# What happened to Spindle AI The room usually agrees that the empty forecast cell is a defect. That agreement is the problem: the close on 21 November 2025 names forecasting, and it does not publish a forecast number. **Published:** 2026-07-22 **Canonical:** https://synconai.com/what-happened-to-spindle-ai **Author:** SynconAI Architecture Team ## In short Settle six things before a forecast cell is filled with a number Spindle AI did not publish. Salesforce completed the acquisition on 21 November 2025. Spindle AI is agentic analytics, placed with Agentforce observability and self-improvement, including forecasting. Forecasting is named. A forecast figure is not. The cited announcement publishes no price and no end date. A bad baseline is still a bad baseline. ## Sources - **Salesforce Signs Definitive Agreement to Acquire Spindle AI** (retrieved 2025-11-21) https://www.salesforce.com/news/stories/salesforce-signs-definitive-agreement-to-acquire-spindle-ai/ > Salesforce completed its acquisition of Spindle AI on November 21, 2025. ## What to take away - Decide the forecast cell stays empty until a reviewer can open the records behind it. - Name the person who can withdraw a published number, and agree what won means before two teams each supply a definition. - Leave both blanks empty: the purchase price, and the forecast figure. Do not use one to fill the other. - Do not move a dashboard onto a bad baseline and call that a result. A chart you cannot trace to a record should wait. ## The guide The room usually agrees that an empty cell is unfinished work. A finance partner asks for the forecast number Spindle AI will produce next quarter. The cell on her sheet is empty, and she would like it filled before the steering meeting. There is no number to give her from the announcement. Spindle AI is agentic analytics. The announcement placed it with Agentforce observability and self-improvement, including forecasting. Forecasting is named. A forecast figure is not. That agreement is the problem. Analytics on top of a bad baseline still reports the bad baseline. If the opportunities are stale, the cases are untyped, or the outcome you care about is not written down, a new chart will repeat the mess with a cleaner legend. Do not migrate a dashboard and call that a result. The close tells you who owns the company and when it happened. It does not certify a number, and it will not borrow one from another business to make the sheet look finished. The completion line is the sentence worth keeping in the brief, unchanged. "Salesforce completed its acquisition of Spindle AI on November 21, 2025." What follows keeps two cells empty. A good answer names a person, a document, a product, or a year. A DVD, a Reddit thread, a YouTube video, a TikTok, a course, a book, or a certification is not a source for the date, the price, or a shutdown. | Question | Who has to answer it | What a bad answer usually means | | --- | --- | --- | | What is the empty cell allowed to hold | The finance partner who asked for the number | A forecast figure is typed because the cell looks unfinished | | Who owns Spindle AI, and who owns the baseline | Salesforce for the company; the person who can defend the records | Two definitions of won are averaged into one certain chart | | What does 21 November 2025 decide | Whoever is writing the timeline, from the announcement | The close is reused as the morning a target was issued | | What may be written about price | Finance, with the announcement and the order form | A forecast is used to fill the missing purchase price | | Is there a shutdown to schedule | The person who can trace a chart point to a record | An old report is retired so the new number has no rival | | What has to be true before anyone forecasts | The person who accepts the baseline | A dashboard is moved, and the bad inputs move with it | ## What is the empty cell allowed to hold? The question is not whether forecasting is in scope. It is whether a figure exists to type. It does not. Salesforce bought Spindle AI. The product is agentic analytics. The announcement placed that work with Agentforce observability and self-improvement, including forecasting. Customers were not given a target, a lift, or a quarter-end figure. Observability means you can see what an agent did. Self-improvement means a change is judged against a result you defined. Forecasting means a forward view built on a baseline you trust. None of those three is a number the announcement is willing to supply. What your numbers look like is local. Some teams can open an opportunity and see a next step, an amount, and a date a colleague would believe. Some teams have a dashboard that disagrees with the meeting. The announcement cannot see either, and it will not average them. Before anyone says forecast, write the outcome in a sentence a reviewer can check. A case closed with the right reason. An opportunity with a next step a manager would accept. An agent action that matches