Skip to content
Works everywhere, not just Gmail/Outlook. The Salesforce and HubSpot plugins only activate inside those specific email clients. Go runs on 500k+ websites and apps. A rep doing outreach in LinkedIn, a prospect's web portal, or a custom tool gets nothing from the native plugin — Go still works.Natural language synthesis, not just field display. The native sidebars surface raw CRM records. The Deal Briefer can synthesize them — "Last touchpoint was 3 weeks ago, deal is stalled at legal review, she mentioned budget concerns in the last call" — as a readable brief, not a list of fields.Indexed unstructured data. Deal notes, call summaries, email bodies, and custom text fields are often too messy to surface in a sidebar widget. The Go knowledge layer can index and semantically search all of it — meaning the brief can surface relevant context that the native plugin would never show.No additional install or IT dependency. Lightning for Gmail requires installing the extension, connecting the Salesforce org, and in some enterprise configurations, admin setup. Go is already in the browser.
Go Vision - Sales Persona
<Prompt>Here are some typical challenges that salespeople are facing.
1. Email / Inbox (~90 min/day) — The single biggest time sink. Reps receive ~117 emails/day on average (Microsoft Work Trend Index). It's the primary communication layer for prospects, customers, and internal coordination — yet most email clients offer no intelligence.
2. CRM — Salesforce, HubSpot, etc. (~75 min/day) — CRM is where 20–25% of rep time goes, mostly to manual data entry and updates. Despite the time investment, 76% of reps say less than half of their CRM data is accurate. The irony: CRM captures less than 1% of actual customer interactions.
What agents could be built with Go that would help prevent salespeople from switching tabs from email to their CRM. I'm looking for agents that will work today and agents that could be built when Go matures.
Potential CRM Agents
Works Today
🗂️ Deal Briefer
While writing an email to a prospect, the agent surfaces everything relevant from the CRM — last touchpoint, deal stage, open tasks, key contacts — without the rep ever leaving their inbox. Rep clicks the agent → it reads the recipient's name/company from the email on screen → semantic search against an indexed sync table of CRM deals, contacts, and activities → returns a compact brief inline.
SDK: setBenchInitializationSkill, ToolType.Knowledge, addSyncTable with titleProperty, index.properties on deal notes and activity history, screen context to extract recipient signal.
User might say: "What do I know about this person?", "Catch me up before I reply", "What's the status of this deal?"
<Prompt> For the Deal Briefer, does Salesforce or HubSpot offer the same feature if someone is in email?
What Salesforce and HubSpot already do in email
Both platforms have native email sidebar plugins for Gmail and Outlook. Here's what they cover:
Salesforce — Lightning for Gmail is Salesforce's official Chrome extension. It adds a sidebar to Gmail that displays Salesforce records related to the email you're viewing — contacts, leads, opportunities, cases. When you view an email, the sidebar scans the sender and recipient addresses, matches them to Salesforce records, and shows you the context. However, it's read-only for most workflows. You can view records and log emails, but you can't run Screen Flows, execute complex processes, or customize the experience beyond basic field visibility.
HubSpot — When you open an email, a sidebar appears with all the essential information about that contact: their job title, company details, past conversations, associated deals, and recent activities. It also supports email templates, meeting scheduling, and email open tracking without leaving your inbox.
So where does the Deal Briefer still have a real angle?
The native tools cover structured CRM data well — fields, stages, activity logs. What they don't do:
Bottom line for your pitch: The native tools solve "show me the record." The Deal Briefer solves "tell me what I need to know right now, in plain English, wherever I'm working." Those are meaningfully different — and the second one is harder to replicate with a sidebar widget.
📋 CRM Note Writer (Deal Update)
After a call or email thread, the agent reads what's on screen and drafts a structured CRM activity note — ready to paste or push via MCP. Rep clicks agent → it reads the email/thread → formats a clean activity log (summary, next steps, sentiment) → uses the Rewrite tool to suggest the note inline, or pushes it directly via the CRM's MCP server.
SDK: addSkill, screen context, ToolType.ScreenAnnotation with Rewrite for drafting, OR addMCPServer (HubSpot/Salesforce MCP) to push the note directly. One or the other — not both in the same agent today.
User might say: "Log this conversation", "Draft my call notes", "Summarize this thread for Salesforce"
<prompt> what is the advantage of the CRM Note Writer agent versus just attaching the email?
What "attaching the email" actually gives you
When a rep logs or attaches an email to a CRM record, the CRM stores the raw email thread. That's it. To get value from it, someone has to open the record, find the email in the activity timeline, and read it. It's an archive, not an insight.
