compute_pricing
ChatGPTCompute prices for given service selections and property info WITHOUT creating a bid, contact, or property. Returns prices for all enabled packages per sub-service. Use list_catalog_services first to get valid sub-service IDs, factor IDs, and option IDs. If what the user asked for doesn't clearly match one catalog service or sub-service by name (several could fit, none fit, or only an industry term matches), ask the user using the catalog's own names. Don't guess. Useful for price checks, what-if scenarios, and previewing costs before creating a bid.
confirm_appointment
ChatGPTBook a previously offered appointment slot for a bid, from a slot id returned by request_scheduling_slots. The slot may have been taken by another customer between the two calls — if so, the result lists the times available now; offer those instead. This books a first appointment only: if the customer wants to move or cancel an appointment they already have, do not attempt it — tell them to contact the company directly. A user may refer to a bid as a "quote" — treat both terms as the same thing.
confirm_signature
ChatGPTRecord a customer signature on a bid proposal, confirming they agreed to its terms. Scheduling cannot be requested for a bid until its signature is recorded. When the signature is recorded AND the company has scheduling enabled, this tool goes on to request that bid's appointment slots itself and returns them with the confirmation — do not call request_scheduling_slots again for the times it already lists. That carries request_scheduling_slots' own side effect: when the calendar has no availability, the bid is moved to the company's configured "could not schedule" status. Offer the returned times to the customer and book one with confirm_appointment. A user may refer to a bid as a "quote" — treat both terms as the same thing.
create_bid
ChatGPTCreate a bid by composing a contact, property, and service selections. A user may refer to this as a "quote" — treat "quote" and "bid" as the same thing. Collect all info from the user first — do not fabricate values. Don't invent IDs: service, area, house size and lead source IDs come from the list_* tools. If what the user asked for doesn't clearly match one catalog service or sub-service by name (several could fit, none fit, or only an industry term matches), ask the user using the catalog's own names. Don't guess. Before creating a new contact, ask the user whether the customer agreed to texts and to emails, and show both answers in the summary. For an existing contact (contact_id, or one found with find_contact_by_phone) don't ask; show its saved opt-ins. By default the bid is saved and priced without contacting the customer: show the user the prices this call returns (the same package lines get_proposal shows), then call start_follow_up once they approve. Pass start_follow_up: true only when the user wants the company's Open follow-up, whose first message can reach the customer right away, started now.
create_contact
ChatGPTRegister a caller as a new contact. A contact group is created automatically — you do NOT need one first. Deduplicates by email then phone: if a contact with the same email or phone already exists it is returned unchanged (wasMatched=true) rather than duplicated. Use this after find_contact_by_phone returns no match. Only affects the authenticated company. Collect details from the caller — do not fabricate values. It can't record text or email consent: for a new customer who gave consent, create them through create_bid's contact instead.
find_contact
ChatGPTLook up contacts by name, email or property address when the phone number is not known. Each given field must match: name matches every word against first or last name, email and address are partial matches (address against the street lines or ZIP of any of the contact's properties). Only contacts belonging to the authenticated company are ever returned, at most limit per page. Present the matches AS-IS, without editorial asides.
find_contact_by_phone
ChatGPTLook up a caller (contact) by their inbound phone number. Formatting is ignored — any format of the same number finds the same contact. Returns matching contacts with their name, phone numbers, and email. A number that cannot form a full phone (fewer than 10 digits) is rejected; a number with no match returns an empty result, not an error. Only contacts belonging to the authenticated company are ever returned. Present the matches AS-IS — list the contacts exactly as returned, without editorial asides, guesses about whether they look like test records, or unsolicited next-step suggestions.
get_bid
ChatGPTRead one bid (quote) by its id, including its follow-up status, closing price, and the contact id and property id it belongs to (not the contact/property objects themselves). Use it after list_bids to answer questions about a specific quote. The closing price is the whole-dollar total stored on the bid, so a selection priced in cents reads rounded here; call get_proposal for the exact price to quote a customer. A bid belonging to another company is reported as not found.
get_catalog
ChatGPTGet the whole catalog in one call: every quotable service with its sub-services, factors and options, plus service areas, lead sources and house sizes — the same lists list_catalog_services, list_service_areas, list_lead_sources and list_house_sizes return. A LARGE_DATA question with no options can't be answered here: it has more than 1,000 options (such as a vehicle list) or none configured. Leave it unanswered and tell the user the quoted price doesn't include it, so the company still has to price that part. Pass services with the service names or IDs being quoted to get only those services; house sizes, service areas and lead sources are always returned, and when services is set, meta.availableServices lists every quotable service name (use it to offer add-on services) and meta.unmatchedServices lists requested names that matched nothing. When asked to list the catalog, present the entries AS-IS — the services and sub-services exactly as returned, without editorial asides, groupings the data does not carry, or judgements about which entries are usable. A user may refer to a bid as a "quote" — treat both terms as the same thing.