the record, not a sentence that sounded fluent. If you cannot point from the chart back to that record, you cannot audit the number that will eventually sit on top of it. **A good answer sounds like** the finance partner leaving the cell empty, and stating that forecasting was named on 21 November 2025 and a forecast figure was not. **A bad answer sounds like** put a percentage in so steering has something to discuss. The percentage was chosen to fill a cell. The announcement did not choose it. ## Who owns Spindle AI, and who owns the baseline? Salesforce owns Spindle AI. Salesforce completed the Spindle AI acquisition on 21 November 2025. Did Salesforce buy Spindle AI? Yes. That is not a forecast engine that arrived with a pre-loaded target, and it is not a partnership still waiting on a signature. Who owns the baseline is the person who can defend the records. Salesforce owning the company does not clean your stages, your case reasons, or the fields an agent writes. If two teams define won differently, a forecast will pick one and look certain. Name the definition before you name the tool. The brand owner is Salesforce. The definition owner is yours. Who owns a published forecast, when you have one, is a finance or sales role you already recognise. The announcement does not appoint that person, and it does not supply the figure they would publish. A model is not the owner. A dashboard is not the owner. A person who can withdraw the number is the owner. The cited announcement does not publish a headcount, a founder note, or a jobs list. Working at Spindle AI after the acquisition is not answered there. A profile is not a baseline. Do not staff a forecast process from a career post. Why Salesforce acquired Spindle AI, beyond agentic analytics for observability, self-improvement, and forecasting, is not a motive you can recover from the completion line, and why the company sold is not stated there. **A good answer sounds like** a named person who can withdraw the number, and a written definition of won that both teams would accept, dated before any chart is trusted. **A bad answer sounds like** the model owns the forecast now. A model cannot withdraw a number. A person can. ## What does 21 November 2025 decide? The date is 21 November 2025. Use that form in the brief. The quoted sentence writes the day month-first, as published. Same day. It is the close. It is not the morning your forecast category changed, and it is not the day a dashboard was declared the number. Completion means the acquisition closed. It does not mean a model ran on your data. It does not mean a contract line moved, unless your order form says so. Open the order form for commercial dates. The announcement has the close only. Before that day, Spindle AI was a separate company doing agentic analytics. On it, Salesforce completed the purchase and placed the work with Agentforce observability and self-improvement, including forecasting. After it, the announcement does not invent a forecast figure, a price, or an end date. If a later Salesforce page publishes a metric, cite that page on the day you read it. Do not paste it back onto this close. **A good answer sounds like** the programme lead reading 21 November 2025 from the announcement, and keeping any target date on the finance calendar, labelled as a different document. **A bad answer sounds like** the forecast was issued on 21 November 2025. The date is the close. It is not a target. ## What may be written in the price cell? The cited announcement does not publish a price, a purchase price, or a licence fee. How much Salesforce paid, and how much Spindle AI was sold for, stay blank. A completion notice is not a consideration figure. There is no free trial on the cited announcement. What you pay for Agentforce, or for the analytics you already run, is on your agreement. Do not type zero. Do not type a forecast and call it the return on the acquisition. A return needs a baseline and a definition of the outcome. You do not have those from the announcement, and you do not have a deal price either. Two blanks. The purchase price, and the forecast number. Do not let one be used to fill the other. The time to make the baseline checkable, fields and owners and a sample a reviewer can open, is the work. It is not a consideration. **A good answer sounds like** the finance partner presenting with both cells empty, and saying what they need from the records before either cell can be filled. **A bad answer sounds like** use the forecast as the value of the deal, since the price was not published. That fills one honest blank with another invention. ## Is there a shutdown to schedule? No end date is published for Spindle AI. There is no sunset quarter, no instruction to switch a dashboard off, and no instruction to treat an old report as retired because a company was bought. A search for shutdown does not create a notice. You can stop using a chart that nobody can trace to a record. That is a quality