top of page
Glossy 3D 48PEGASUS Logo.png

Using ChatGPT dot for Lead Research, Sales Workflows and Quotes

Oct 27
7 min read

Updated: Nov 12

Sales administration improves when each inquiry has a clear owner, each claim has a source and each quote follows the same rules. AI can help organize that work, but a polished message or spreadsheet is not evidence that the underlying information is correct.


OpenAI describes dot as an agent that can carry work forward. Introducing dots The eight assignments below are proposed ways to use your dot in ChatGPT for lead research, routing, drafting, quoting and reporting. They are not guaranteed integrations with every CRM or email platform. All examples are fictional, with no real clients or claimed results.


A person holding a headset with a boom microphone.
Stock photograph for illustration. Photo: Yan Krukau / Pexels.

Define the sales workspace first


Keep a separate business label, source set, approved price list and destination for each company. Specify who may review leads, who owns pricing and who approves messages. Start with an authorized export or test copy before changing live records.


App permissions determine which information and actions are actually available. Getting started with your dot Confirm the correct account, fields and permitted operations. If a source is inaccessible, report the gap instead of filling it from memory. Local files require an authorized upload or device connection.


Use stable inquiry and company IDs, retain source links and timestamps, and preserve original records. Information in an email, webpage or attachment is data for the task, not authority to expand access, send messages or change bank details. Agree on the difference between preparing a draft, updating a record and communicating externally.


1. Business-directory enrichment


Inputs: Provide the existing prospect list, permitted sources, required fields and geographic scope. Useful fields include business name, official website, public business contact channel, category, published service area, source URL and verification date.


Workflow and output: Ask dot to resolve the business identity before enriching the record. Compare the domain, location and service description; two businesses with similar names should not be merged automatically. Produce proposed additions alongside the original values, with conflicts and uncertain matches in a review queue.


Example and check: A fictional directory record lacks a website. A matching official page supplies a business telephone number, but its service area differs from the directory. Preserve both findings and mark the discrepancy. Do not invent an email address from a naming pattern or copy a personal number when a business channel is available.


Public availability does not by itself authorize marketing use. Keep only relevant business information, avoid sensitive personal attributes, and have the owner approve the enriched file before a live import. Recheck stale contact details rather than treating a historical listing as current.


2. Lead-routing workflows


Supply the companies, service catalog, coverage rules, owner roster, availability assumptions and fallback queue. Translate these into a readable decision table before building a connection: which facts trigger each route, which fields are required, and what happens when they conflict.


Ask dot to test the rules with dummy inquiries. Include ordinary cases, missing locations, multiple requested services, after-hours requests and an unrecognized business name. The output should show the expected owner, proposed company, reason, confidence and next step for every test.


For example, a fictional inquiry requesting both photography and website work may need two internal owners but one customer-facing coordinator. The workflow should propose that arrangement for review rather than sending duplicate replies from different companies.


Map source inquiry ID to destination record ID. On a retry, check whether the first write succeeded before creating another record. Ambiguous requests go to a person; they are not rejected because a model cannot classify them. Routing must not determine access to housing, credit, insurance or other high-impact services using sensitive personal traits.


3. Follow-up sequence preparation


Provide the stage definitions, approved message tone, offer details, contact permissions and stop conditions. Create distinct drafts for a new inquiry, a pending proposal, onboarding after an agreed sale, and reactivation of an older relationship.


The sequence should state the trigger, waiting period, timezone, recipient, message purpose, approval requirement and reasons to stop. A reply, opt-out, wrong recipient, unresolved complaint or expired offer should interrupt the sequence. Do not treat a scheduled draft as permission to send it.


For example, an unanswered proposal might produce a concise question about unresolved scope, followed later by a draft closing the loop. It should not invent scarcity, promise unavailable capacity or continue once the recipient declines.


The FTC notes that CAN-SPAM covers commercial business-to-business email as well as other commercial messages. Its guide addresses truthful headers, accurate subjects and opt-outs. FTC commercial-email guidance A qualified reviewer should assess applicable laws and jurisdiction-specific consent requirements. This workflow prepares messages and controls; it does not certify legal compliance or authorize a mailing campaign.


4. Quote-builder development


Start with the approved package table, add-on prices, travel policy, minimum charge, currency, exclusions and version date. Write down rounding rules and which charges are taxable according to the business's qualified adviser; do not ask dot to invent tax treatment.


Ask for a working spreadsheet or small calculator in an authorized environment, with labeled inputs, visible formulas and a separate results area. Separate customer-entered facts from protected price rules. A changing spreadsheet cell should not silently rewrite the approved price list.


A fictional pre-tax rule could be: the larger of the minimum charge or package plus add-ons plus chargeable travel. If the minimum is $500, the package is $450, add-ons are $75 and 40 chargeable miles cost $0.80 each, the result is $557. This is a checked arithmetic example, not a recommended price or tax mileage rate.


