- BFSIBanking, Financial Services & Insurance
- HealthcareComing soon
- ManufacturingComing soon
- Retail & E-CommerceComing soon
Snowflake CoWork
Your data already sits in Snowflake. Your business teams still wait days for an answer — Snowflake CoWork, formerly Snowflake Intelligence, lets business users ask questions of governed data in plain language, inside a Snowflake account stored in Mumbai or Pune, billed by the token in AI credits.
Buy through TechBag
Same software. Better outcome — at a lower cost.
How it’s rated
Full scoreboard ↓Quick answer
The Snowflake platform family
This page covers Snowflake CoWork — the business-user work agent, formerly Snowflake Intelligence. The rest:
Quick facts
30-second orientation- Product
- A work agent that answers business users’ plain-language questions over governed Snowflake data
- Maker
- Snowflake Inc. (NYSE: SNOW), whose chief executive is Sridhar Ramaswamy
- Name
- Launched as Snowflake Intelligence; renamed Snowflake CoWork at Summit on 2 June 2026
- Status
- Ramaswamy announced general availability in November 2025; 5,800 accounts by July 2026
- Price
- Tokens billed in AI credits: $2.00 per Global credit or $2.20 per Regional credit
- Needs
- An existing Snowflake account; CoWork is not sold as a standalone subscription
- Features
- Artifacts, Cortex Sense, User Skills, User Memory and a Skill Catalog, as press reports list them
- India
- Its host account’s data can sit in AWS Mumbai or in Pune on Azure Central India
- Analysts
- No analyst placement covers CoWork as a product; TechBag found none
- In India via
- TechBag — use-case scoping, an AI-credit budget, the quote in INR with GST
Understand AI analytics agents before you buy one
Most product pages skip this. We start here — so you buy a capability, not a buzzword.
What is a data work agent?
An agent that takes a plain-language question and answers it from governed company data, without a report request.
A ticket to the analyst team vs a work agent in the warehouse — the honest table
What consolidation actually replaces, dimension by dimension.
| Dimension | Requests queued for the analyst team | Snowflake CoWork |
|---|---|---|
| Who answers a data question | An analyst, from a ticket queue | The person asking, in plain language |
| Where the data goes | Extracts copied into a BI tool | Stays in the Snowflake account |
| Whose permissions apply | A second set kept in the BI layer | The account’s own roles and policies |
| How it is paid for | Named-user licences, used or not | AI credits by the token, as used |
| Time to a first answer | Days, once the request is triaged | One conversation with the agent |
| What it is NOT | — | Standalone, seat-priced, or analyst-ranked |
The cheapest test is a three-question pilot: one team, tables already described, and an AI-credit budget alert set before day one.
The five pieces of the platform
Vendors love diagrams; buyers need to know what they’re actually operating. Here’s the whole platform, demystified.
Account
CoWork is not a separate SaaS tenant: it runs inside a Snowflake account and reads tables in whichever edition and cloud region that account uses, Mumbai and Pune included.
Policies
Snowflake pitches CoWork at governed data. The access roles and Horizon Catalog policies live in the same account, and Horizon ships with the platform rather than as a priced extra.
Cortex
Behind the chat window, Cortex Analyst and Cortex Agents do the work; their calls are metered in AI credits per million tokens, at a rate that depends on the model chosen.
Reach
Press coverage of the June 2026 relaunch lists user skills, memory, a skill catalog and MCP connectors to Google Drive, Salesforce and Slack; confirm each in your region.
An agent inside your Snowflake account — Cortex does the reasoning, your roles decide what each user sees.
Six capabilities. Ask, act, govern.
Snowflake CoWork answers business questions from governed data without leaving the Snowflake account.
Ask without writing SQL
A sales head types a question about margin or churn and gets an answer built from tables already loaded into the Snowflake account.
Context that carries over
User Memory, one of the features listed at the relaunch, is meant to keep a person’s context between sessions; test what it retains.
Repeatable tasks as skills
User Skills and a shared Skill Catalog let a team package a recurring analysis once, so colleagues run it instead of rebuilding it.
Drive, Salesforce, Slack
MCP connectors to Google Drive, Salesforce and Slack, as reported, bring documents and CRM records into the same conversation.
Answers inside the rules
The agent works in the account where roles and Horizon Catalog policies are set, so data stewards govern one place, not two.
One gate for agent traffic
Cortex AI Gateway, built on Snowflake’s 2026 Natoma purchase, is Snowflake’s new layer for agent identity and MCP access control.
Watch Snowflake CoWork in action
The June 2026 CoWork introduction, the general-availability announcement under the Snowflake Intelligence name, and a plain-language demo from 2025.
Work Smarter with, Snowflake CoWork, Your Personal Work Agent
Snowflake’s own introduction to CoWork under its new name, aimed at business users rather than data engineers.
