This section walks through exactly how to configure Sai for complete meeting lifecycle automation — from connecting your tools to running your first automated scheduling + prep + follow-up loop. Each step includes what to say to Sai, what happens behind the scenes, and what you need to review before approving.
Step 1 — Connect Google Calendar and Gmail
Before Sai can schedule meetings or draft emails, it needs access to your calendar and inbox. Open Sai and type:
"Connect my Google Calendar and Gmail"
Sai will prompt you to authorize both services. You will see two separate approval requests:
- Google Calendar — allows Sai to read your events, check availability, and create new calendar entries
- Gmail (read + write) — allows Sai to search your sent folder, read email threads, draft messages, and send emails after your approval
Both authorizations use OAuth, meaning Sai never sees your Google password. You can revoke access at any time from your Google account security settings.
Important: Grant Gmail write access, not just read. Sai needs write access to send scheduling emails, follow-ups, and stale-thread re-engagements. If you grant read-only, Sai can still generate meeting briefs and draft emails — but you will need to copy-paste and send them manually.
Step 2 — Define Your Scheduling Preferences
Sai applies a set of scheduling rules every time it proposes meeting times. The default rules work for most professionals, but you can customize them. Tell Sai:
"Set my scheduling preferences: business hours 9am to 6pm Pacific, 15-minute buffer between meetings, prefer Tuesday through Thursday, no meetings before 10am or after 4pm unless I specify otherwise"
Sai stores these as persistent rules. Here is what each rule controls:
Business hours (9am-6pm). Sai will never propose a time outside this window unless you explicitly say "schedule an early morning call" or "find a time after 6pm."
Buffer time (15 minutes). If you have a meeting ending at 11:00am, the earliest Sai will propose the next meeting is 11:15am. This prevents back-to-back fatigue and gives you transition time. You can set this to 0, 10, 15, or 30 minutes.
Preferred days (Tuesday-Thursday). When multiple slots are available, Sai ranks Tue-Thu higher than Monday or Friday. Monday mornings are often consumed by planning, and Friday afternoons have the highest no-show rate. Sai still offers Monday/Friday if they are the only options.
Peak hours (10am-4pm). Sai treats 10am-4pm as the optimal window for external meetings — high enough energy for productive conversations, late enough that both coasts in the US are in business hours. Early morning and late afternoon slots are reserved for focus work unless overridden.
Meeting type durations. Sai applies default durations based on meeting type keywords:
- "intro" or "coffee chat" = 30 minutes
- "demo" or "walkthrough" = 60 minutes
- "check-in" or "sync" = 30 minutes
- "proposal review" or "deep dive" = 45 minutes
- "strategy session" or "workshop" = 90 minutes
You can override any of these. Tell Sai: "My demos are 45 minutes, not 60" and it updates the rule permanently.
Step 3 — Schedule Your First Meeting
Now test the full scheduling flow. Tell Sai:
"Schedule a 30-minute intro call with Sarah Chen (sarah@acmecorp.com) sometime next Tuesday or Wednesday"
Here is exactly what happens:
3a. Sai parses the request. It identifies:
- Contact: Sarah Chen, sarah@acmecorp.com
- Meeting type: intro (30 minutes)
- Timeframe: next Tuesday or Wednesday
- Constraints: your stored preferences (business hours, buffers, peak hours)
3b. Sai checks your Google Calendar. It pulls all events for next Tuesday and Wednesday, identifies blocked times, applies the 15-minute buffer rule, and generates a list of available 30-minute slots.
3c. Sai ranks and selects 3 options. From all available slots, Sai picks the three with the highest acceptance likelihood. The ranking logic:
- Mid-morning (10am-11:30am) and early afternoon (1:30pm-3pm) rank highest
- Slots immediately after lunch (12:30pm-1:30pm) rank lower
- Slots at the edges of the day (9am, 5pm) rank lowest
- Tuesday slots rank slightly higher than Wednesday (earlier in the week = higher commitment)
3d. Sai drafts the scheduling email. You see a preview:
Subject: Intro Call — Sai x Acme Corp
Hi Sarah,
Great connecting with you. Would any of these times work for a quick intro call?
- Tuesday May 13, 10:00am - 10:30am PT
- Tuesday May 13, 2:00pm - 2:30pm PT
- Wednesday May 14, 11:00am - 11:30am PT
Let me know which works best, or suggest an alternative. Happy to adjust!
Best, [Your name]
3e. You review and approve. Sai will not send the email until you explicitly approve. You can:
- Approve as-is
- Edit the email text (adjust tone, add context, change subject line)
- Swap out a time slot ("replace the Wednesday option with Thursday 3pm")
- Cancel entirely
3f. Sai sends and monitors. Once approved, Sai sends the email via Gmail. It then monitors your inbox for Sarah's reply. When she confirms a time, Sai:
- Creates a Google Calendar event with both attendees
- Adds a Google Meet or Zoom link (based on your default conferencing tool)
- Sends a calendar invite to Sarah
- Notifies you: "Meeting confirmed: Intro with Sarah Chen, Tuesday May 13 at 10:00am PT"
If Sarah replies proposing a different time, Sai checks your availability for that slot and either confirms automatically (if you are free) or asks you to approve the new time.