decision. It is not a product sunset. Leave the chart in place until the replacement can be opened back to the same records. A migration of pictures, with the baseline still bad, is how the old argument survives under a new title. On a risk register, write four lines. No end date. No published price. No forecast number. A bad baseline is still a bad baseline. Then name the outcome you would forecast, and the object that holds it. If you cannot, the risk is the missing outcome, not a hidden shutdown. ::: warn Do not publish a Spindle AI forecast number, a purchase price, or a shutdown quarter. The close is 21 November 2025. Forecasting sits on a baseline a reviewer can open. A dashboard moved onto a bad baseline still reports the bad baseline. ::: **A good answer sounds like** the baseline owner stating there is no published end date, and that the current chart stays until a reviewer can open the same records from the replacement. **A bad answer sounds like** retire the old dashboard this quarter so the new number has no rival. The quarter was chosen so the invented figure would not be argued with. ## What has to be true before anyone forecasts? When the lines above are accepted, name the outcome, the record that proves it, and the person who may say the baseline is good enough to forecast from. Ask the partner to present with both cells empty, and to open ten records before either cell is filled. A pass asks to open those records. A fail arrives with a number already chosen. A demo after the acquisition is a trace from a chart point to a record, or a demo Salesforce will show you now. A roadmap is a Salesforce page read on the day you read it. Self-improvement, when you reach it, is a change you can reverse because you kept the baseline. It is not a promise that the next chart will be higher. **A good answer sounds like** ten records a reviewer can open from the chart, the outcome written in one sentence, and the person who can withdraw the number if the records do not support it. **A bad answer sounds like** move the dashboard and type next quarter's figure. The baseline moved with the picture, and the figure is not on the announcement. ## What to do with the answers Read the six as a set. Specific answers mean the baseline has an owner and both cells are still empty. General answers mean the decision rests on an empty cell being treated as a defect. If the answers are specific, the next discipline is the outcome and the record, set out in [Spindle AI and Agentforce](/spindle-ai-agentforce). If they are not, the useful outcome is a sheet whose forecast cell you can still clear. Keep this page for the lines a finance partner will check: owned by Salesforce, completed 21 November 2025, agentic analytics for observability and self-improvement, including forecasting, price not published, no end date, no forecast number. ## Questions ### What happened to Spindle AI after Salesforce bought it? Salesforce completed the acquisition on 21 November 2025. Spindle AI is agentic analytics. The announcement placed it with Agentforce observability and self-improvement, including forecasting. Observability means you can see what an agent did. Self-improvement means a change is judged against a result you defined. No forecast number is published. A chart on a bad baseline still reports that baseline. ### Is Spindle AI owned by Salesforce? Yes. Salesforce completed the Spindle AI acquisition on 21 November 2025. That is not a forecast engine with a pre-loaded target. Ownership of the company does not clean your stages or appoint a forecast owner. If two teams define won differently, name one definition. The person who can withdraw a number is still yours to name. ### When did Salesforce acquire Spindle AI? The cited announcement completes the acquisition on 21 November 2025. That is the close. It is not the day a forecast was issued, it is not the morning a forecast category changed, and it is not a shutdown date. It also does not mean a model ran on your data that day. ### How much did Salesforce pay for Spindle AI? The cited announcement does not publish a price, a purchase price, or a licence fee. Leave the cell blank. Do not fill it with a forecast figure. That figure is not on the announcement either. Do not type zero. There is no free trial. What you pay for Agentforce, or for analytics you already run, is on your own agreement. ### Did Salesforce shut Spindle AI down? No end date is published for Spindle AI. There is no sunset quarter. Stopping a chart you cannot trace to a record is a quality decision. Leave that chart until a replacement opens back to the same records. A migration of pictures, with the baseline still bad, keeps the old argument under a new title. --- SynconAI is a Salesforce Select Partner and an OpenAI Select Partner, delivering architect-led Salesforce and AI programmes across the United States and Australia. Related service: Agentforce consulting (https://synconai.com/salesforce-agentforce-consulting).