Snowflake CEO Sridhar Ramaswamy Announces Snowflake Intelligence General Availability
The GA announcement, made under the product’s former name, Snowflake Intelligence, seven months before the rename.
Talk To Your Data: Snowflake Intelligence Demo
A walk-through of plain-language questions over account data, recorded when CoWork was still called Snowflake Intelligence.
Want a live, India-context walkthrough for your environment?
Book a guided demo →Every data question becomes a ticket for a small team. CoWork lets the person asking get the answer.
Here’s what genuinely sets it apart — and exactly where it stops.
No second copy of the data to govern
Most AI analytics tools sit outside the warehouse and query into it, so permissions live in two places. CoWork runs inside the Snowflake account itself, next to the roles and Horizon Catalog policies, and that account’s storage can be in AWS Mumbai or Azure Central India (Pune).
Paid for by questions, not by seats
Snowflake bills the Cortex calls behind CoWork in AI credits by the token, at $2.00 for a Global credit or $2.20 for a Regional one. A finance head asking five questions a month costs little; a team running heavy analyses daily costs far more, so set a cap.
Adopted fast inside the existing base
General availability came in November 2025, when the product was still Snowflake Intelligence. By the quarter ended July 2026 Snowflake reported 5,800 accounts on CoWork, alongside new Cortex Sense and Cortex AI Gateway. Expect features to keep moving.
Where it stops
It needs a Snowflake account; unlike Observe it is not sold on its own. Data outside Snowflake arrives only through connectors. Snowflake prints no cost per answer, no analyst has placed it, and the October 2026 end of password-only sign-ins applies to the account it runs in.
The numbers behind the platform
What your Snowflake CoWork rollout looks like
Pick three questions that hurt
List the questions business teams wait longest for, and check the tables that answer them are already in Snowflake.
Describe the data properly
Add plain descriptions to the tables and columns involved, and confirm the roles that limit who sees which rows.
Pilot with one business team
Give a single team CoWork, log the questions they ask, and compare each answer against the analyst’s own version.
Read the AI-credit bill
Total the AI credits the pilot used, find the heaviest users and models, and set a budget alert before you widen access.
Add skills and connectors
Package repeat analyses as skills, decide which MCP connectors are allowed, and roll out team by team with a cap.
The review scoreboard
Modelled on Gartner Peer Insights structure. *Counts and breakdowns are illustrative pending verified review collection.
Quick poll — what’s driving your evaluation?
Talk to an advisor“Regional managers now ask for weekly sell-through themselves; our two analysts stopped answering the same question forty times.”
“Security liked that nothing left the account we already audit. The review took a week instead of the quarter we feared.”
“Token billing surprised us in month one when one team ran long analyses daily. A resource budget fixed it, but set it first.”
“Answers were only as good as our table and column descriptions. Cleaning those up mattered more than any prompt tweak.”
Where everyone sits — the grids
Analyst firms bury this view behind paywalls, and G2 retired its Grid. So here’s TechBag’s synthesis of the AI analytics agent market — tap any vendor to see why it sits where it does.
TechBag AI Analytics Agent Grid
Execution strength vs product vision — the classic market map, minus the paywall.
5,800 accounts by July 2026; metered in AI credits.
Data Reach × Governed Grounding
The grid nobody publishes — how far an agent reaches beyond one platform’s data vs how firmly its answers are tied to governed definitions and permissions.
One account’s rules; reaches out by MCP connectors.
Positions are TechBag’s illustrative synthesis of public review-platform data and vendor documentation — not a reproduction of any analyst graphic. Verify before relying on it.
Snowflake CoWork vs the AI analytics agents
Against ThoughtSpot Spotter, Power BI Copilot, SAP Analytics Cloud, Databricks Genie and Salesforce Agentforce — on data reach, pricing, limits, trust and India.
| Dimension | Snowflake CoWork | ThoughtSpot Spotter | Microsoft Power BI Copilot | SAP Analytics Cloud | Databricks Genie | Salesforce Agentforce |
|---|---|---|---|---|---|---|
| What it is | Agent in the warehouse | Agentic AI analyst | Copilot inside BI | BI, planning, predict | Databricks AI for users | CRM agent platform |
| Deployment | Inside your account | ThoughtSpot Cloud | Fabric capacity | SaaS run by SAP | In your workspace | Salesforce platform |
| Data it reads | Snowflake + connectors | Several warehouses | Fabric workloads | SAP-centred sources | Lakehouse tables | Customer data first |
| Pricing model | Tokens in AI credits | Quote, premium | Draws on capacity | Quoted by SAP | Free to 31 Jan 2027 | Three models |
| Published entry price | $2.00 per AI credit | Not published | F2 ~$262 a month | Not printed | $0 for users, for now | ~$2 per conversation |
| Included vs add-on | Needs a Snowflake bill | Sold with ThoughtSpot | Needs Fabric capacity | Joule licensed in | Genie Code is metered | CRM seats underneath |
| Scale and limits | Spend is the limit | Not published | Can throttle Fabric | Not published | Free window ends | Credits drawn down |
| Trust and grounding | Same-account policies | Semantic model | Models plus Q&A | SAP data layer | Domain agents | Trust layer, Data 360 |
| Integrations | MCP: Drive, Slack, CRM | Embeds in apps | M365 and Excel | SAP applications | Dashboards and Apps | Slack and voice |
| Access and sign-in | MFA mandatory now | Not detailed here | Admin and group control | Ask SAP | Principals are billed | Guardrails in Builder |
| India storage region | Mumbai or Pune | No India region | Prompts go to the US | Not published | Mumbai workspace | Hyperforce India |
| Trial and support | Trial with credits | Quote first | No trial capacity | Quote per deal | Free until 2027 | INR pricing |
| Lock-in and exit | Tied to Snowflake | Warehouse-agnostic | Tied to Fabric | SAP-centred models | Tied to Databricks | Tied to Salesforce |
| Best fit | Snowflake-first firms | Multi-warehouse BI | Microsoft estates | SAP finance teams | Databricks lakehouses | Customer-facing work |
Which approach fits you?
