Automate pre-call research by using a booked meeting to identify the correct CRM account, retrieve previous conversations and commitments, check relevant public changes, and prepare a short, sourced brief for the meeting owner. Put five useful facts ahead of a company dossier: who is coming, why they booked, what was promised, what changed, and what remains unknown. Context engineering connects that evidence to your offer and sales rules. Bombos is the memory layer that keeps your company context and supplies the relevant part when preparation starts. Keep human approval as the default for actions.
By Bombos. Published November 12, 2026. Sources checked November 12, 2026.
A call appears between a product discussion and a forecast review. You open the website, LinkedIn and the CRM. The useful question is whether the buyer is still waiting for the answer your colleague promised last week. Build preparation around that continuity.
The workflow, templates and selection criteria below are our recommendations. Every scenario is an illustrative example, not a customer case or measured result. Product capabilities come from the linked makers; their descriptions do not establish independent performance. The distinction matters when deciding what to buy.
What should automated meeting prep help a seller do?
Automated meeting prep should help a seller understand the buyer’s situation well enough to ask a relevant question and honor previous commitments. The buyer should leave with a clearer understanding of the problem, the available choices and the next useful step, whatever they buy. Design the brief for that conversation.
Begin with the relationship: booking answers, introductions, earlier messages, call notes, unresolved objections and promised work. Then examine the company’s own description, relevant announcements and the attendee’s professional responsibilities. Public facts become useful when they change a question you would ask. Avoid a personalized opening that merely demonstrates how much information you collected.
Choose research categories from the decision ahead: business model and markets for relevance; an announced product or expansion for changed priorities; hiring and leadership for possible ownership changes; documented technology for integration questions. Include funding only when its date and implications are relevant. Treat a technology-detection result as evidence about the observed website, not a complete map of internal systems. Stop researching when another fact would not change the conversation you are preparing.
Example: a company announces expansion into another market. Your previous call says the immediate difficulty is an approval process that crosses departments. Prepare a question about whether expansion changes that process. Do not replace the buyer’s stated problem with an invented expansion strategy, buying deadline or budget.
Separate a discovery call from validation, consensus building and risk reduction. For an SDR, a useful context card might take thirty seconds to scan: verified identity, reason for booking, one relevant fact and the question to resolve. An account executive may need an expandable stakeholder map and evidence behind each open requirement. These are proposed reading budgets, not measured productivity claims.
Would your next buyer recognize their own situation in the brief, or mostly recognize your sales pitch?
What belongs in a useful pre-call research brief?
A useful pre-call brief contains a small set of account-specific facts, their sources, an explicitly tentative interpretation and questions that could disprove it. Keep the main card readable in the two minutes before a call, with deeper evidence behind links. A blank answer should identify what needs learning rather than invite the model to complete the story.
Use this example brief template:
| Part | Put on the main card | Keep available behind it |
|---|---|---|
| Who and why | Confirmed account, attendees, meeting purpose and booking reason | CRM account, contact and opportunity identifiers; invitation and booking answers |
| Relationship | Last relevant exchange, unresolved objection and outstanding promise | Message link or attributed transcript passage, date, owner and completion status |
| Relevant change | One dated development that could affect this conversation | Original company announcement, role page or other attributable evidence |
| Working hypothesis | Why the change might matter to this account and your supported offer | Supporting evidence, contradictory evidence and what remains unconfirmed |
| Questions and next step | Three priority questions and a possible next step to discuss | Stage-specific requirements, approved proof and boundaries on commitments |
Attach a compact evidence label to each substantive statement: source, event date, verification date, entity, status and relevance. Status can be buyer-confirmed, documented public fact, seller note, hypothesis, disputed or unknown. Confidence should explain the evidence: a direct statement in the latest call is different from a role inferred from a title. Avoid an unexplained percentage that looks calibrated but is simply generated.
Worked example, entirely illustrative: the booking answer says an approval process is slowing delivery. A previous transcript attributes a request for a security document to the operations contact. A current job advert describes an internal automation project. The brief should preserve all three separately. The advert supports a question about an existing alternative; it does not establish that the buyer wants your software.
