vendor's cloud I have to audit separately. Everything compiles to native SQL and runs where the data already sits, which means my row-level security and access controls stay authoritative in one place. Compare that to tools like Looker or Tableau's extract-heavy modes, where you're reconciling two sources of truth.
Input Tables worried me at first pass. Write-back into planning workflows is exactly the kind of feature that quietly seeds unreconciled data drift. It's manageable if governed tightly, but it's a real operational tax on my team, not a free lunch.
The $25 to $75 per-seat ladder is reasonable for BI, but AI Applications and Agents will need enterprise contracting to govern at scale. Three years in, this looks like a durable warehouse-native bet, not a lock-in trap. That's the right shape for how data teams actually operate today.
Sits between traditional BI (Looker, Tableau) and AI-native agent platforms, competing on both fronts at once.
Governed metrics layer plus warehouse-native execution matches how data teams actually want BI to behave in 2024.
Direct connectors to Snowflake, Databricks, BigQuery, and Redshift mean it slots into an existing modern data stack, not around it.
No data extraction reduces lock-in risk, though Input Tables write-back needs governance discipline to avoid drift over time.
Spreadsheet UI, SQL/Python-in-workbook, and native compilation show genuine architectural investment beyond a chat wrapper.
Data teams standardized on Snowflake, Databricks, BigQuery, or Redshift who want governed BI without an extraction layer.
Avoid if your team needs transparent, self-serve pricing for AI agent workloads without a sales cycle.
Three tiers visible. Fourth tier hidden. Warehouse compute cost isn't in the sticker at all.
“$25 to $75/seat, three tiers public. Enterprise is a black box, and warehouse query costs sit outside the invoice entirely.”
$25/month Essential, $50 Pro, $75 Business. Viewer seats free — real savings if your ratio is 10 viewers to 1 creator. 50 creators × $50 × 12 = $30K/year at Pro. Add Snowflake or BigQuery compute, since Sigma queries live, it doesn't store. That's a second invoice finance teams forget to model.
Enterprise is contact-sales, custom term length, custom everything. No published overage rate for query volume — that's the real risk, not the seat price.
Compare to Looker, which bundles similar warehouse-native querying but with its own opaque enterprise pricing. Sigma's three public tiers beat that on transparency. Procurement can price two-thirds of this without a call. Compliance stack (SOC 2, HIPAA, GDPR, SSO/SCIM) reduces security review friction, which shortens onboarding cycles.
SOC 2, HIPAA, GDPR, SSO/SCIM listed upfront, which shortens security review for procurement.
No published term length or auto-renewal terms for Enterprise tier.
Three tiers and prices public; Enterprise still requires a sales call.
Input Tables and prebuilt AI Applications give measurable workflow use cases, not just dashboards.
Seat cost is clear; warehouse compute cost from live queries sits outside the invoice.
Teams already on Snowflake or BigQuery who want to avoid extract-based BI tools.
Avoid if your warehouse compute budget is fixed and can't absorb variable query costs.
Warehouse-native BI that lets you live in a spreadsheet without extracting data
“Sigma pushes computation into Snowflake/BigQuery instead of pulling data out, which is the right architecture for governed analytics. The daily workflow question is whether spreadsheet-as-interface holds up once workbooks get complex.”
Querying live off the warehouse instead of maintaining extracts is the correct call — no stale cache debugging, no refresh-schedule babysitting like you'd get in legacy Tableau deployments. Input Tables for write-back is the feature I'd actually use weekly: planning and forecasting loops usually require exporting to a spreadsheet and re-importing, and Sigma closes that loop natively.
Where I'd want day-3 evidence: spreadsheet formulas compiling to SQL under the hood is elegant until you're debugging a slow pivot and need to see the generated query. Docs cover SQL/Python-in-workbook mechanics, but nothing here shows query plan visibility or how governance (row-level security, metrics layer) holds up once fifty analysts are building simultaneously.
Pricing starts at $25/month (Essential) scaling to $75 (Business), with free viewer seats — reasonable against Looker's enterprise-only posture. Tradeoff: power depends on warehouse performance, so a slow Redshift cluster becomes Sigma's problem too.
Spreadsheet-to-SQL compilation is a strong pitch but daily debugging visibility isn't evidenced.
Docs exist and cover SQL/Python workbook use, but no changelog detail on query optimization or debugging shown in evidence.
Write-back via Input Tables removes a classic export/re-import fight, though multi-tool AI Toolkit adds surface area to learn.
SQL and Python inside workbooks alongside spreadsheet UI covers beginner-to-advanced without forcing a tool switch.