Step 4 — Enable Automated Pre-Meeting Intelligence Briefs
This is where Sai goes beyond scheduling into meeting preparation. Tell Sai:
"Before every meeting with an external attendee, generate a pre-meeting brief the evening before"
Sai will now run this workflow automatically for each qualifying meeting:
4a. Identify meetings needing prep. Every evening at the time you specify (default: 7pm), Sai scans your Google Calendar for tomorrow's meetings. It filters for external attendees — anyone whose email domain does not match your company domain. Internal standups, team syncs, and 1:1s with coworkers are excluded.
4b. Research each attendee on LinkedIn. For each external attendee, Sai opens a browser on its cloud desktop and navigates to LinkedIn. It searches for the attendee by name and company, then captures:
- Current title and company
- Time in current role
- Previous positions (last 2-3)
- Recent LinkedIn posts or articles (last 30 days)
- Mutual connections
- Education and certifications (if relevant to your industry)
This is browser automation — Sai navigates LinkedIn the same way you would. It reads publicly available profile information. It does not use the LinkedIn API or require your LinkedIn credentials.
4c. Pull email history from Gmail. Sai searches your Gmail for all threads involving the attendee's email address. It extracts:
- Date of first contact
- Date of most recent message
- Key discussion topics across the thread
- Any open commitments or questions that were never answered
- Attachments sent (proposals, decks, docs)
4d. Scan Google News for company updates. Sai searches Google News for the attendee's company name, filtered to the last 30 days. It looks for:
- Funding announcements
- Product launches or major updates
- Leadership changes (new CEO, VP hires)
- Earnings reports or revenue milestones
- Acquisitions or partnerships
- Layoffs or restructuring (flagged as a sensitive topic)
4e. Compile the intelligence brief. Sai assembles everything into a structured document and delivers it to you. The brief includes:
- Attendee profile summary (title, tenure, background)
- Relationship timeline (first contact, last contact, total interactions)
- Email thread summary with open items
- Recent company news with source links
- 3-5 suggested talking points tailored to the attendee's role and your relationship history
- Warning flags (unanswered emails, recent company layoffs, long silence periods)
4f. Delivery. You receive the brief in Sai's chat interface. You can also configure delivery via email or Slack. For a day with 4 external meetings, you receive 4 separate briefs — each taking Sai 2-3 minutes to generate, versus 15-30 minutes each if you did the research manually.
Customization options:
- "Only generate briefs for meetings longer than 30 minutes" — skips quick check-ins
- "Skip briefs for contacts I have met more than 5 times" — focuses prep on newer relationships
- "Include CRM notes if available" — Sai can pull deal stage and notes from your CRM via browser automation
- "Generate briefs 2 hours before the meeting instead of the evening before" — for same-day prep
Step 5 — Configure Post-Meeting Follow-Up Automation
After each meeting, Sai can draft a follow-up email immediately — while the conversation is fresh. Tell Sai:
"After every external meeting, ask me for key takeaways and then draft a follow-up email"
Here is the post-meeting workflow:
5a. Sai prompts you after the meeting ends. Within 15 minutes of a meeting's scheduled end time, Sai sends you a message:
"Your meeting with Sarah Chen just ended. Want me to draft a follow-up? Share a few bullet points on what you discussed, or say 'use calendar context' and I will infer from the meeting description and email history."
5b. You provide brief notes. You can be as minimal or detailed as you want:
"Discussed their Q3 eval timeline, they want a demo for their eng lead, I need to send the GitHub+Slack workflow recording, she'll share their security checklist"
5c. Sai drafts the follow-up email. Using your notes, the meeting context (title, attendees, duration), and your email history with the attendee, Sai generates:
Subject: Re: Intro Call — Next Steps
Hi Sarah,
Thanks for the great conversation today. Here is a quick summary and next steps:
Key Discussion Points:
- Your team is evaluating workflow automation tools for Q3 deployment
- Primary pain points: manual standup reports and CI/CD monitoring gaps
- Interest in seeing the GitHub + Slack integration in action
Action Items:
- [Our team] Send demo recording focused on GitHub + Slack workflow (by Thursday May 15)
- [Sarah] Share internal security requirements checklist
- [Both] Schedule a 45-minute demo for your engineering lead (targeting week of May 19)
I will send the recording over by Thursday. Let me know if anything else comes to mind!
Best, [Your name]
5d. You review and approve. Edit anything — add a personal note, adjust the tone, remove sensitive details, modify action items. Sai sends only after your explicit approval.