The resulting question could be: would it help to compare the internal project with an external option, or is the meeting about another part of the process? The disconfirming signal is explicit: an existing solution already meets the requirement. If that is true, the useful next step may be sharing a document or ending the evaluation.
For objections, prepare the unanswered issue, the approved evidence and the person who can resolve it. A concern about access permissions calls for a precise explanation of access permissions. Do not let the model translate it into resistance to change. An objection is information about the conditions a good solution must satisfy.
What event should trigger research, and how do you handle reschedules?
Start pre-call preparation when an eligible external meeting is booked, then recheck the current event before delivering the brief. Keep the meeting identity, current schedule and account match together. A research job created at booking should never assume that the time, owner or attendee list will remain unchanged.
The scheduling systems provide different starting points. Calendly documents invitee.created and invitee.canceled; rescheduling triggers both. Chili Piper documents separate new-meeting, update and cancellation webhooks, including reassignment or rescheduling. For HubSpot Meetings, the sales workspace exposes connected-calendar meetings. Choose the CRM meeting record or connected calendar as your operational starting point after confirming the events your configuration supplies.
Use this example operating policy, rather than treating every notification as another research request:
| Event or condition | Recommended response |
|---|---|
| New eligible external meeting | Create one preparation record, match the account and collect available history |
| Same event received again | Update that record; do not deliver another brief |
| Rescheduled meeting | Associate the replacement with the earlier booking, invalidate the old delivery time and schedule a fresh check |
| Cancellation | Stop pending delivery and mark any existing card as canceled |
| New attendee or reassigned owner | Recheck account association, relevant history and recipient permissions |
| Booking shortly before the call | Prioritize booking answers and verified relationship history; mark external research incomplete |
| No unambiguous account match | Request a match from the owner and withhold account-specific private history |
Example timing policy: collect stable context at booking, refresh changing evidence on the meeting day and deliver twenty minutes before the call. Read the latest event again immediately before delivery. Store the actual start time with its timezone, and distinguish it from the briefing time. These intervals are a starting design, not universal best practice.
Use the provider’s meeting identifier and the relevant workspace or calendar as the meeting key. Store the provider’s relationship between an old and replacement booking when available. Reuse approved company context across meetings, but create a separate meeting-specific brief for each purpose and attendee group. Two calls with the same account can concern different opportunities.
Example event record: provider, calendar, event ID, last-updated timestamp, replacement-booking reference and delivery status. Process changes to the same event in sequence, rejecting updates older than the stored timestamp. Before sending, compare the job’s event details with the latest record and suppress any delivery based on superseded details.
Deliver a link into a private place the owner already checks, such as their CRM workspace or approved team channel. Keep internal objections and relationship notes out of an invitation visible to the prospect. If research is unavailable, deliver the useful confirmed history with an explicit gap, and alert the owner when even that cannot be supplied.
How do you prevent research landing on the wrong account?
Prevent wrong-account research by requiring an explicit match between the attendee, the intended company and the CRM relationship before retrieving private history. Use identifiers and evidence together. A similar name, a logo or a familiar domain fragment is insufficient to establish which organization and opportunity the meeting concerns.
Our recommended matching order is the linked booking or CRM contact, the verified work email and canonical company domain, then the relevant account and opportunity identifiers. Confirm subsidiaries and consultants separately. A parent company’s announcement can be relevant context without being evidence that its subsidiary shares the same buying process.
Example: an attendee books with a personal email and writes only a common company name. Show the booking answer and possible matches to the seller. Do not select the most prominent search result or expose the history of another customer. A useful uncertainty message states what would resolve it: the work domain, correct contact or meeting purpose.
Keep the booker, attendee and buyer roles separate. An assistant booking on behalf of an executive is not automatically the decision maker. A consultant’s email identifies the consultant, not necessarily the client they represent. Where multiple opportunities exist, ask which one the call serves rather than silently choosing the latest.
