Tableau Pulse or ThoughtSpot Spotter? Compare what happens after the digest.
Both products are good at telling you that a metric moved. The evaluation question is what you can do in the next five minutes: ask why, ask about a column nobody turned into a metric, and check each number against the row it came from.
What Pulse and Spotter get right.
These are polished products from teams that have thought hard about how people read numbers. The credit is earned.
Pulse computes first and writes second
Tableau documents that Pulse uses deterministic statistical models to detect facts about a metric, and that those facts are the ground truth for the language model that writes the insight summary. That is the right architecture for numbers, and it is close to the one Ward uses.
The digest arrives where people are
Pulse sends insights to followers in email, Slack and Microsoft Teams digests. ThoughtSpot Monitor sends scheduled, threshold and anomaly alerts on KPIs, hourly to monthly. Nobody has to remember to open a dashboard.
Spotter shows its interpretation
Spotter surfaces the search tokens it used to build an answer, lets you click a token to change it, and highlights what changed on a follow-up. There is a Double check response control, and ThoughtSpot describes Spotter 3 as planning, validating assumptions and self-correcting.
Security follows the user
ThoughtSpot states that row-level security applies to KPI alert subscriptions and that Spotter respects the asking user's permissions. ThoughtSpot documents live queries against Snowflake and Databricks connections rather than copying the data in.
A digest explains a metric. The next question is usually about something else.
Both products are scoped by design. Know the scope before the first follow-up.
A Pulse metric definition is created by connecting to a published data source, with a measure, a time dimension and optional filters. Ask Q&A works on a single metric and matches your question against insights Pulse has already detected for it. Tableau Agent in Pulse, which Tableau ties to Tableau+, looks across the metrics you follow, and Tableau documents that it cannot answer questions about columns that are not used in the metrics. Pulse is available for Tableau Cloud.
Spotter answers from ThoughtSpot data models. A Spotter Analyst can be pointed at up to five of them, and in Spotter 3 cross-model search picks a source per question. That is wider, and it is still the set of models someone built.
In practice the digest tells a merchandising lead that margin fell in the West. Her next question is whether it was freight, a promotion or a supplier price change, and the answer is often in a column nobody turned into a metric, or in a system nobody modeled.
A report that regenerates, says what changed, and takes the follow-up on the same data.
This is where Ward is built differently. The difference is narrower than a feature list suggests, and it is the one that matters on a Monday morning.
In Ward, a scheduled dashboard regenerates on its schedule and describes what changed since the last run. The model writes the description, but every figure in it is a citation computed by Ward's data lake, not text the model produced. Hover any figure to see the source row and the query.
The follow-up runs on the same data. Ask why the West moved and the AI analyst plans new queries across every source you connected, not only the measure behind the report: the order database, the freight table, Stripe, a supplier contract that arrived as a PDF. Ward queries live databases in place through a proxy layer that joins them as one database, and syncs sources that are not databases into its lake.
Every answer is checked before it returns: valid data, query efficiency, and whether it answers the question. A failure loops back. When the data is too thin to support an answer, Ward suggests questions it can answer instead of guessing.
The same question, answered three ways.
An illustrative question with no customer data in it. Read the middle columns as strong: both vendors compute before they write.
The question: margin fell in the West last week. Why?
| What you get | Tableau Pulse | ThoughtSpot Spotter | Ward |
|---|---|---|---|
| How the change is found | Deterministic statistical models detect facts such as period over period change and unexpected values | Monitor alerts on a schedule, a threshold or an anomaly. Analyze change on a KPI chart runs change analysis | The scheduled dashboard regenerates and describes what changed since the last run |
| What computes the figure | The Pulse insights platform, on the metric's published data source | ThoughtSpot, as a live query against your connected warehouse | Ward's data lake, across every source you connected |
| How you check the figure | Tableau Agent answers carry citations to the metrics considered | Search tokens on the answer, formula hover, and a Double check response control | Each figure is a citation. Hover it for the source row and the query |
| Checked before it returns | Detected facts act as ground truth for the summary text | Spotter 3 plans, validates assumptions and self-corrects | Valid data, query efficiency, answers the question. A failure loops back |
| The follow-up: why | Within followed metrics. Columns not used in the metrics are out of scope for Tableau Agent | Within the selected data models, up to five per Analyst | Any source you connected that your policy allows, in the same conversation |
The difference is reach and granularity. Ward cites the row rather than the metric, and the follow-up is allowed to leave the metric.
A vendor-neutral way to test all three on your own data, scored out of 100. The 48-hour test protocol →
When Pulse or Spotter is the right answer.
Often. These are real cases, and a lot of teams are in them.
- Your metrics are defined and stable, and what people need is to hear when one moves. A digest in Slack, Teams or email is the right product for that, and Pulse does it well.