Honest fit signals — because the fastest way to lose your trust is to pretend one product wins every scenario.
Choose Snowflake CoWork if…
- ✓Your data already lives in a Snowflake account and you want business users answering their own questions inside it
- ✓You would rather pay per question asked than buy named-user licences for people who log in twice a quarter
- ✓The account must be stored in India, and AWS Mumbai or Azure Central India (Pune) suits your regulator
Compare alternatives if…
- ✓Your data is spread across several warehouses — ThoughtSpot Spotter queries Snowflake, Databricks and BigQuery live
- ✓Your users live in Excel and Teams on Fabric capacity — Power BI Copilot is already paid for there
- ✓You run Databricks and want a pilot at no user cost before February 2027 — Genie One is free until then
Do not expect…
- ✓A version you can buy without a Snowflake account behind it
- ✓A published cost per answer; spend depends on tokens and the model chosen
- ✓Any analyst-firm ranking for CoWork under its old or new name
TechBag has no AI analytics agents guide yet, so Snowflake CoWork sits outside the category guides. Browse all products to compare it with the rest of the catalogue. →
What do queued data requests cost you?
Drag the sliders (business users who ask data questions; their loaded hourly cost). Estimates model time lost raising, chasing and waiting on ad-hoc data requests at an assumed 1.5 hours per user a year, with 70% of it removed by self-service answers. Both figures are assumptions, and AI-credit spend is not included. Illustrative.
Loaded cost = salary + overheads per productive hour. Illustrative only — your TechBag quote models your actual environment and modules.
Three ways to consume it
Published: Snowflake bills the Cortex calls behind CoWork by the token in AI credits — $2.00 for a Global credit, $2.20 for a Regional one — with no flat fee, and CoWork needs a Snowflake account whose warehouse credits ($2–$6 by edition in Mumbai and Pune) are billed separately. TechBag pilots first, sets a monthly AI-credit cap, then quotes in INR with GST.
On demand
Best for a first pilot with one team
- AI credits by the token, billed monthly in arrears
- $2.00 Global or $2.20 Regional per AI credit
- Warehouse credits billed on their own meter
+ Platform add-ons
Best for a broader rollout
- Scoped to your estate
- Add-on modules as needed
- Phased, right-sized deployment
Capacity
Best once usage is known
- Prepaid Snowflake capacity, quoted by Snowflake
- Ask whether AI credits draw on the same commitment
- TechBag quotes in INR with GST
Buy it for less — TechBag pricing beats list
Whatever the list prices above, TechBag negotiates a significantly better deal — with GST-compliant INR invoicing and local support. Ask us for your discounted quote.
Get an India-ready quote
Tell us your requirements and current tools — we’ll model it against what you spend today.
The 8 questions to ask every vendor
Take this into your next vendor call — including ours.
Is the data already in a Snowflake account, and is that account in AWS Mumbai or Azure Central India (Pune)?
Which five questions should a pilot answer, and who checks the agent’s answers against a known result?
Are tables and columns described clearly enough for a business user’s wording to land on the right field?
Do existing roles and row policies already stop people seeing what they should not, before CoWork reaches them?
Which models will answer, Global or Regional AI credits, and what monthly cap stops one team running away with it?
Which MCP connectors — Google Drive, Salesforce, Slack — are allowed, and who approves adding another one?
Before October 2026, which service accounts still sign in by password alone, and who will re-key them?
Does the quote separate warehouse credits from AI credits? Ask for INR with GST and a capacity-versus-on-demand view.
Questions buyers ask
Ready to evaluate Snowflake CoWork?
Model how many people will ask questions and how often, or let a TechBag advisor scope a pilot that gives one business team CoWork on data you already hold in Snowflake.
Stats, ratings, review counts and pricing are illustrative and sourced from public materials; verify before purchase.