An empty contact record should prompt a search across permitted account associations and prior attendees, not the conclusion that no relationship exists. Preserve the scope of that search: no relevant conversation in the connected sources is a defensible statement. No previous conversation ever happened is a much stronger claim.
Which wrong company could your current process confidently confuse with the account on tomorrow’s calendar?
What company context does the research agent need?
Context engineering for meeting prep means supplying the agent with the company knowledge, account evidence and decision rules needed for this particular conversation. Public account research answers what a prospect does. Company context answers why it matters to your offer, what your team already knows and what it may responsibly promise.
Keep durable shared knowledge in an approved company brief: supported use cases, ideal-customer criteria, exclusions, implementation requirements, product limits, approved evidence and sales-stage definitions. Keep live ownership, relationships and opportunity status in the CRM. Keep attributed conversation evidence linked to its recording or original note. This is our proposed source map, not a claim that any connected system is automatically correct.
| Context to supply | Example instruction for the agent |
|---|---|
| Current offer and exclusions | Suggest questions only about supported work; identify a requirement outside scope |
| Segment experience | Use an approved, relevant buying or churn lesson as a hypothesis, without exposing another customer’s private information |
| Account commitments | Surface the promised security answer before suggesting another demonstration |
| Evidence policy | Separate buyer statements, seller interpretation and generated inference |
| Sales method | Identify missing evidence for the next stage; do not invent it |
| Access and action policy | Read only permitted sources; prepare proposed changes for human approval |
Include the previous correction, not just the previous summary. Example: a rep corrects an earlier brief because a contact advises the buyer but cannot approve the purchase. Keep the corrected role with its evidence and date. Future preparation should ask who approves rather than promote the advisor back into that role.
Gong’s brief types use questions and optional instructions, with custom types managed by administrators. Clay’s account agents describe persistent memory per account. These capabilities support explicit context design; neither establishes that every company’s rules are understood without configuration.
Bombos’s role is to keep that shared company context and supply the relevant part when the preparation task needs it. The aim is that no seller has to repeat the same background for each call. Start with giving AI agents your company context before adding more research sources.
Which promise would a colleague miss if they prepared tomorrow’s meeting using only the CRM summary?
Can HubSpot, Salesforce, Gong, Clay, Apollo or Sales Navigator already do this?
Existing sales products can cover substantial parts of pre-call research, so compare their output with your brief before adding another automation layer. Distinguish account research, conversation history, scheduled preparation and delivery. A tool that answers a research question on demand has not necessarily been configured to deliver an updated brief before every booking.
| Existing product | Documented capability | Boundary to check in your environment |
|---|---|---|
| HubSpot | The sales workspace supports preparation, attendee and deal information, suggested activities and prospect insights | Standard preparation is documented for Sales Hub Professional and Enterprise; confirm calendar connection and actual access |
| HubSpot’s additional preparation experience | Pain points from earlier emails, notes, calls and transcripts; recent engagements | The additional experience is labeled beta, requires an assigned Sales seat and Sales Access, and requires an external CRM contact attendee |
| Salesforce Sales Account Research | External company research within Salesforce | Setup requires Einstein generative AI, Data Cloud, the Record Research component and Research Records permission; edition and Agentforce entitlement matter |
| Gong AI Briefer | Structured account, deal, call and contact summaries using conversations, emails, web information and CRM data | Gong Foundation and seat access apply; configure scope and questions, and confirm the upcoming-call delivery you need |
| Clay Account Research Agents | Ongoing research across an Audience segment, combining connected CRM, transcripts and signals into structured fields with account memory | Open beta; needs a connected data source and defined segment, with access to Audiences; account monitoring is distinct from a calendar-specific brief |
| Apollo pre-meeting insights | Templates for priorities, decision makers, company information and objections from prior meetings | The documented workflow covers external meetings and selecting templates, running and reviewing insights; a confirmation email links to those insights |
| LinkedIn Sales Navigator Account IQ | Account summaries from LinkedIn and public information, with product personalization | Coverage varies; available with Core, Advanced and Advanced Plus, subject to administrator controls; public insights do not establish your private relationship history |
For Salesforce, the setup documentation lists Enterprise, Performance and Unlimited with an Agentforce for Sales or Industry add-on, or inclusion in Agentforce 1 Sales Edition. Do not assume a general AI subscription enables this component. For every product, ask for the actual account, meeting type, sources, access restrictions and recipient to be demonstrated together.