get_follow_up_sequence
ChatGPTRead one follow-up sequence with every step it runs, in the order they fire, and each step's current wording verbatim. This is the whole context a rewrite needs: it also reports the exact merge fields this sequence can resolve. Call this before update_follow_up_step, and use only the merge fields it lists.
get_proposal
ChatGPTGet the proposal for an existing bid — returns the structured pricing breakdown with resolved service and package names, per-line pricing, bundles, requested services, contact and property details, and signature status. Present the returned proposal AS-IS: relay the contact, property, services, packages, bundles, and status exactly as given, without editorial asides, guesses about the data (e.g. whether records look like tests), or unsolicited analysis. The closing price is the whole-dollar figure stored on the bid; when a package's exact price sums to a fractional dollar amount that figure cannot hold, the summary reconciles the two and quotes the exact total — always relay that reconciled price rather than the rounded one when the two are shown separately. A bundle marked [selected] is the one the customer has chosen on this bid; at most one is, and none being marked means the customer has not chosen a bundle. A tier marked [recommended] is the one the bid's selected bundle puts forward; it is the bundle default, not a "most popular" pick the company chose, so describe it as what the bundle includes rather than as the company's advice. A user may refer to a bid as a "quote" — treat both terms as the same thing.
get_service_pricing
ChatGPTGet one service's full pricing setup: what each part means in a quote. Each sub-service has up to five packages (tiers the customer chooses between, e.g. Basic / Premium); a package only appears in quotes when enabled, and its hourly rate converts hours/minutes prices into dollars. A "price per…" rule (pricePer) relabels a package price and multiplies or divides it by a quantity (e.g. "per window"); null means no rule. House sizes are the size brackets a property is placed in; each has a type, and only residential sizes are home sizes (commercial and other types describe businesses or other properties, so they are not used for homes). label is how the app shows a size: its name, or for commercial sizes (which have no name) their square-footage range; use it when naming a size to the user. A square-footage question (pricing "band") sets a base price per house size and package; an access-charge question (pricing "access") adds a charge per house size, package and floor (e.g. "2story", "basement"). Write band and access prices with set_house_size_prices. A measurements question (pricing "rate") charges a price per sq ft per package: its cells are that rate, and mirroredCells holds the mirrored side; write them with set_measurement_rate. Other questions price by their answers (pricing "options", write with set_option_price): each answer's cells are its price effect per package; a how-many-story answer's cells are keyed by floor ("basement", "1_story" … "4_story") then package. A cell is { value, priceType, priceUnit }: upcharge adds, discount subtracts, bid-in-person means the job must be quoted on site, ignore has no effect. Cells are keyed by house size id, then package number (then floor for access charges); a missing cell is $0. unpricedHouseSizeIds lists house sizes with no price on that question, which quote $0; orphanedHouseSizeIds lists prices kept for house sizes that no longer exist. suffixUnreadable on a sub-service means its stored package rules are damaged, so they cannot be shown; damaged question or option prices fail the whole read. exactSizesBypass names the property types (residential / commercial) priced by separate exact-size rates when the property's actual square footage is known, so the bands do not apply to those properties.
list_bids
ChatGPTList the bids (quotes) belonging to the authenticated company, newest first. A user may refer to a bid as a "quote" — treat them as the same thing. Use this to answer follow-up questions about work already quoted, and narrow it with contact_id (from find_contact_by_phone) or status. Only the authenticated company's bids are ever returned. Results are paginated — the reply says when another page is available. Each closing price shown is the whole-dollar total stored on the bid, so a selection priced in cents reads rounded here; call get_proposal for the exact price to quote a customer.
list_catalog_services
ChatGPTList every service in the catalog with its sub-services, factors and options. A service with no enabled sub-service is left out: it cannot be quoted. A LARGE_DATA question with no options can't be answered here: it has more than 1,000 options (such as a vehicle list) or none configured. Leave it unanswered and tell the user the quoted price doesn't include it, so the company still has to price that part. To show the full catalog, call get_catalog. If what the user asked for doesn't clearly match one catalog service or sub-service by name (several could fit, none fit, or only an industry term matches), ask the user using the catalog's own names. Don't guess.