Live warehouse connectivity (Snowflake, Databricks, BigQuery, Redshift) avoids extract/reload habits analysts already hate.
Analytics teams already on Snowflake, Databricks, BigQuery, or Redshift who want spreadsheet-style exploration without building extracts.
You need a lightweight, cheap BI tool and don't already have a governed cloud warehouse in place.
A spreadsheet that actually talks to your warehouse, but you'll need a warehouse first
“Sigma's pitch is live queries, not extracts, and that's a real difference if you've been burned by stale dashboards. But this is a tool for teams already deep in Snowflake or BigQuery, not a casual Sunday project.”
Sigma's whole trick is compiling your pivots and formulas into SQL that runs where the data actually lives, no extracts, no waiting on someone's overnight job. That's a genuinely different feel from tools like Tableau that got built in the extract era. Input Tables for write-back is the feature I'd actually use daily, planning and forecasting without exporting to Excel and losing the thread.
But day one is homework, not welcome. You need Snowflake, Databricks, BigQuery or Redshift already wired up before Sigma does anything useful. That's not a knock exactly, it's warehouse-native by design, but it means no scrappy solo-user path. Pricing starts at $25/month per creator with free viewer seats, climbing to $75 for Business tier governance and embedding.
The AI Toolkit and Sigma Agents sound promising on paper. Whether Ask Sigma feels trustworthy after month three, with no public benchmarks shown, is the open part. Mobile isn't mentioned anywhere, and for a warehouse-tethered spreadsheet tool, that's probably fine.
Pixel-perfect reporting and row-level security suggest real investment in details that matter to analysts running this daily.
Spreadsheet interface lowers the floor, but SQL/Python and Sigma Agents raise the ceiling considerably by month three.
Platforms listed as web only, no mobile story mentioned anywhere in the evidence.
Requires an existing warehouse connection (Snowflake, BigQuery, etc.) before value shows up, that's a real barrier for first ten minutes.
SOC 2 Type II and HIPAA compliance plus live-query architecture signal a mature, governed backend rather than a bolted-together demo.
Teams already running a cloud warehouse who want live, spreadsheet-style analysis without extracts.
Avoid if you don't have a cloud data warehouse in place or need a real mobile app.
"AI runtime for business" is a big claim for a spreadsheet on a warehouse.
“Sigma's warehouse-native architecture is real and the pricing page is unusually transparent for BI. The AI framing is where I'd slow down.”
$25 a month, viewer seats free, live query on Snowflake or BigQuery instead of extracts. That's a legible product. Looker did warehouse-native years ago and got acquired by Google; Looker Studio still exists. Sigma's spin is the spreadsheet-first UI plus write-back via Input Tables, which is a real differentiator against Tableau and Looker both.
The "AI runtime" headline is the tell. Ask Sigma and Sigma Agents sound like every BI vendor's 2024 chat-layer bolt-on. No API listed in the capabilities scan, no public pricing page despite four tiers quoted elsewhere — that's a gap between marketing polish and what's actually documented.
Exit story: it's SQL compiled against your warehouse, so your data isn't trapped. But dashboards, Input Tables, and embedded apps don't travel. SOC 2 and HIPAA are checked boxes, not differentiators anymore.
Spreadsheet-plus-write-back combo is a real gap versus Tableau and Looker, not just a copycat.
No data extraction and native SQL compilation means your warehouse stays clean if you leave.
No public funding data or API in the evidence scan, four pricing tiers but no pricing page found.
"AI runtime for business" oversells what's described as chat-in-cells plus governed agents.
Warehouse-native BI is a proven pattern (Looker) though the AI-agent layer is unproven at scale.
Teams already on Snowflake or BigQuery who want spreadsheet-style self-service without extracts.
You need a documented API or fixed enterprise pricing before you'll sign.
Common questions answered by our AI research team
Sigma connects directly to cloud data warehouses including Snowflake, Databricks, BigQuery, and Redshift.
No. Sigma queries your data warehouse directly and is warehouse-native by design, so data stays in place without extraction or copying.
Yes. Input Tables support write-back, letting you combine live warehouse data with manual entries for planning and forecasting workflows.
Sigma's foundation includes SOC 2 Type II security, HIPAA compliance, GDPR privacy, and SSO & SCIM for identity management.
Yes. Users can interact with data through a spreadsheet interface, natural language chat, SQL, or Python, all compiling into queries that run in the warehouse.
Sigma Computing is a San Francisco-based analytics platform that lets users explore and build on cloud data warehouses using a spreadsheet-like interface, SQL, or Python.
