Skip to main content

Weflow's AI Features Explained: Ask Weflow AI, AI Playbooks, AI Field Updates and Agents

What Weflow's four AI features do: Ask Weflow AI, AI Field Updates, AI Playbooks and agents, and when to use each.

Weflow has four AI features. Ask Weflow AI answers your questions when you ask. AI Field Updates write what was said on one call into Salesforce fields. AI Playbooks score a whole deal against your sales methodology and keep that score current. Agents run a prompt on a schedule and send you the result.

Feature

What it reads

When it runs

Where the result goes

Writes to Salesforce

Ask Weflow AI

Records, emails, meetings, transcripts, web

When you ask

A text answer

No

AI Field Updates

One call transcript

After each recorded call

Suggested field values

Yes, automatically or after approval

AI Playbooks

Every activity and field on a record

Every two hours, on changed records

Sidebar and pipeline views

Only with write-back on

Agents

A filtered set of records

On a schedule

Email, Slack, Teams, CSV or PDF

No

Which Weflow AI feature to use

  • A question right now: use Ask Weflow AI, for example which open deals are missing methodology information

  • Keeping fields current after every call: use AI Field Updates, for fields such as next steps or a website that one call can settle

  • A field whose answer builds up across a deal: use AI Playbooks, which keep the field at the current best answer rather than the last call's, such as the champion or the metrics

  • A recurring report: use an agent, for example a Monday deal review or a monthly win/loss analysis

Ask Weflow AI explained

Ask Weflow AI answers questions across your Salesforce records, captured emails and meetings, and call transcripts. It can also search the web and read a document you attach. You point it at a single call, a deal, a filtered pipeline view, an account or your whole workspace. In the Weflow app it opens from the floating Ask AI button, scoped to the view or record on screen, and the Weflow Chrome Extension opens it on Salesforce record pages.

Ask Weflow AI Pro is included in every Weflow plan and is not metered separately. Your admin turns it on for everyone or for specific teams.

Ask Weflow AI limitations

  • Answers in text: it does not write to Salesforce

  • Question size: a question too large for one AI model is split across several, up to about twenty million tokens per question. Anything spanning your whole book of business belongs in an agent

AI Field Updates explained

AI Field Updates read the transcript of one recorded call and propose new values for your Salesforce fields where the call gave new information. Each field is set to write automatically or to wait for your approval, so an unambiguous field such as a website can update itself while a field such as the pain stays under review. Required fields stay locked. They work on Accounts, Contacts, Opportunities and Leads, and on custom objects that Weflow support sets up.

Admins build the templates per Salesforce object in Admin Console > Conversation Intelligence.

Note: AI Field Updates need a recorded call, so they come with Conversation Intelligence.

AI Field Updates limitations

  • One call at a time: each update reflects only that call

  • Replace, never append: the update replaces the field's content. To keep a record across calls, map the updates to fields on the meeting or recording object

AI Playbooks explained

AI Playbooks score a whole record, such as an opportunity, against your sales methodology by reading every email, meeting, transcript and field on it, and recommend a next step. You start from MEDDIC, BANT, SPIN, SPICED, Challenger, Solution Selling or Sandler, or from scratch, and assign each playbook to teams in Admin Console > Weflow AI > AI Playbooks. Playbooks work on accounts, contacts, leads, opportunities, cases and custom objects.

A playbook re-evaluates a record every two hours once something on it has changed. Records that existed before the playbook are scored with Generate AI Playbooks. Weflow switches AI Playbooks on for customers with Deal Intelligence or Forecasting.

Note: a playbook's result stays in Weflow unless your admin switches on write-back for it. With write-back on, the result is written to the chosen Salesforce fields automatically, with no approval step.

Weflow agents explained

Agents are built by admins in Agent Builder. An agent runs on a schedule (daily, weekly or monthly), filters a set of records such as every open opportunity in a stage, runs a prompt over their emails, meetings, transcripts and Salesforce fields, and delivers the result by email, Slack or Microsoft Teams, as a CSV table or a PDF report. A test run emails the result to the agent's own recipients with TEST RUN in the subject.

Every plan includes Agent Builder Free with 25 agent actions a month. An action is measured in tokens, roughly one million per action: every run uses at least one, a simple agent about one per run and heavy reasoning up to about five. Summaries, field updates, coaching scorecards and other standard workflows do not use agent actions.

Weflow agent limitations

  • Schedule only: a change to a Salesforce record cannot trigger an agent

  • No Salesforce writes: an agent's output reaches a record when a person enters it

  • Actions run out: when a workspace has used its monthly actions, its scheduled agents pause

Frequently asked questions about Weflow's AI features

Can a Weflow agent update Salesforce fields?

No. Agents deliver by email, Slack or Microsoft Teams. To write AI results to Salesforce fields, use AI Field Updates after calls or AI Playbooks with write-back on.

What is the difference between an AI Field Update and an AI Playbook?

An AI Field Update reflects one call. An AI Playbook reads everything on the deal and keeps the field at the current best answer, which suits anything discussed across several calls.

Does Ask Weflow AI cost extra?

No. Ask Weflow AI Pro is included in every Weflow plan.

Is my data used to train AI models?

No. Weflow never uses customer data to train AI models.

Did this answer your question?