Example buying decision: your existing HubSpot preparation contains the right account history and questions. Use it. If the missing piece is a reminder or delivery channel, add that narrow step. If the wrong assumptions come from an outdated description of your offer, correct the context. Another enrichment subscription will not answer that problem.
Do Fathom, Fireflies, Granola and Otter prepare upcoming meetings?
Meeting assistants differ in how they use earlier conversations for upcoming preparation. Some document a dedicated pre-meeting experience; others document access to transcripts and summaries that you can use in your own workflow. Check the actual feature rather than assuming a recording assistant either does everything or only writes after-call notes.
| Product | What its own documentation supports | Practical implication |
|---|---|---|
| Fireflies | Meeting Prep combines available company, participant, email and previous-meeting context; briefs generate about three hours before the meeting, with an on-demand option | Requires calendar access; company and participant insights depend on corporate email domains; late bookings may need on-demand preparation |
| Fathom | Its public API provides summaries, transcripts and action items for workflows | Useful for grounding a brief in earlier meetings; user-level keys cannot access other users’ unshared meetings, including through an administrator key |
| Granola | Its usage guide describes external-meeting Briefs with attendees, previous discussion and open items; connected Gmail can add relevant email context | Start with calendar and profile setup; inspect the brief’s actual account coverage and sharing rules |
| Otter | Its Work Mode product description describes pre-meeting briefs using transcripts and connected apps | A vendor-described preparation option; confirm which apps, permissions and delivery behavior apply to your workspace |
Fireflies describes email context from the last ninety days with up to five external attendees. That is a useful documented boundary: it should not be mistaken for a complete account correspondence archive. Its preparation uses AI credits, and available information can be incomplete.
Example: Fathom has the previous buyer conversation, but your workflow can access only the meeting owner’s shared recordings. Preserve that access boundary. Do not solve an incomplete brief by copying everyone’s private transcripts into an unrestricted folder. Ask whether the missing evidence can be appropriately shared or whether the seller should supply a short approved note.
When should you use n8n or Zapier for meeting preparation?
Use orchestration when you have a clear gap between the booking event, permitted sources, synthesis and the place the seller needs the brief. An automation platform coordinates those steps. You still need to define identity, evidence quality, access and how incomplete work is reported.
n8n’s Google Calendar trigger documents created, updated, canceled, started and ended events. Zapier’s Google Calendar integration includes a trigger for a specified time before an event starts. It also distinguishes instant triggers from polling triggers; some polling checks occur every fifteen minutes on its entry plan. Select the exact event and cadence your preparation window requires.
Example workflow: booking notification, account match, permitted CRM and conversation context, targeted public research, evidence-linked synthesis, owner review. Store a preparation status and source links. Before notifying the owner, recheck the event and recipient. If the CRM step fails, report the missing relationship context instead of generating an apparently complete public-only brief.
Reuse unchanged account evidence, refresh the fields that can affect the upcoming decision, and keep retries from creating duplicate notifications. Assign an owner for failed jobs and connector access. A diagram connecting five tools is incomplete until someone can tell which meetings did not receive useful preparation.
For occasional meetings with little account history, a maintained template and manual source selection may be the better starting point. For repetitive meetings already handled by a native feature, keep that feature. Add orchestration when its reduction in repeated work exceeds its setup, maintenance and review effort in your own measurements.
How should preparation change between discovery, demos and follow-up calls?
Meeting preparation should follow what the buyer needs to decide next. Discovery explores the situation; validation checks whether the proposed approach meets a requirement; later conversations resolve ownership, agreement or risk. Use qualification frameworks to identify missing evidence, while letting the conversation follow the buyer’s answers.
