/

Sales

Ask Your Pipeline in Plain English: AI Tools for Pipeline Questions (2027)

Lokesh Sharma

Lokesh Sharma

Product Manager

·

8

min read

Ask Your Pipeline in Plain English: AI Tools for Pipeline Questions (2027)

Yes. Several AI tools let you ask questions about your sales pipeline in plain English and get an answer in seconds. HubSpot's Breeze Assistant answers from CRM records, Gong's Ask Anything from calls and emails, and Proshort's Ask Aria from deals, calls and CRM records together, with sources shown. How accurate the answer is depends on which data the tool can read.

Disclosure: we build Proshort, one of the tools below. Every competitor capability and price was checked on the vendor's own site or help center in November 2026.

Key takeaways

  • Natural-language pipeline Q&A turns a question such as "Which deals have been in the same stage for more than 30 days?" into a search of your CRM and, in some tools, your calls and emails.

  • The data behind the answer matters more than the chat box. CRM-only assistants miss what buyers said; call-only assistants miss stage, amount and close date.

  • Trust answers that show their sources. Of the tools checked, Proshort and Backstory state that answers come with sources.

  • Before rollout, test every tool with questions you already know the answers to.

What is natural-language pipeline Q&A?

Natural-language pipeline Q&A is an AI assistant that answers questions about deals, accounts and forecasts typed in everyday English, instead of making you build a report or a filter. It works in three steps:

  1. Interpret the question. Which records (my deals, this quarter, a stage) and which measure (count, value, risk, reason).

  2. Retrieve the evidence. Query CRM fields and, depending on the tool, search call transcripts, emails and notes.

  3. Answer. Return text, a table or a chart, ideally with links to the records and call moments it used.

It replaces the Monday spreadsheet export and the "can you pull a report" message to RevOps. It does not replace judgment: the answer is only as good as the data and the question.

What questions can you ask your pipeline in plain English?

Useful questions fall into five groups. Examples marked with a vendor come from that vendor's own pages; the others are illustrations.

  • Pipeline hygiene: "Which deals have been in the same stage for more than 30 days?" and "What's the total pipeline value for deals closing this quarter?" (HubSpot).

  • Deal attention: "Which deals need my attention today?" (Proshort); "Which late-stage deals have no meeting booked in the next two weeks?"

  • What buyers said: "Which deals mentioned pricing?" and "Where did competitors come up?" (Salesforce); "What did the CFO ask for on the last call with this account?"

  • Team and coaching: "Rank my reps by closed-won revenue this quarter" (HubSpot); "Who needs coaching this week?" (Proshort).

  • Call prep: "Brief me on all my upcoming calls today" (Proshort).

The first group needs only CRM fields. The third needs conversation data. The second and fourth are best answered with both. For prompts to use in a weekly review, see our pipeline review prompts.

What does a good pipeline answer need?

  • CRM and conversation data together. Stage, amount and close date live in CRM fields; the reason a deal is stuck usually lives in calls and emails. Gong's help center says Ask Anything "doesn't take into account any information synced from CRM fields" (Gong), while HubSpot's Slack help article lists CRM objects, notes, meetings and emails as Breeze Assistant's data (HubSpot).

  • Sources you can click. Every claim should link to the CRM record, email or call moment behind it. Without sources, a wrong answer looks exactly like a right one.

  • A visible scope. Know how much the tool reads. Gong documents up to 60 calls and 500 emails per deal or account.

  • Permissions. A rep should not get answers about deals they cannot open in the CRM.

  • Freshness. An answer given this afternoon should include this morning's call and stage change.

  • Usable output. A table to paste into a review, a chart, a draft email, or a scheduled brief.

Which AI tools let you ask pipeline questions in plain English?

Tool

Where you ask

Data it reads (per vendor)

Scope

Sources shown

Published price (checked Nov 2026)

Proshort (Ask Aria)

Proshort; Proshort app in ChatGPT; Claude (MCP)

Every deal, call and CRM record

Any record (tag with @) or across your deals; questions can be scheduled

Yes

Included in the $75 per rep per month platform

Gong (Ask Anything)

Gong homepage, deal boards, account pages, mobile app

Calls and emails (up to 60 calls and 500 emails); not CRM fields

One deal, account or contact at a time

Not stated

Not published; any Gong plan, paid seat required

HubSpot (Breeze Assistant)

HubSpot and Slack

CRM objects (deals, contacts, companies, tickets), notes, meetings, emails

Pipeline-wide: counts, totals, stalled deals, rep rankings

Not stated

Included with every HubSpot subscription at no extra cost

Salesforce (Call Explorer)

Sales Cloud

Calls, emails and meetings captured by Conversation Intelligence

Across conversations

Not stated

Conversation Intelligence included in every Sales Cloud edition

Backstory (formerly People.ai)

Claude, ChatGPT, Copilot, Slack; MCP server

Every call, email and CRM record

Across the quarter's pipeline

Yes: meeting, stakeholder and dollars at stake

Not published

Salesloft (with Clari)

Your own AI tool through the Salesloft MCP server

Pipeline, deals, accounts and call data

Any deal or rep

Not stated

Not published

Avoma (Ask Avoma)

Avoma

Accounts and deals (Revenue Intelligence add-on); conversations (Conversation Intelligence add-on)

Account and deal level

Not stated

Revenue Intelligence add-on $29 per seat per month, annual, on a plan from $19

"Not stated" means the vendor page we checked does not say; ask in the demo. Gong publishes no prices; third-party estimates put its license at about $1,360 to $1,600 per user per year at list plus a platform fee.