What the CRM Note Writer does differently
1. It synthesizes, not stores. The agent reads the thread and produces a structured note: summary of what was discussed, sentiment, next steps, open questions, blockers. That's the thing a manager or future-you actually needs when they open the deal record — not the raw email.
2. It extracts structured fields, not just text. A well-prompted agent can pull out specific CRM-relevant signals: "budget mentioned: $50k," "decision maker identified: Sarah," "timeline: Q3." Those can be mapped to CRM fields, not just dumped into a notes field. Attaching an email never populates a field.
3. It creates a consistent format across all reps. Every rep writes emails differently. Some are terse, some are verbose, some bury the key detail in paragraph four. The agent normalizes all of that into the same structure every time, which makes pipeline reviews, forecasting, and handoffs dramatically cleaner.
4. It works for inbound threads, not just outbound. Attaching an email is a deliberate act — reps mostly do it for emails they send. The agent can read an inbound reply and generate a note capturing what the prospect said, which is often the more valuable signal.
The honest limitation
If all a company needs is a paper trail — "was this conversation logged or not" — attaching the email is fine and simpler. The Note Writer earns its keep when the CRM is being used for forecasting, coaching, or handoffs, where the quality of what's logged matters as much as the existence of a log.
That's the real pitch: attaching an email is compliance, the Note Writer is intelligence.
✍️ Personalization Filler
Detects placeholder tokens in outbound email drafts — [mutual connection], [recent news], [pain point] — and fills them in using indexed CRM context about the recipient. No tab switch, no copy-paste. The agent reads the draft, matches tokens to CRM data via the knowledge layer, and suggests rewrites inline.
SDK: setBenchInitializationSkill, ToolType.ScreenAnnotation with Rewrite, ToolType.Knowledge, addSyncTable of CRM contacts with indexed notes/custom fields.
User might say: "Fill in my template", "Personalize this", "What do I know about them that I can use here?"
When Go Matures
These ideas are blocked by current platform limitations but are strong future builds once those gaps close.
⚡ Inbox Intelligence (blocked: no proactive assist)
As a rep opens any email, the agent automatically surfaces deal context, sentiment history, and recommended next action in the side panel — no click required. Today the agent must be manually triggered; once proactive/assist mode ships, this becomes ambient CRM intelligence layered over every email.
🔗 Cross-App Sync (blocked: one connector per agent)
An agent that simultaneously reads email context AND writes to CRM in a single flow — using the email's MCP for thread history and the CRM's MCP for deal/contact writes. Today you have to pick one. Once multi-connector support lands, this becomes the killer "close the loop" workflow: email in, CRM updated, no tab switch ever.
📎 Attachment-Aware Briefer (blocked: no file/attachment reading)
Reads a PDF proposal or contract attached to an email and cross-references it against indexed CRM deal data — flagging mismatches in pricing, scope, or contacts. File reading is in dev but not available yet.
Where to start
Deal Briefer is the strongest first build. It requires no MCP server, no auth complexity — just index your CRM data once and the knowledge layer does the heavy lifting. The "wow" moment is immediate: rep opens an email, clicks the agent, and gets a deal brief without leaving their inbox. That's the tab-switch problem solved in one demo.
<Prompt> What agents could be built with Go that would help prevent salespeople from switching tabs from email to Conversation Intelligence (e.g., Gong, Chorus). I'm looking for agents that will work today and agents that could be built when Go matures.
Works Today
🎯 Call Brief
Before writing a follow-up email, the agent surfaces what happened on the last call — summary, key topics, objections raised, next steps — so the rep never has to open Gong to remember. Rep clicks agent → it reads the recipient from the email on screen → semantic search against an indexed sync table of Gong call summaries and transcripts → returns a tight pre-call brief inline.
SDK: setBenchInitializationSkill, ToolType.Knowledge, addSyncTable with Gong's REST API indexing call summaries + AI topics + next steps per contact/deal, screen context to extract recipient signal.
User might say: "What did we talk about last time?", "Catch me up before I reply", "What did she say about pricing?"
✍️ Follow-Up Drafter
Reads the last call summary from indexed Gong data and drafts a follow-up email that directly references what was discussed — specific pain points, commitments made, agreed next steps — as inline rewrite suggestions. No generic "great chatting" emails.
SDK: addSkill, ToolType.Knowledge (indexed call summaries), ToolType.ScreenAnnotation with Rewrite, screen context to read recipient and any draft already in progress.
User might say: "Draft a follow-up based on our last call", "Help me write a recap email", "Turn my call notes into an email"
<Prompt> Does Gong currently have the feature or capability for a Follow-up drafter agent for Gmail or Outlook?
