Amdahl Platform API

The user's resume payload. Must match the run's persisted pending_input_schema.

Optional operator note describing why the run was canceled.

Optional stable kebab-case dispatch key (lowercase letters, digits, single dashes). Omitted: derived from the name and uniquified automatically.

Display name.

Optional one-line blurb for the agent directory.

Operation ids to subtract from this agent tool kit at run time (all kit tools are on by default; empty list = no restriction). Runner-inline tools are refused.

The prompt the agent runs with.

New display name.

New directory blurb.

New prompt body.

Operation ids to subtract from this agent tool kit at run time (all kit tools are on by default; empty list = no restriction). Runner-inline tools are refused.

Human label. Each fired Chat is named "{name} — {date}".

The Chat input every fire starts the Master turn with.

Cron expression (e.g. "0 9 * * 1"). Validated before the row is stored.

IANA timezone the cron evaluates in. Defaults to UTC.

LEGACY single-agent pin the Master turn runs as (library slug/uuid or workspace agent id/slug). Prefer config.agents; passing both rejects.

Per-fire Chat config. on_question is always none (headless); agents picks the roster ("all", one pin, or a delegate allowlist).

Whether the routine fires. Defaults to true.

New human label.

New Chat input for future fires.

New cron expression (validated before storing).

New IANA timezone.

LEGACY single-agent pin, or null to clear back to the plain Master. Prefer config.agents; passing both rejects.

Replacement per-fire config (replaces the whole object, including the agents roster).

Enable or disable firing.

Human label. Each fired Chat is named "{name} — {date}".

Which event source fires this subscription (e.g. "calendar.event_upcoming"). Read subscription://kinds for the catalog + each kind's config fields.

The Chat input every fire starts the Master turn with; the triggering event is appended after it automatically.

Per-kind source config, validated against the kind's schema (for calendar.event_upcoming: connection_id, lead_minutes, optional calendar_id + filters).

Optional agent pin the Master turn runs as (library slug/uuid or workspace agent id/slug).

Per-fire Chat config. on_question is always none (headless); every knob defaults conservative (standard depth, no market fan-out).

Whether the subscription fires. Defaults to true.

Replacement per-kind config, validated against the subscription's existing source_kind.

New agent pin, or null to clear it (plain Master).

Catalog connector type to connect (e.g. hubspot, fathom, twitter).

Optional display label for this connection instance.

Ownership: business (workspace-wide) or personal. Defaults from the catalog.

For api_key connectors: the API key / token to store (encrypted).

For an api_key connector whose credential is a PAIR (today only Aircall, whose API is HTTP Basic): the non-secret id half, pasted alongside the token in api_key. Read connections.list_catalog to see whether a connector asks for it.

For the computer_agent connector: its deployment base URL (http/https).

For social (handle) connectors: the resolved bare account handle.

For social (handle) connectors: the provider's stable account id.

For social (handle) connectors: display name snapshot.

For social (handle) connectors: avatar image URL snapshot.

For social (handle) connectors: follower count snapshot.

For oauth connectors: console URL to redirect back to after the handshake (first-party origins only).

For an api_key CALLS connector (e.g. Fireflies): an optional atomic pre-sync call filter applied in the SAME request, so the source connects AND its filter decision lands together (an enabled filter syncs filtered; { enabled: false } is the "ingest all" opt-out). Omit to leave the calls source paused until a filter is set later. Own domains are stamped server-side. Requires workspace admin.

For a CRM connector (Salesforce / HubSpot / Pipedrive): how much activity/comms history (emails, meetings, calls, notes, tasks) the first sync imports. window_months (1–120) OR all_history. Reference records (accounts, contacts, deals) always land in full. Omit for the 12-month default.

For a connector that supports it (HubSpot and Salesforce emails; Zero meetings, whose transcripts follow their meeting): a SOFT filter over which comms records (e.g. emails or meetings) the first sync imports — keep only external / rep-involved messages, dropping internal-only or irrelevant ones before they land. Applies only to the named comms streams; account/contact/deal records are untouched. Own domains stamped server-side. Omit to import comms unfiltered.

New display label for this connection instance (trimmed; must be non-empty). Renames any connection so multiple instances of the same connector type can be told apart.

The workspace member (auth user id) this personal connection belongs to. Pass null to clear the owner. Only valid for personal connectors (Gmail, Outlook).

For api_key connectors: the replacement API key / token to store (encrypted) on the existing source. Ignored for oauth and social-handle connectors.

For an api_key connector whose credential is a PAIR (today only Aircall): the non-secret id half, re-stored alongside the replacement token. Required when the catalog entry asks for it.

For oauth connectors: console URL to redirect back to after the re-handshake (first-party origins only).

The call-filter config (inclusion polarity). Own domains are stamped server-side.

The recency window: window_months (1–120) OR all_history. Omit both for the 12-month default.

The comms filter: which streams + predicates. Own domains are stamped server-side.

Free-text ask. Meaning-shaped wording ("what do customers say about…") routes to the semantic lane; other questions route to the NL-to-SQL fuzzy lane.

How many independent sub-questions a multi-part ask may fan out into. Default 3; up to 5 synchronously and 12 with async:true. RAISE THIS when a reply comes back with retry_guidance.action = "raise_max_subqueries" — it means parts of your question were cut by this cap, and retry_guidance.suggested_params carries the value to use.