MEDDPICC covers metrics, economic buyer, decision criteria, decision process, paper process, pain, champion and competition. In our proposed brief, these are evidence categories. An unanswered category stays unknown. A job title alone does not confirm buying authority, and an enthusiastic attendee is not automatically a champion.
| Meeting | Research priority | Example question, not a script |
|---|---|---|
| First discovery | Booking reason, current process and observable consequences | What happens between the request arriving and someone being able to act? |
| Demonstration or validation | Buyer-confirmed requirement, applicable product capability and unresolved constraint | Which part would you need to see working before this becomes a credible option? |
| Stakeholder alignment | Known roles, competing priorities and evidence each person needs | Who else needs confidence in this approach, and what would they need to establish? |
| Follow-up or risk discussion | Previous promises, completed work, objections and changed circumstances | Did the material we sent resolve the concern, or did it expose another question? |
Prepare competing alternatives, including the current process and an internal project. Use approved proof where it matches the requirement, and name the mismatch where it does not. For renewals, start with the existing commitment and unresolved delivery issues before treating a public growth signal as an expansion opportunity.
Keep initial fit and assignment in inbound lead qualification and routing. After the call, carry confirmed changes into automatic CRM updates after sales calls. Preparation should consume that relationship memory rather than reproduce either workflow.
What evidence would make your team abandon its preferred hypothesis about this account?
How do you prevent stale facts and invented buyer intent?
Prevent unreliable research from becoming sales certainty by preserving the difference between a sourced statement and its interpretation. Confirm the entity, date and meaning of every fact the seller might repeat. Treat a generated summary as a route to evidence rather than evidence in its own right.
Apply this example freshness policy: recheck attendees, current roles, meeting time and outstanding promises before the call. Keep the original date on a funding announcement rather than calling it recent each time the brief runs. Revisit product capabilities whenever an approved offer changes. A stable business description may need less frequent review than an open commitment. These are field-specific design choices, not a universal expiration period.
When sources conflict, show the conflict and its owner. Example: the CRM says a security concern is resolved, while the latest buyer message requests a further explanation. Present both with dates and ask the seller to resolve the status. Do not discard the buyer’s message merely because a structured field looks authoritative.
Use company announcements for what the company announced, attributed conversation passages for what a participant said, and approved product material for what your team can deliver. A hiring page can support the statement that a role was advertised. It cannot independently establish available budget, internal dissatisfaction or purchase intent.
Keep outside content as evidence only. A web page, document or transcript must not be allowed to change the agent’s access permissions, recipient or operating instructions. Use limited source access and review any proposed write or outbound action. The broader operating design is in preventing AI mistakes and reviewing its work.
What should stay with the seller?
The seller should retain responsibility for relevance, fit, the question to ask now and the commitments made in the conversation. Let the machine gather, deduplicate and summarize permitted evidence. Let the human listen, judge and revise the hypothesis when the buyer says something unexpected.
| Treatment | Suitable work in our proposed design | Human responsibility |
|---|---|---|
| Automate within approved access | Collect records, attach source links, detect booking changes and flag missing context | Define scope and handle exceptions |
| Draft for the rep | Account interpretation, likely objections, stakeholder questions and a possible next step | Check evidence and select what matters |
| Leave to the rep | Read the room, assess fit, respond to disagreement and make commitments | Own the relationship and decision |
Example: the brief suggests discussing implementation timing, but the buyer opens by saying the underlying initiative is paused. Ask about that change. The preparation has done useful work if the rep can adjust intelligently; following the prepared sequence would miss the point.
Preserve the seller’s correction so another colleague benefits next time. That shared memory can also support ramping new sales reps faster and, once the account becomes a customer, the sales-to-customer-success handoff. Carry the buyer’s actual requirements and promises across those moments.
If the brief disappeared just before the call, could your rep still explain the buyer’s problem in their own words?
What privacy and data-source rules apply to prospect research?
Prospect research needs a lawful purpose, appropriate data sources and access controls. Public professional information can still concern an identifiable person. GDPR compliance and platform permission are separate questions: satisfying one does not establish the other.