5e. Sai creates follow-up tasks. After sending the follow-up, Sai optionally:
- Creates a calendar reminder for your own action items (e.g., "Send Sarah the demo recording" on Thursday)
- Adds the thread to stale email detection — if Sarah does not reply within 5 days, Sai drafts a gentle check-in
Tone matching. Sai reads your original email thread with the attendee and mirrors your writing style. If your outreach was formal ("Dear Sarah, I hope this finds you well"), the follow-up maintains that register. If your original tone was casual ("Hey Sarah!"), the follow-up matches. This prevents the jarring tonal shift that makes AI-generated emails obvious.
Step 6 — Activate Stale Email Detection and Auto Follow-Up
This is the workflow that catches revenue and relationships that would otherwise slip through the cracks. Tell Sai:
"Every Monday morning, scan my sent folder for emails from the past 45 days that never got a reply. Draft follow-ups for each one."
Here is what Sai does every Monday:
6a. Scan sent folder. Sai reads your Gmail sent folder for the past 45 days. For each sent email, it checks whether the recipient ever replied in that thread.
6b. Filter out noise. Sai automatically excludes:
- Internal emails (same company domain)
- Newsletters and marketing emails
- Auto-replies and out-of-office responses
- Threads where you sent the last message as a "thank you" or closing statement (no reply expected)
- Contacts you have already met with since the email was sent (checked against calendar)
6c. Categorize by urgency. Each stale thread is classified:
- HOT (3-5 days): Recent enough for a casual bump. Draft is short and light: "Wanted to make sure this landed — any thoughts?"
- WARM (5-10 days): Needs added value to justify the follow-up. Draft includes a new angle, a relevant resource, or a reference to a recent event: "Saw Acme announced the Series B — congrats! Circling back on our earlier conversation..."
- COOL (10-21 days): Requires a topic redirect. Draft asks a new question or proposes a different approach: "Quick question on a related topic — have you evaluated X for your Q3 rollout?"
- COLD (21-45 days): Last-chance follow-up. Draft is a break-up style message that leaves the door open: "Totally understand if the timing is not right. If this becomes relevant down the road, I am here."
- DEAD (45+ days): Archived. No follow-up drafted.
6d. Cross-reference calendar context. For each stale thread, Sai checks your Google Calendar:
- Did you meet this person recently? If yes, reference the meeting: "Great chatting last Thursday — following up on the items we discussed."
- Is a meeting scheduled soon? If yes, skip the follow-up — the upcoming meeting will cover it.
- 연락처가 귀하가 주최한 웨비나 또는 이벤트에 참석했습니까? 그렇다면 자연스러운 접점으로 언급하세요.
6e. 개인화된 후속 이메일 초안 작성. 각 초안은 원래 스레드 내용, 긴급성 수준 및 캘린더 컨텍스트에 맞춰 작성됩니다. Sai는 귀하의 원본 이메일을 읽고 일반적인 템플릿이 아닌, 귀하가 직접 작성한 것처럼 들리는 후속 이메일을 작성합니다.
6f. 모든 초안을 검토용으로 제시. Sai가 일괄 처리 결과를 전달합니다:
"이번 주에 6개의 오래된 이메일 스레드를 찾았습니다:
- 2개 HOT (3-5일): Sarah Chen, David Kim
- 1개 WARM (8일): Lisa Park
- 2개 COOL (14-18일): James Ng, Priya Patel
- 1개 COLD (32일): Michael Torres
초안을 보여드릴까요?"
각 초안을 개별적으로 검토합니다. 승인, 편집 또는 건너뛰기를 할 수 있습니다. Sai는 귀하가 승인한 초안만 하루 동안 간격을 두고 보냅니다(자동화된 것처럼 보이지 않도록 한 번에 모두 보내지 않습니다).
발송 주기. Sai는 승인된 후속 이메일을 하루 종일 30-60분 간격으로 보냅니다. 이는 수신자에게 자동화임을 알리는 "오전 9시 1분에 8개의 후속 이메일 발송" 패턴을 방지합니다. 발송 주기를 구성할 수 있습니다: "오전 9시에서 오후 2시 사이에 최소 45분 간격으로 후속 이메일 발송."
7단계 — 반복 일정 설정 (선택 사항)
최대 자동화를 위해 Sai가 다음 워크플로우를 일정에 따라 실행하도록 구성하세요:
"매주 평일 저녁 7시에 내일 외부 회의를 위한 사전 회의 브리핑을 생성합니다. 매주 월요일 오전 9시에 오래된 이메일을 스캔하고 후속 이메일을 작성합니다. 모든 외부 회의 후에 메모를 요청하고 후속 이메일을 작성합니다."
이렇게 하면 세 가지 반복 자동화가 생성됩니다:
- 매일 (오후 7시): 다음 날을 위한 사전 회의 정보 브리핑
- 매주 (월요일 오전 9시): 오래된 이메일 감지 및 후속 이메일 작성
- 이벤트 트리거: 각 외부 회의 후 회의 후속 조치 프롬프트
언제든지 시간, 빈도 및 트리거를 조정할 수 있습니다. Sai에게 "오래된 이메일 스캔을 매주 수요일로 변경"하거나 "브리핑을 오후 7시 대신 오전 8시에 생성"하라고 말하면 일정이 업데이트됩니다.