list_follow_up_sequences
ChatGPTList the follow-up sequences a company runs — the campaigns that send messages automatically after a quote or a job. START HERE when the user wants to review or rewrite their follow-ups: pick the sequence they mean, then call get_follow_up_sequence for its steps. Each entry reports which merge fields it can resolve. No sequence listed here is locked against a rewrite of its steps, so never skip one or divert a rewrite to a copy of it. These are company-wide templates, not tied to any one bid or customer.
list_follow_up_steps
ChatGPTList the follow-up steps a company sends after a quote or job, grouped by the sequence they belong to. Call this BEFORE rewriting anything: it returns each step's current wording verbatim and, per sequence, the exact merge fields that sequence can resolve. These are company-wide templates, NOT tied to one bid. Editing a step also changes the messages already scheduled for customers currently in that sequence, because a pending message takes its wording from the step when it sends. The only exception is a customer whose message was edited by hand in the app, which keeps that hand-edited wording. To work on one sequence, prefer get_follow_up_sequence, which returns that sequence and its steps in run order in a single call; use this tool to sweep steps across sequences. Note a sequence with no steps cannot appear here at all — list_follow_up_sequences is the complete list. The copy of a step is never locked by the settings of the sequence that owns it: never report a step as uneditable, skip it, or retarget the rewrite to a differently named copy of the sequence. The one step update_follow_up_step refuses is one listed here with no owning sequence at all, because its merge fields cannot be resolved.
list_house_sizes
ChatGPTList every house size the company prices by. To show the full catalog, call get_catalog.
list_lead_sources
ChatGPTList every lead source a contact can be attributed to. To show the full catalog, call get_catalog.
list_service_areas
ChatGPTList every service area the company serves. To show the full catalog, call get_catalog. When unsure which service area the user means, confirm with the user instead of choosing.
manage_house_sizes
ChatGPTCreate, rename, delete or reorder house sizes. House sizes are the size brackets a property is placed in when quoted; square-footage and access-charge questions price by house size. Each has a type, and only residential sizes are home sizes (commercial sizes describe businesses, so never offer them as home sizes). create needs name and type (residential or commercial), optionally max_sq_ft (the largest square footage in the bracket); a new size quotes $0 until its prices are set. rename needs house_size_id and name. delete needs house_size_id; prices stored for it stay but no longer apply, and properties placed in it quote bid in person on services priced by house size. reorder needs order: every house size id of one type (residential or commercial), in the new display order; the app orders each type separately. Commercial sizes have no name: the app shows them by square-footage range, so they cannot be renamed. Previews unless apply is true.
request_scheduling_slots
ChatGPTRequest bookable appointment slots for a signed bid from the company's scheduling calendar. When no slots are available, the bid is moved to the company's configured "could not schedule" status as a side effect. A user may refer to a bid as a "quote" — treat both terms as the same thing.
select_proposal_bundle
ChatGPTSelect the bundle a customer chose on a bid, or clear the current selection by passing null. The bundle must be one of THAT bid's bundles — pass an id from this bid's get_proposal response, not from another bid. Selecting a bundle ALSO sets the package on every SUB-SERVICE and CUSTOM ITEM the bundle carries: an existing package the customer already chose is kept when the bundle allows switching to it, otherwise the bundle's own default package is applied. Sub-services and custom items outside the bundle are left alone. Clearing removes the bundle discount and DESELECTS the packages on that bundle's sub-services and custom items — including any the customer had chosen independently before the bundle was selected, so confirm before clearing. Returns the updated proposal, so the price in the response is the price that persisted. Bundles must be ENABLED and configured on the bid beforehand (done in the ResponsiBid app, not through this API): a bid created via create_bid has no bundles, so both selecting and clearing return 409 BUNDLES_DISABLED. Only act on a bid whose get_proposal already lists bundles. A user may refer to a bid as a "quote" — treat both terms as the same thing. When the customer refers to a choice positionally or vaguely ("the middle one", "the first one", "the latter", "that one", "the cheaper one"), do NOT call this tool yet: state the exact name you resolved it to, with its price, and ask the customer to confirm. Call only after the customer explicitly confirms that choice, or names it directly.