Under the GDPR, Article 5 requires minimization, accuracy, security and purpose-limited retention. Article 6(1)(f) requires necessity and balancing for legitimate interests. Article 14 covers indirectly obtained data: provide required information within a reasonable period, at most one month, or earlier at first communication or disclosure where applicable, subject to its exceptions. Article 21 requires stopping direct-marketing processing after objection. Article 28 governs processors; Chapter V addresses international transfers. Recording rules also require jurisdiction-specific review.
LinkedIn prohibits third-party scraping and unauthorized automated activity. Buying enrichment does not establish that its source is permitted.
Our recommended setup records the purpose, source, relevant data categories, notice responsibility, retention rule, objection handling and approved processors before enabling research. Ask suppliers about provenance, correction, deletion, subprocessors, hosting and model-data use. Restrict transcripts and briefs to people who need them. Exclude irrelevant private-life details and sensitive inferences.
Example: a contact objects to marketing research. Apply the restriction across research jobs and enrichment imports so the next booking does not silently recreate the profile. Keep only the minimal suppression information appropriate to honoring the objection. For recordings, agree on notification and consent handling for the relevant countries and offer a non-recorded conversation where required. A recorder’s settings are not a legal conclusion.
What evidence supports time savings, and how should you measure results?
Measure automated meeting prep by useful preparation delivered, total human effort and factual reliability before attributing changes in sales outcomes to it. A faster generated brief is not necessarily a faster preparation process once review and correction are included. The available evidence supports evaluating the workflow locally rather than promising a universal win-rate increase.
Salesforce’s 2026 State of Sales announcement reports that sellers expect fully implemented agents to reduce prospect-research time by 34%. Its methodology is a double-anonymous survey of 4,050 sales professionals conducted in August and September 2025. That figure measures an expectation, not observed time saved by an automated pre-call workflow.
The strongest counter-evidence against delegating the whole interpretation is Dell’Acqua and colleagues’ randomized experiment involving 758 knowledge workers. For a managerial task selected to be outside GPT-4’s capability frontier, the control group was correct about 84.5% of the time, versus 60% and 70.6% in the two AI conditions, an average decline of approximately 19 percentage points. The task combined quantitative material and interview evidence. It was not a sales-preparation trial or a comparison of current meeting-prep products.
Our inference is to automate evidence collection and draft interpretation while keeping consequential judgment and conflicting evidence visible to the seller. That boundary is our recommendation, not the authors’ finding about sales. The practical counterargument is equally strong: native preparation may already solve the job. Start there before adding a custom workflow.
HBS Working Knowledge describes an analysis of validation attempts by 72 BCG consultants in which AI’s persuasive responses could lead professionals to accept incorrect recommendations. This does not measure an error rate for sales briefs. Our recommendation is to retain human approval while checking consequential claims against original evidence outside the model conversation. Approval should follow that independent check; reassurance from the model is insufficient.
Use this example evaluation scorecard:
| Measure | Definition |
|---|---|
| Timely coverage | Eligible meetings with a useful brief available before the agreed cutoff, divided by eligible meetings |
| Correct identity | Reviewed briefs matched to the correct account, contact and opportunity |
| Supported facts | Audited factual statements supported by the linked evidence, with correct entity and date |
| Unsupported intent | Instances where an inferred need, role or deadline appears as confirmed |
| Total preparation effort | Reading, source checking, correction and manual preparation time, plus allocated maintenance |
| Operational failures | Missing briefs, duplicate deliveries, stale attendees and unresolved access failures |
| Conversation outcome | Agreed next step and subsequent qualified-opportunity progression, reported separately from attendance |
Log whether the rep actually used the brief. Compare similar meeting types, sources and stages within reps; record changes in lead quality and sales process. Where practical, assign eligible meetings prospectively to existing preparation or the new workflow. Avoid withholding information needed to meet a commitment. Account for repeated meetings and knowledge carrying over between them.
Where account knowledge carries across repeated meetings, assign whole accounts to the comparison groups so their meetings stay in the same condition. Specify the primary measure and evaluation period before starting. Report uncertainty so a small observed difference is not presented as proof of an improvement.