- You already run Tableau Cloud or ThoughtSpot, and your follow-up questions stay inside the published data sources or models you have built.
- You have analysts who will maintain metric definitions or data models. Both products get better in proportion to that work.
- Your data is in one cloud warehouse that ThoughtSpot already reads live. The cross-source argument on this page does not apply.
- Your users like to steer. Spotter's editable search tokens are a good interface for people who want to adjust the query themselves.
Keep the digest. Add the follow-up.
Ward does not replace Tableau or ThoughtSpot dashboards and does not try to. Teams that run both keep the digest people already read, and take the follow-up question to Ward when it crosses systems or needs a figure checked to the row. Ward connects to the same databases on read-only credentials, so adding it moves nothing.
The longer version, for the person who owns the warehouse: for data and analytics leaders. The category argument: a semantic layer an agent cannot route around.
The questions that arrive with this one.
Answered without assuming you have picked a side. Every claim about Tableau Pulse and ThoughtSpot Spotter here is checkable in their documentation.
Pulse is metric-first. Creators define metrics on a published data source in Tableau Cloud, followers get digests in email, Slack or Teams, and statistical models detect what changed before a language model writes the summary. Spotter is conversation-first: an AI analyst that answers questions from ThoughtSpot data models, shows the search tokens it used, and supports follow-ups. ThoughtSpot's Monitor feature covers the alerting side, with scheduled, threshold and anomaly alerts on KPIs.
No, according to Tableau. Its documentation says Pulse uses deterministic statistical models to detect facts about metrics, and those facts act as ground truth when a large language model writes the insight summary. The summary appears only when AI in Tableau is turned on for the site.
Within limits. Ask Q&A works on a single metric and matches questions to insights Pulse has already detected. Tableau Agent in Pulse, which Tableau ties to Tableau+, looks across the metrics you follow for a definition or data source, but Tableau documents that it cannot answer questions about columns that are not used in the metrics.
Yes. Spotter shows the search tokens used to construct the answer, lets you edit them, and lets you hover a formula to see how it is calculated. There is a Double check response control, and ThoughtSpot describes Spotter 3 as creating a plan, validating assumptions and self-correcting.
A scheduled dashboard in Ward regenerates on its schedule and describes what changed since the last run. The model writes the description, and Ward's data lake computes every figure in it. Each figure is a citation: hover it to see the source row and the query. You can ask a follow-up on the same data, across every source you connected, and every answer is checked for valid data, query efficiency and whether it answers the question before it returns.
If your metrics are well defined and your follow-ups stay inside them, you may not need one. The case for an alternative is when the next question leaves the metric: it needs a column nobody modeled, or a system outside the warehouse, or a figure checked against the row it came from. Ward is built for that case. It queries your databases in place, syncs the sources that are not databases, and links every figure in an answer to its source.
Check it rather than take it from us.
Free to deploy, 5% of model spend
No license, nothing per seat. Setup is one-time: $0 self-guided, $7,500 guided, $20,000 managed, $50,000 and up for a program.
See the numbers →The audit log, with the schema
What a complete agent audit record contains, field by field, with copy-pasteable DDL and the SOC 2 mapping.
Read the schema →The governed MCP server
The tools an agent can call, the three permission layers that gate them, and a worked example with the warehouse grants.
Read the reference →The 48-hour test protocol
A vendor-neutral acceptance test you can run against us, against them, or against both.
Run the protocol →Claims about Tableau Pulse on this page are taken from Tableau's own help documentation, and claims about ThoughtSpot from ThoughtSpot's documentation for its 26.x cloud releases, checked on 11 November 2026. Specifically: digests, channels and Tableau Cloud availability come from About Tableau Pulse; metric definitions on a published data source come from Create Metrics with Tableau Pulse; deterministic statistical models, ground truth and insight summaries come from The Insights Platform and Insight Types in Tableau Pulse; Ask Q&A, Tableau Agent in Pulse, the Tableau+ requirement and the column limitation come from Ask Questions and Discover Insights in Tableau Pulse. For ThoughtSpot: search tokens, editing and Double check response come from Getting started with Spotter; Spotter 3 behavior comes from Spotter versions; the five-model limit and permissions come from Spotter Analysts; alert types and row-level security come from Monitor; live queries come from the Snowflake and Databricks connection overviews. ThoughtSpot ships monthly, so check the version your cluster runs. No Tableau or ThoughtSpot pricing is quoted here. If you find a row that is out of date, it is a mistake rather than a position: write to brian@askotter.ai and it will be corrected or removed.
Run the test rather than take the page for it.
Deploying is free, so an evaluation costs you a credential and two days.
Read-only to start · your LLM keys · SOC 2 Type II underway · or book a call directly
Find out what your data has been hiding.
Tell us about your operation. We’ll show you the problems Ward catches, and the ones your current tools miss.