select_proposal_package
ChatGPTRecord the package a customer chose on a bid, per sub-service. Identify each package by its NAME exactly as get_proposal shows it for that sub-service (names are set by the company — e.g. "Full Clean Up", "Half Clean" — never assume defaults). Pass null to unselect. Each sub-service is identified by its moduleSubServiceId from get_proposal — NOT the id listed beside it, which this tool does not match. This covers MODULE services only: a CUSTOM_ITEM has no moduleSubServiceId, and its package cannot be selected here — say so rather than guessing an id for it. If a bundle is selected on the bid, that bundle's sub-services are constrained: you may switch to another package only when the bundle allows switching AND that package is offered by the bundle, and you may never unselect a bundle sub-service — such requests are refused (409) and NOTHING in the batch is written. Sub-services outside the bundle are unrestricted. The tool matches on the exact name only. Returns the updated proposal. A user may refer to a bid as a "quote". When the customer refers to a choice positionally or vaguely ("the middle one", "the first one", "the latter", "that one", "the cheaper one"), do NOT call this tool yet: state the exact name you resolved it to, with its price, and ask the customer to confirm. Call only after the customer explicitly confirms that choice, or names it directly.
set_bundle_auto_add
ChatGPTList bundle sets, or turn a bundle set's auto-add on or off. A bundle set groups bundles (packaged combinations of services offered together). With auto-add on, ResponsiBid automatically adds the set's bundles to a quote whose requested services fit them, recommending the bundle to the customer; with it off, bundles are only added by hand. Without bundle_set_id this lists the sets and changes nothing. With bundle_set_id, enabled is required. Previews unless apply is true.
set_house_size_prices
ChatGPTSet the prices a square-footage or access-charge question charges per house size and package. House sizes are the size brackets a property is placed in; each has a type, and only residential sizes are home sizes (commercial sizes describe businesses). A square-footage question sets the base price for each house size and package (its "band"); an access-charge question adds a charge per house size, package and floor segment ("basement", or "1story", "2story", …), so access rows need a segment and band rows must not have one. Each row edits one cell; fields left out keep their stored value, and cells not named are left unchanged. Previews unless apply is true.
set_measurement_rate
ChatGPTSet the rate a measurement question charges per package. The rate is the price per measured sq ft for each package (area x rate, raised to the question's minimum when minimums are on). A measurements-mirrored question has a second rate per package: give those rows side 'mirrored'. Each row edits one package's rate; fields left out keep their stored value, and packages not named are left unchanged. Answers of measurement questions carry no price. For a question priced per answer (pricing "options" in get_service_pricing), use set_option_price instead. Previews unless apply is true.
set_option_price
ChatGPTSet an answer's (option's) price effect per package, and optionally its minimum. When a customer picks this answer, each package adds (upcharge) or subtracts (discount) the amount, needs an on-site quote (bid-in-person), or is unaffected (ignore). minimum (how-many questions only) is the least quantity the customer is charged for on this answer: a smaller count is raised to it when the sub-service has minimum quantities turned on in the app; "0" means no minimum. Each row edits one package; fields left out keep their stored value, and packages not named are left unchanged. Answers of square-footage, access-charge and measurement questions carry no price: use set_house_size_prices or set_measurement_rate. Previews unless apply is true.
set_packages
ChatGPTChange a sub-service's packages. Packages are the up-to-five tiers a customer chooses between in a quote (e.g. Basic / Premium); only enabled packages appear in quotes. A package's hourly rate converts prices set in hours or minutes into dollars. A "price per…" rule (price_per) relabels the package price and multiplies or divides it by a quantity (e.g. "per window"); null removes the rule. Fields left out keep their stored value, and packages not named are left unchanged. Only price_per is guarded against concurrent edits; name, enabled and hourly_rate overwrite whatever is stored. Previews unless apply is true.
start_follow_up
ChatGPTStart the follow-up of a bid saved without starting it (create_bid with start_follow_up left off), moving it into the company's Open status. Call it only after the user has seen and approved the prices. It starts the follow-up once: a bid that already has a status is refused, so a repeated call never sends twice.
update_follow_up_step
ChatGPTUpdate one follow-up step of a company-wide sequence. Only the fields you pass are changed. Use this to rewrite the wording of the messages a company sends after a quote or job. IMPORTANT: subject and body are rendered by a merge engine that DELETES anything in curly braces it cannot resolve — an invented merge field, a field belonging to another sequence type, or ordinary prose written in braces all become blank text in the real customer message, with no error. So call list_follow_up_steps first and use ONLY the merge fields it reports as in scope for that sequence, and never write literal curly braces in prose. Bad merge fields are refused here before anything is written. Each call updates ONE step, so a multi-step rewrite is several calls and a failure partway leaves earlier steps already changed — report what landed if one fails. Rewrite the sequence the user named — do not silently retarget a differently named copy of it.