Example interpretation: shorter preparation with more unsupported statements is a failed improvement. Better progression after a stronger inbound campaign cannot be credited entirely to briefing. Keep counts, denominators and uncertainty visible, particularly before enough completed opportunities exist to judge sales outcomes.
Which question did the brief help your seller ask that they would otherwise have missed?
Where Bombos fits
Good meeting prep depends on continuity: what a colleague promised last week, why the buyer booked, what is still open. Bombos keeps that in one shared memory under the AI your team already uses, so the brief your AI prepares includes what another seller learned on the previous call. Rules learned from how your team sells, such as which offer applies to which segment, are applied before anyone asks. Bombos also reads what already sits in mail, call notes and documents, and every fact in the brief shows its source. CRM changes and outgoing messages stay with the meeting owner. Book your half hour to map what your AI forgets before a call.
What could this page not establish?
This page cannot establish a universal preparation-time saving, a causal increase in win rate, or error rates for the named products on your accounts. Product documentation establishes described functionality and prerequisites, not a controlled comparison. The knowledge-worker experiment establishes a task-specific risk with GPT-4, not the performance of current sales tools. Account coverage, entitlements, source provenance and recording obligations need confirmation for your environment. The workflow and measurement rules here are proposed operating choices.
Which sources support this page?
Product documentation describes capabilities and access conditions. Vendor research is identified as such; the experimental study concerns knowledge work rather than sales preparation. Living documentation has no inferred publication date.
- Calendly Developer Documentation: Automate Your Process. Living documentation; booking, cancellation and rescheduling events.
- Chili Piper, Chili Hub: Custom Webhook Configuration. Updated July 13, 2026; event types and meeting identifiers.
- HubSpot Knowledge Base: Use meetings in the sales workspace. Updated July 31, 2026; standard preparation, calendar and subscription scope.
- HubSpot Knowledge Base: Prepare, review, and follow up on meetings in the sales workspace. Updated August 20, 2026; beta preparation, contact requirements and permissions.
- Salesforce Help: Turn On Sales Account Research. Living documentation; setup, editions and permissions.
- Gong Help Center: AI Briefer. Updated February 24, 2026; source types, brief templates and access.
- Clay University: Account Research Agents. Living documentation; open beta, Audiences and account memory.
- Apollo Knowledge Base: Research and Prepare for Upcoming Meetings. Updated October 14, 2025; templates, manual generation and email links.
- LinkedIn Sales Navigator Help: Account IQ in Sales Navigator. Living documentation; subscriptions, sources and coverage.
- Fireflies Knowledge Base: Meeting Prep: Get Briefed Before Every Meeting. October 2026 update; generation timing, email scope and eligibility.
- Fathom Help Center: Public API. Edited November 12, 2025; meeting evidence and user-level access.
- Granola, Jo Barrow: How to get the best from Granola. Displays August 3; calendar, Briefs and Gmail context.
- Otter: Meeting workflow. Living vendor product description; Work Mode and pre-meeting briefs.
- n8n Docs: Google Calendar Trigger. Living documentation; supported calendar events.
- Zapier: Google Calendar integrations. Living product documentation; event timing and polling behavior.
- MEDDICC: MEDDPICC sales methodology. Living framework description; qualification categories.
- European Union, EUR-Lex: General Data Protection Regulation, Regulation 2016/679. April 27, 2016; Articles 5, 6, 14, 21 and 28, and Chapter V.
- LinkedIn Help: Prohibited software and extensions. Living policy; scraping and unauthorized automation.
- Salesforce Research: State of Sales report announcement for 2026. February 3, 2026; seller expectations and survey methodology.
- Dell’Acqua and colleagues, Organization Science: Navigating the Jagged Technological Frontier. Published March 11, 2026; randomized knowledge-worker experiment, particularly Section 4.2.
- Harvard Business School Working Knowledge, Michael Blanding: AI’s ‘Persuasion Bombing’ Can Make Experts Question Themselves. October 29, 2026; analysis of validation attempts by 72 BCG consultants and the limits of human oversight.