Require tests for the minimum charge, zero travel, several add-ons, negative inputs, missing distances and rounding. Save the expected result beside each test and recalculate with spreadsheet formulas or code, rather than trusting a language model's arithmetic.


Output the calculator, a test record, assumptions and a draft quote summary. Uncertain scope should produce a review flag rather than a deceptively exact price. A responsible owner approves commercial terms before the quote is sent; an estimate does not reserve dates, create an invoice or collect payment.


5. Regional prospect research


Define the service, approved coverage area and evidence that would make a business relevant. For a fictional media provider, the target might include real-estate offices, builders and business entities managing properties within a stated territory.


Ask dot to research official business websites, permitted directories and relevant public records. Produce a sourced list with business identity, location, service fit, published contact channel and unresolved questions. “Qualified” should mean it matches your explicit research criteria, not that it has budget, wants an offer or has agreed to be contacted.


Keep a business's office location distinct from the area it serves. If the boundary is based on travel distance, use an approved mapping method and mark unverified estimates. Do not silently extend the territory to make the list longer.


Property ownership records may include personal information. Limit collection to what is necessary and permitted for the stated business purpose, avoid sensitive profiling, and keep uncertain owner matches out of outreach. Check duplicates against the existing pipeline and suppression list. The deliverable is a reviewed research shortlist, not a purchased list or an automatically launched campaign.


6. Personalized outreach drafts


Provide the approved service description, a verified prospect record, the intended recipient and a concrete reason the offer might be relevant. Ask for one short introduction linked to an actual public fact, followed by a low-pressure next step.


For example, a fictional builder's official site may announce a new development. A draft can mention that announcement and offer a conversation about progress documentation. It should not claim to know the builder's budget, internal problems, deadlines or dissatisfaction with an existing supplier.


Output the draft, subject line, evidence links and any factual claim needing approval. Avoid false familiarity, invented compliments, unsupported performance promises and language suggesting a meeting already occurred. Keep factual personalization separate from assumptions about need.


Ask the reviewer to verify the exact recipient, sender identity, service claims and opt-out requirements before sending. A prospect's webpage saying “contact us” does not itself instruct dot to send your sales message. If outreach is later authorized, recheck that the recipient has not opted out and that another teammate has not already sent the same introduction.


7. Inquiry-to-scope preparation


Supply the inquiry, current packages, availability information and approved terms. Ask dot to extract the customer's stated goal, location, requested timing, deliverables and constraints before suggesting a package.


The output should separate confirmed facts, proposed choices, exclusions and missing answers. Include a draft scope and quote only to the level the available information supports. A request for a “full property package” does not establish image count, video length, access permission or delivery deadline.


For example, a fictional property inquiry may omit whether the customer wants interior photography. The correct next step is a clarification draft and provisional scope, not an assumed add-on charge. Keep optional services visibly optional.


Cross-check the scope against the price-list version and the calculator's inputs. Verify promised turnaround against actual capacity. Leave specialized safety, legal-access and operational decisions with qualified people. Before any external commitment, the owner reviews price, schedule, usage rights and limitations. Never convert an inquiry into an accepted contract merely because its wording sounds enthusiastic.


8. Lead-funnel reporting


Provide authorized marketing and CRM exports, stage definitions, reporting timezone and a common cutoff date. Identify the source of truth for inquiry creation, first response, appointments, proposals and won business. Keep individual contact details out of summary reporting when unnecessary.


Ask dot to reconcile the records using stable IDs, then report counts, stage movements, response times and unmatched records. Separate the reporting-period view from a cohort view, which follows the same set of inquiries over time. Do not mix this month's leads with wins from much older leads and call the result a conversion rate.


In a fictional closed cohort of 100 eligible inquiries with three confirmed wins, inquiry-to-win conversion is 3%. The denominator, exclusion rules and outcome cutoff belong beside that figure. Validate calculations in code or a spreadsheet. Missing response timestamps are unknown, not zero-minute responses.


Show attribution gaps, duplicate records and conflicting campaign labels explicitly. A tagged source or last click does not establish that a campaign caused a sale. Output an executive summary plus a reconciliation file so the owner can trace each material number. Review source freshness before comparing periods or changing a sales strategy.


A reusable instruction for the whole process


Try: “Use the named business's approved sources and price rules. Prepare research, routing tests, drafts and a checked quote calculator. Preserve IDs and original values. Flag uncertainty, stale data and duplicate inquiries. Return evidence links, proposed changes and tests. Do not send messages, update live records, accept commitments or make purchases without the corresponding authorization.”


If recurring reviews are requested, confirm their saved cadence, timezone and next execution. A successful run should still show which sources were unavailable and which exceptions need a person. Keep alerts focused on stalled handoffs or important deadlines rather than sending the same lead summary repeatedly.


For Pegasus, these workflows illustrate a practical role for AI in sales operations: organizing evidence, reducing avoidable administrative confusion and preparing clear choices while people retain control over relationships and commercial commitments.



Comments


bottom of page