Executive synthesis

Executive Summary

agents.executive_summary

Turns structured project, research, evaluation, or workflow notes into a concise executive-ready summary. It accepts seven named source sections, generates an overview, extracts key insights, and recommends focus areas for next steps.

POST /v1/agents/agents.executive_summary/invokeagents:invoke

When to use it

Good fit

Creating a concise brief after another workflow has collected discovery notes, research findings, evaluation results, or stakeholder input.
Generating dashboard summaries for project intake, operational status, product discovery, grant planning, coalition work, or curriculum review.
Turning curated material into structured output that can be rendered as overview text, insight cards, and focus-area lists.
Adding a synthesis step after search, interview, assessment, or report-building workflows.

Use something else when

Fact checking or evidence verification. The agent summarizes what you provide.
Responses that need citations, source snippets, or independent evidence lookup.
Single arbitrary transcripts that have not been mapped into the documented section fields.
Legal, medical, financial, safety, or other high-stakes decisions without expert review.

Use after another workflow has collected and structured project, research, evaluation, or stakeholder context.

Required inputs

session_id: A caller-provided identifier for the source session, workflow, or brief.
sections: Object containing the seven source section strings used for synthesis.
sections.challenge_summary: Core challenge, problem, opportunity, or question being summarized.
sections.desired_outcomes: What success should look like, including goals, criteria, or expected results.
sections.functional_needs: Functions, capabilities, actions, or jobs the work needs to support.
sections.operating_context: Relevant environment, timeline, scale, geography, organization context, or execution setting.
sections.constraints: Boundaries, non-negotiables, risks, policy limits, resource limits, or implementation constraints.
sections.stakeholders: People, teams, communities, partners, decision-makers, or affected groups.
sections.assessment: Current state, what is working, gaps, pain points, evaluation findings, or readiness signals.

Optional inputs

project_name: Optional name for the initiative, report, workflow, or document set.
configuration.metadata: Optional caller-owned metadata used to connect job records, webhooks, and local workflow state.

What comes back

The completed job result is structured for integration. Preserve citations, warnings, confidence fields, and source context when the agent returns them.

Representative output fields

job_id: Identifier used to poll GET /v1/jobs/{job_id}.
status: Job state, such as queued, running, completed, or failed.
agent_id: The invoked agent, here agents.executive_summary.
poll_url: Relative polling URL for the job result.
estimated_cost_usd: Preflight estimate used before the job runs. The final charge can differ.
created_at / started_at / completed_at: Job lifecycle timestamps when available.
result.overview: Narrative summary suitable for executive briefings or dashboard summaries.
result.key_insights[]: Insight objects for cards, bullets, or review queues.
result.recommended_focus_areas[]: Areas that deserve attention in follow-up decisions or workflows.
result.charge_usd: Customer-visible amount charged for the completed public API call when billing applies.

Result card fields

key_insights[].insight: Concise synthesized takeaway.
key_insights[].source_section: Display guidance about which supplied sections shaped the insight. Treat it as orientation, not citation proof.

How output is produced

Authenticates the API key and confirms permission to invoke agents.
Checks whether the account can start a billable synthesis job before work begins.
Accepts the request as an asynchronous job and returns job_id, status, agent_id, poll_url, estimated_cost_usd, and created_at.
Reads session_id, optional project_name, and the seven strings supplied under inputs.sections.
Arranges the sections into a synthesis task that asks for a professional third-person executive brief.
Validates generated content against the expected result shape and attempts one formatting correction when output is not usable.
Returns overview, key_insights, recommended_focus_areas, and charge_usd when the completed job is billable.

Trust and caution

The agent summarizes supplied material. It does not verify facts or retrieve missing evidence.
source_section is useful display guidance but should not be presented as a verified citation.
If input sections are sparse, contradictory, or biased, the summary can inherit those limitations.
Review generated material before high-stakes use, external publication, or customer delivery.
Best results come from compact but information-rich section strings. Newline-separated lists are fine inside string fields when they preserve structure.
The sections value itself must be an object, and any provided section field must be a string.

