Alta

Create a custom prospect field on this account so add_prospects can carry values for it, and return the field key to pass those values under. The account's existing fields are the ones list_prospect_fields returns; a field whose name is already taken is returned as it stands rather than duplicated. The key is derived from the label and is not guessable from it.

create_prospect_field

Remove a manually added identifier from the communication opt-out (do-not-contact) list. Prospects paused when it was added stay paused.

delete_communication_opt_out

Answer a free-form research question about a person using an AI agent with web search. First resolves the person (provide a LinkedIn profile URL for best results; alternatively a known email, or firstName + lastName with organizationName), then runs the prompt against their enriched profile and live web research. Pass the question in prompt. Optionally pass outputFields (key/label/type) to get structured values back; otherwise a single free-text "answer" field is returned, alongside reasoning, confidence, and sources. The answer is returned only and is not saved. Consumes the account's enrichment allowance, which cannot be refunded.

enrich_ai_fields

Enriches records in an audience, filling in emails, phones, profile data or company data from third-party enrichment providers, and returns how many enrichment jobs were queued. The filters select which records are enriched, and count caps how many of them the run takes. It consumes the account's enrichment allowance, which estimate_audience_enrichment reports on beforehand for the same arguments — with the one difference that a run must be sized, so a quote taken without a count still needs one here. Enrichment is all-or-nothing: a request larger than what the account has left enriches nobody and fails with the shortfall. Jobs run in the background, so enriched values are not available immediately. A queued count of 0 means no records matched the request, not that enrichment failed.

enrich_audience

Enrich a person and return their full profile: contact details (email, phone), title, headline, summary, location, company, employment history, education, skills, and buying signals. Provide a LinkedIn profile URL for the best results; alternatively a known email, or firstName + lastName together with organizationName (organizationDomain narrows that match further). The profile is returned only and is not saved to any prospect. Consumes the account's enrichment allowance, which cannot be refunded, so run it when the user asks to enrich or look up a person.

enrich_prospect

Reports what an audience enrichment would consume, without enriching anything or queueing any work. Returns how many records match, what the run would draw against the account, whether it fits, and a one-line summary in the units this account is billed in — credits on credit-based accounts, plan enrichments otherwise. The filters select which records the run would cover, and count caps how many of them it takes; naming no count and no ids prices the whole audience, so a quote never needs a size, while the run that follows it does. Enrichment is all-or-nothing, so a run that does not fit enriches nobody.

estimate_audience_enrichment

Return an audience as CSV text so the app can hand it to the host as a file download. Takes the same filters as list_audience_people, and can also name records outright with audienceListItemIds, so the export matches what is on screen; with neither it exports the whole audience. App-only — the CSV is never routed through the conversation.

export_audience

Get an audience by its ID — name, entity type, how many people it holds, which sources fed it, which campaigns use it, and the countries, industries, companies, signals and sources its records carry, which are the values list_audience_people filters on. Each source carries its daily sync (status, batchSize, nextSyncAt), or null when it was a one-time import — pause or resume it with update_audience sourceSync. Call this when the user wants to see, review, or export an audience; the rendered card carries the export action.

get_audience

Report how far the enrichment of an audience has got, as one progress column per enrichment goal and per AI field: how many cells are waiting, running, done, came back with no match, failed, or have not been run at all. enrich_audience only reports how many jobs it queued, so this is how to answer whether a run has finished. Counts are per cell — a record carries one result per goal — so the columns must be read side by side rather than added up, and they describe the audience's state rather than one run. This is a snapshot, not a subscription: report it and let the user ask again rather than calling it in a loop.

get_audience_enrichment_progress

Get the fields of a CRM object and the mapping Alta picks automatically from them (Alta field key -> CRM field key): field-name matching first, then AI suggestions for the fields still unmapped, as in the audience builder. Use it only when the user wants a specific CRM field to fill an Alta field — pass the adjusted mapping in update_audience crmList.mapping. Without one, the automatic mapping is applied.

get_crm_fields

Get a call by its ID

get_call

Get a campaign by its ID with its full configuration: pitch, assigned reps, audience (prospect source type, audience list and each source's config, or search filter), run mode, sending schedule, prospect limitations, the Settings page fields (settings — including the reply assistant, replyAgentId), and workflow. The workflow is either agentic (agenticSettings) or step-based (the full step tree, each step with its content). The pitch, settings and an agentic workflow come back in the shape update_campaign accepts, so read them here before changing them.

get_campaign

Get a company by its ID, including name, domain, industries, location, and employee count

get_company

Get an email message by its ID — sent, received or draft

get_email_message

Get a LinkedIn message by its ID

get_linkedin_message

Get a person by their ID

get_person

Get a prospect by its ID. The result is the prospect's own ids; the detail is drawn on the card it opens — identity and buying signals alongside the outreach timeline: the campaign the run belongs to, every step the sequence sent or skipped with the message it carried, and the replies, opens and connection requests that came back. That detail is rendered on the card rather than returned as text.

get_prospect

Get a prospect's buying signals — person signals (job changes, LinkedIn posts and engagement, work anniversaries) and company signals (funding, product launches, hiring, leadership changes)

get_prospect_signals

Return what has happened with one prospect, in order: the campaign it belongs to, when it was sourced, every outreach step the sequence ran with the message it sent, and the replies, opens and connection requests that came back. Steps the sequence has not reached yet come back separately, and a step that was skipped or failed carries the reason. Talkative channels are capped at their most recent events — 10 for email, 20 for LinkedIn — and `omitted` counts how many older ones exist beyond them.