Yes — Gong has this capability, but with meaningful limitations that leave room for the Go agent. Here's exactly what they have and where the gap is.
What Gong actually offers today
Inside Gong (call page): Gong lets you generate a follow-up email directly from the call page, with up to three variations: a general summary with next steps, an action-oriented version for established relationships, and a detailed summary for internal calls. You copy and paste the text into your preferred channel.
In Gmail (Gong Anywhere Chrome Extension): Gong Anywhere for Gmail includes tracking, the ability to browse and apply Engage templates and blocks, and an AI rephraser that gives suggestions while writing emails. It also includes an "Ask anything" feature that lets you ask open-ended questions about a contact and get AI-generated answers based on recent conversations.
In Outlook (Gong Anywhere Add-in): The Gong side panel within Outlook updates dynamically to give insights about contacts being engaged, including historical activities and summaries of previous conversations, plus an AI Composer for rephrasing suggestions.
The critical gap
Gong's follow-up email generator lives on the call page inside Gong. The workflow is: finish call → open Gong → find the call → click "Generate follow-up email" → copy the output → switch back to Gmail → paste it in. That's still a tab switch — it just moved the friction rather than eliminating it.
What Gong Anywhere brings into Gmail/Outlook is context and rephrasing, not a call-grounded draft. The rep still has to write the email themselves; the extension just helps them polish it and look up contact history on demand.
Where the Go Follow-Up Drafter is differentiated
1. The draft comes to the rep, not the other way around. The Go agent reads the recipient from the email the rep is already composing, pulls the relevant call summary from the indexed knowledge layer, and surfaces a full draft as an inline rewrite suggestion — without the rep ever navigating to Gong.
2. It works everywhere Go works. Gong Anywhere only works in Gmail and Outlook. A rep doing outreach from LinkedIn, a customer portal, or any other surface gets nothing from Gong's extension. Go runs on 500k+ sites.
3. Requires Engage license to get the good stuff. Gong Anywhere requires an Engage license to enable the extension. That's an upsell many Gong customers haven't made. The Go agent works off the base Gong REST API.
Bottom line for your pitch: Gong's version requires the rep to go to Gong first. The Go version meets the rep where they already are. That's the tab-switch problem — and Gong hasn't fully solved it yet.
🚩 Deal Risk Spotter
Reads the email thread on screen and cross-references it against indexed Gong deal intelligence — flagged risks, stalled stages, competitor mentions, sentiment shifts — and surfaces a warning inline before the rep sends. Catches things like: "Gong flagged budget concerns on your last two calls with this account."
SDK: addSkill, ToolType.Knowledge (indexed Gong deal warnings and AI-scored risks per account), ToolType.ScreenAnnotation with Rewrite to suggest copy adjustments that address the flagged risk, screen context.
User might say: "Any red flags I should address?", "What does Gong say about this deal?", "Am I missing anything before I send this?"
When Go Matures
⚡ Real-Time Call Companion (blocked: no proactive assist)
During or immediately after a call ends, the agent automatically pushes the Gong summary into the Go side panel — no click required. Rep opens their email to send the follow-up and the brief is already waiting. Once proactive/assist mode ships, this becomes ambient call intelligence layered over every post-call email.
🔗 Live Transcript Search (blocked: no multi-connector)
An agent that queries Gong's API live at the moment the rep is writing — not from a pre-indexed sync table, but fresh from Gong's search endpoint — while simultaneously reading the email client's thread via its own MCP. Today you pick one connector. Once multi-connector lands, this closes the loop between what was said on the call and what's being written right now.
📊 Coaching Signal Surfacer (blocked: no proactive assist + richer indexing needed)
Detects when a rep is writing an email that misses a coaching opportunity Gong has flagged — for example, not acknowledging a champion who was identified on the call, or failing to reference the business case the prospect mentioned. Requires both proactive triggering and richer indexed coaching metadata than most Gong REST APIs expose today.
Where to start
Call Brief is the clearest first build. Gong has a solid REST API for call summaries and AI-generated topics — indexing it into the knowledge layer is straightforward. The "wow" moment is immediate and visceral: rep opens a reply, clicks the agent, and gets a plain-English summary of the last call without touching Gong. That's the exact tab-switch this category is designed to eliminate.
The secondary pitch is that it works anywhere — not just Gmail or Outlook. A rep doing outreach from LinkedIn, a customer portal, or any other surface gets the same call brief. Gong's own app obviously doesn't do that.
Want to print your doc?
This is not the way.
This is not the way.
Try clicking the ··· in the right corner or using a keyboard shortcut (
CtrlP
) instead.