Run as a background job instead of blocking. Returns {job:{job_id,status}} straight away; collect it by calling this verb again with that job_id and no query. USE THIS when a reply comes back with retry_guidance.action = "run_async" — the synchronous path is bounded at ~15s and your ask needs longer, and the async path also allows a higher max_subqueries. Fuzzy lane only; others name it in corpus.ignored_params.

Collect a job started with async:true. Send this ALONE (no query). While status is "queued"/"running", poll again; on "complete" the full envelope is on detail.

Lane override. Default auto: filters-with-no-query → filter; meaning-shaped query → semantic; else fuzzy. Lanes read DIFFERENT stores at different grains — read `corpus` on the response to see which one answered.

Warehouse surface for the filter lane. Default interactions. Ignored by the fuzzy and semantic lanes, which report it back in corpus.ignored_params when you pass it.

Who counts as the voice you want. Default "all". Use "customer_voice" to keep only genuine customer / prospect speech and drop investor, advisor, media, partner, vendor and competitor chatter that merely shares an account record — while STILL keeping a real buyer on such an account (someone on a live opportunity, or a scored champion / economic buyer). This applies the platform's own canonical rule server-side; it is a disjunction you cannot express with `filters`, which are ANDed. interactions only. Applied on the filter lane, and on the semantic lane with hydrate (turned on for you unless you send hydrate:false). A request that would land on the fuzzy lane, which writes its own SQL, is rerouted to semantic with hydrate and says so in rerouted_from and notes. Refused on the lexical lane, which matches exact terms.

Semantic lane only. Semantic matches carry a truncated preview and no labels, so they cannot be quoted or attributed. Set true to bridge each match back to its full warehouse row — full `content` plus company, speaker title, pushback and buyer-signal columns. One row per match, containing the matched utterance, so a hit is citable. SPEAKER comes from the ATOMIC utterance that matched, not from the turn stamp: read speaker_attribution on each row. "utterance" = verified from the atomic source. "ambiguous" = two matched utterances share this row and disagree on speaker, so the speaker fields are NULL and nobody is named — check turn_speakers for who is in play. "turn" = the atomic source had no row, so the fields are the turn's FIRST member and are NOT verified. Separately, speaker_ambiguous says the row's TEXT spans more than one speaker (13.58% of turns on the tenant measured 2026-08-15) — a row can be cleanly attributed AND span two voices, so quoting the whole content as one person is still wrong there. Both fields ship on every turn-grain row including the clean ones; corpus.speaker_ambiguous_rows / corpus.speaker_unattributable_rows are the rates. speaker_ambiguous_source says where that verdict came from: "stamped" = read from the column the pipeline writes on the turn itself, "overlay" = reconstructed at read time because the pipeline recorded none for that turn. A missing record is NOT "one speaker" — permanently-excluded interactions never get one — so an unrecorded turn falls back rather than being reported unambiguous, and a "stamped" row also carries speaker_count. corpus.speaker_stamped_rows / corpus.speaker_fallback_rows are the split. Stakeholder columns (role_level, is_champion, is_economic_buyer, champion_score, inferred_persona) are derived from the turn stamp, so they are NULLED when the row is re-attributed to a different speaker rather than left describing someone else. Slower — two reads. Check corpus.hydrated_grain: "turn" is that row-per-match answer; "call" means the turn lookup failed and the rows are arbitrary turns from the right conversations (context, not the matches; also flagged corpus.degraded). Matches whose utterance belongs to no turn have no warehouse row and are counted in corpus.unresolved_matches; matches sharing one turn are corpus.collapsed_matches. Read those two directly rather than inferring a gap from a match_count/row_count delta — corpus.match_count is how many nearest neighbours the lane pulled and is BOUNDED BY limit, not a count of how many things genuinely matched. Caps at 200 distinct ids, reported as corpus.truncated.

Opt into gap detection on corpus.coverage. Every response ALREADY carries corpus.coverage with max_timestamp / min_timestamp and an age histogram, computed free from the rows returned. This flag adds `gaps`: windows where the corpus HELD rows and this result set had none — the "your evidence has a hole in it" signal, as opposed to the ordinary quiet periods a plain empty-bucket scan would flag. Costs one extra bucketed read, so it is off by default. Read coverage.basis first: "matched_set" means the numbers describe your whole match set, "returned_rows" means they describe a truncated slice, and "unavailable" means coverage could not be measured — never read that as zero.

Typed predicates (ANDed — there is no OR; for the customer-voice disjunction use `audience`). Semantic mode supports only company_id / occurred_at / speaker_type, and cannot apply any of them on its theme-index fallback — check corpus.filters_applied. The filter lane supports every field the catalog advertises.

Filter lane: order by a vocabulary field or a metric alias (e.g. "count").

Row / match cap; max 1000. A larger value is clamped to 1000 rather than refused, and the response reports it on limit_clamped and in notes. Default is lane-dependent: 100 on the filter and semantic lanes, 50 on the fuzzy lane (where an unforced plain-language question usually lands), so pass it explicitly when the row count matters.

Filter lane: group fields (requires at least one metric).

Filter lane: aggregations (sum/avg need a numeric field).