get_prospect_timeline

Get a rep (sender) by their ID

get_rep

Report which buying-signal types this account has turned on, as the list of signal types an enrichment run can actually return. An account that has never configured its signals has the default set on. App-only: it describes what the enrichment screen may offer, not anything about a particular audience.

get_signal_configurations

Get one task by its ID — the instruction to follow, its status, type, priority, the prospect it was raised for, and the result recorded once it was completed.

get_task

Return the records inside an audience — the fields Alta holds on each one, including the account's custom field values, minus the social feed. Reads 25 records at a time by default. The filters run in the database, so a filtered read returns the matching subset rather than a page of the whole audience. The reachability flags (hasEmail, hasPhone, hasLinkedin) keep a record matching any of the ones set, not all of them. The countries, industries, companies, signals and sources an audience holds are reported by get_audience.

list_audience_people

Return the records inside an audience with every field Alta holds on each one, including the social feed. Takes the same filters as list_audience_people, so the table can search and narrow in the database rather than paging the whole audience. App-only: the rows fill the audience table and are never routed through the conversation.

list_audience_records

List audiences, most recent first, with how many records each one holds and which sources fed it

list_audiences

List the CRM lists that can be imported into an audience. With no source, returns the connected CRMs that have lists — each with its people source (sourceType, the CRM objects people come from, and whether it supports a daily sync) and its companies source. With a source, returns that CRM's lists for one object type (id and name). Import the chosen list with update_audience (crmList).

list_crm_lists

List calls, sorted by most recent first

list_calls

List the messaging campaigns the user can access, sorted by most recent first

list_campaigns

List communication opt-outs (do-not-contact list), sorted by most recent first

list_communication_opt_outs

List email messages (sent, received or draft), optionally filtered by thread, campaign, prospect, rep, type, or tag, sorted by most recent first

list_email_messages

List LinkedIn messages, optionally filtered by campaign, prospect, rep, type, or tag, sorted by most recent first

list_linkedin_messages

List the account's meetings booked with prospects — start time, duration, status, the calendar it was booked on, host and guest emails, the rep, and the prospect. Filter by prospect, campaigns, reps, status, or integration; sorted by start time, latest first, unless sort says otherwise. Also returns the total number of matching meetings, so page with skip and limit to read past the first page.

list_meetings

List persons in the account

list_persons

List the prospect fields this account accepts — key, label, type, and whether the field is built in or custom. The keys returned here are the only ones add_prospects accepts; any other key is rejected.

list_prospect_fields

List prospects, optionally filtered by campaign, sorted by most recent first

list_prospects

List reps (senders), sorted by most recent first

list_reps

List the account's tasks — the human to-dos a campaign raises when its sequence reaches a manual step, plus tasks created outside a sequence. Each one carries the instruction to follow, its status, priority, and the prospect it was raised for. Filter by prospect, campaign, status, type, or priority; sorted by scheduled time, most recent first.

list_tasks

Load a detailed playbook into context before that kind of work (silent, internal). Topics: campaign-creation: pipeline order, preview/confirm steps, and which tools to call. Always invoke this first, before any other campaign step, whenever the user wants to build, draft, set up, or modify a campaign.), workflow-prompt-builder: full instructions for building Alta workflow outreach prompt templates (channel rules, {{variables}}, banned phrases, structure). weekly-report: a bespoke per-rep recap with its own prospect scope, which differs from the Performance page. Load it ONLY when the user asks for the weekly report or weekly recap by that name. A request to pull, show or check performance for any period, the past week included, goes straight to the data tools. audience-form: use when on audience builder or campaign audience step - full rules for filling the target audience form (filter logic, job titles, departments, locations, filter value lookups, audience size checks). exa-web-sets: use when the audience source is exa_people - query and criteria best practices, and how to fill the Exa web-set form with fill_exa_websearch when available. reply-to-prospect: use whenever the user wants to reply to or answer a prospect on email or LinkedIn - how to gather context, preview the reply in chat, and then send it or save it as a draft. alta-search-query: Convert natural-language audience descriptions — ICP notes, briefs, job postings, lead lists — into Alta search queries over people and companies. Use when the user wants to write or refine an Alta advanced search query, or describes leads or companies they want to find in Alta. In the app, load it before calling fill_search_query — whenever the audience source is in Advanced search, or the user wants to switch a Classic search to an advanced query. support-ticket: load FIRST, as soon as the user reports something broken or asks for support, and always before offering or calling create_support_ticket — whenever the user asks to contact support / open a ticket, or has an account-specific problem (billing, access, bugs, data) you cannot resolve — how to gather details, write the subject/description, and follow up. semantic_models: load FIRST, when the semantic model tools are available, for any question about this account's numbers, trends, breakdowns or metric definitions and any dashboard request, including creating, renaming, changing, adding to, removing from or reordering a dashboard — when to use the semantic models or fall back to the Ask Alta agent, how to compose and present the explores, and when to save them with save_dashboard_from_explores or edit a dashboard with add_dashboard_widgets, update_dashboard_widget, remove_dashboard_widgets and update_dashboard.

load_skill

Lists reps and returns the best match for the given name or email fragment. Does not persist.

match_campaign_rep

Closing step of campaign creation: returns a link to the draft campaign in the Alta campaign builder, where the user reviews it and decides whether to launch. Call once the pitch and workflow have been persisted. Read-only — it does not launch, activate, schedule, or send anything; the campaign stays a draft until a person launches it in Alta.

open_draft_campaign_in_alta