How do the main options differ?

Proshort Ask Aria: CRM and calls in one answer

Ask Aria sits on top of every deal, call and CRM record. You ask in plain English, tag any record with @, and get an answer with sources as text, a chart, a table or a ready-to-send doc. Questions can run on a schedule, for example a daily 8:00 email. Best for: leaders and RevOps teams whose questions need both what the CRM says and what buyers said.

Gong Ask Anything: deep on one deal's conversations

Gong's help center describes natural-language questions about deals, accounts and contacts, answered from calls and emails. You pick one deal or the whole account; for a wider answer, Gong suggests switching to all deals associated with the account. Best for: Gong customers inspecting individual deals before a review.

HubSpot Breeze Assistant: pipeline math from the CRM

HubSpot lists questions such as open deals in the "Proposal" stage, deals closed last month by rep, and deals stuck in a stage for more than 30 days, asked in HubSpot or in Slack (Breeze Assistant). Best for: HubSpot teams that want fast pipeline counts and totals at no extra cost.

Salesforce Call Explorer: search what was said

Salesforce says Call Explorer lets you search with natural language, asking things like "Which deals mentioned pricing?", to surface key moments, risks and pipeline trends (Salesforce). Best for: Sales Cloud teams that record calls in Salesforce.

Backstory and Salesloft: answers inside your AI assistant

Backstory (the new name for People.ai) says it reads every call, email and CRM record across the quarter and answers in Claude, ChatGPT, Copilot or Slack, with answers sourced to the meeting, the stakeholder and the dollars at stake (Backstory).

Salesloft, which now lists Clari Forecast among its products (clari.com redirects to salesloft.com), says its MCP server lets your AI answer questions about any deal or rep, grounded in call transcripts and engagement data (Salesloft). We could not find a current vendor page describing a built-in Ask Clari feature, so it is not listed separately. Best for: teams that want pipeline answers inside the AI tools they already use.

How do you check whether the answers are accurate?

  1. Build a test set. Write 15 to 20 questions with known answers from last quarter: a deal that closed, one that slipped, one lost for a known reason.

  2. Ask in your managers' words. Use the shorthand your team actually uses, not tidy demo phrasing.

  3. Score every answer. Correct, partly correct, wrong, or "not enough data". An honest "not enough data" beats a confident wrong answer.

  4. Click every source. Confirm the call moment or CRM field says what the answer claims.

  5. Test the gaps. Ask about a deal with no recorded calls, a deal with stale CRM fields, and a rep who logs little.

  6. Test permissions. Sign in as a rep and ask about a deal they should not see.

  7. Re-run after a week. Check that new calls and CRM changes appear in the answers.

Where Proshort fits (and where it does not)

Proshort fits when the answer to a pipeline question sits partly in the CRM and partly in what buyers said. Ask Aria answers from every deal, call and CRM record with sources, and the deal pipeline review shows every active deal with probability, 14-day momentum and MEDDICC, with evidence from the calls rather than guesses in the CRM.

You can also ask about calls, deals and accounts from ChatGPT or Claude. The same connected graph of calls, objections and wins powers Aria Live, which hands the rep a sourced answer during the call when the buyer asks.

"The contextual insights it delivers are not only precise but also perfectly timed." Kanika Bhalla, Chief Revenue Officer, Increff

Read the Increff case study. Proshort is not the right first step if your questions are only counts and totals over CRM fields and you run HubSpot: Breeze Assistant is included in your subscription. If your main need is a formal forecast roll-up process, a dedicated forecasting product is built for that job.

Frequently asked questions

Can I ask questions about my sales pipeline in plain English?

Yes. HubSpot's Breeze Assistant, Gong's Ask Anything, Salesforce's Call Explorer, Backstory (formerly People.ai), Avoma and Proshort's Ask Aria all accept questions typed in everyday English. They differ in the data they read: some use only CRM records, some only calls and emails, and some combine both. Check which data a tool reads before trusting its answers.

Are AI answers about the pipeline accurate enough for a forecast call?

Only after you test them. Ask 15 to 20 questions you already know the answers to, click every source, and check deals with missing calls or stale CRM fields. Use answers that show their sources for preparation and inspection, and keep the commit number a human decision.

Can ChatGPT or Claude answer questions about my CRM and calls?

Yes, through connectors. Proshort lets you ask about calls, deals and accounts from its app in ChatGPT or from Claude over MCP. Backstory answers in Claude, ChatGPT, Copilot and Slack, and Salesloft's MCP server is listed in Claude's connector directory. The answer can only be as good as the data each connector exposes.

What is the difference between Ask Aria and Gong Ask Anything?

Gong's help center says Ask Anything answers from calls and emails for one deal, account or contact at a time, up to 60 calls and 500 emails, and does not use CRM field data. Ask Aria reads every deal, call and CRM record, lets you tag any record with @, shows sources, and can run questions on a schedule.

Is CRM data enough to answer pipeline questions?

For counts, totals and stage ages, yes: a CRM assistant handles those well. For why a deal is stuck, who the real decision-maker is, or whether pricing came up, you need call and email data, because that information rarely reaches CRM fields. The most useful tools combine both sources.

To ask Aria about your own pipeline, book a 30-minute Proshort demo.

Share

Lokesh Sharma

Written by

Lokesh Sharma

See Aria on your own calls
30-minute demo. Your work email and a time that suits you.
Book a demo