Runtime

Treat this endpoint as asynchronous even when short jobs complete quickly. A typical integration should show queued or running state, poll with backoff, and render the completed result when status becomes completed.

Cost behavior

The queued response includes estimated_cost_usd for display or logging before completion. Use charge_usd from the completed result for final customer-visible accounting.

Billing notes

The queued response includes estimated_cost_usd as a preflight estimate.
Completed billable jobs include charge_usd in result. Treat charge_usd as the customer-visible final charge.
The final charge can be lower or higher than the preflight estimate depending on the work performed.
Organization credits are deducted for billable completed work according to the account's public API pricing.

Representative input

{
  "inputs": {
    "session_id": "brief_coalition_001",
    "project_name": "River Coalition Governance",
    "sections": {
      "challenge_summary": "A coalition of 12 watershed organizations struggles with centralized decision-making that slows action and reduces trust among smaller members.",
      "desired_outcomes": "The coalition wants equitable voice, faster decisions, clearer accountability, and a governance model that can adapt as funding and climate conditions change.",
      "functional_needs": "Distribute authority\nBuild trust\nMaintain accountability\nEnable rapid response",
      "operating_context": "Pacific Northwest watershed region; 12 organizations ranging from volunteer groups to regional institutions; five-year planning horizon.",
      "constraints": "No single organization can hold veto power. Decisions must be reachable within two weeks. The process must work for low-capacity member organizations.",
      "stakeholders": "Executive directors, community organizers, tribal representatives, funders, municipal water agencies, and state regulators.",
      "assessment": "The rotating chair model builds participation but creates inconsistency. Smaller organizations feel unheard, and urgent decisions often move through informal channels."
    }
  },
  "configuration": {
    "metadata": {
      "workflow_id": "wf_coalition_001",
      "step": "executive_summary"
    }
  }
}

Queued response

{
  "job_id": "8f4c2e5b9a16487abf07b78d5b87d0b3",
  "status": "queued",
  "agent_id": "agents.executive_summary",
  "poll_url": "/v1/jobs/8f4c2e5b9a16487abf07b78d5b87d0b3",
  "estimated_cost_usd": 0.12,
  "created_at": "2026-05-26T20:30:00Z"
}

Representative completed result

{
  "job_id": "8f4c2e5b9a16487abf07b78d5b87d0b3",
  "agent_id": "agents.executive_summary",
  "status": "completed",
  "created_at": "2026-05-26T20:30:00.000000+00:00",
  "started_at": "2026-05-26T20:30:01.000000+00:00",
  "completed_at": "2026-05-26T20:30:12.000000+00:00",
  "result": {
    "overview": "River Coalition Governance is addressing a coordination challenge in which uneven organizational capacity and centralized decision paths limit trust, speed, and accountability. The initiative seeks a governance model that gives smaller members meaningful voice while preserving the ability to make time-sensitive watershed decisions.",
    "key_insights": [
      {
        "insight": "Decision speed and perceived legitimacy are linked; faster processes will only help if smaller organizations can see how authority and accountability are distributed.",
        "source_section": "Section 3a, Section 3b, Section 6, Section 7"
      },
      {
        "insight": "The rotating chair model has participation value but lacks the consistency needed for urgent or high-stakes decisions.",
        "source_section": "Section 8, Section 6"
      }
    ],
    "recommended_focus_areas": [
      "Distributed decision rights with explicit escalation paths",
      "Trust-building routines for low-capacity members",
      "Accountability mechanisms that do not depend on a single chair"
    ],
    "charge_usd": 0.008
  }
}

Errors and retries

Common failures

401: the API key is missing or invalid.
403: the key does not have agents:invoke scope or the agent is not available to the caller.
402: the account does not have enough credits to start the job.
429: the client exceeded the allowed request rate.
503: the platform could not complete preflight checks for the job.
failed: the accepted job could not complete, often because required input was missing, input shape was invalid, synthesis failed, or billing could not be completed.
404: the job ID is wrong, expired, or not owned by the caller.

Retry guidance

After a job_id is returned, poll GET /v1/jobs/{job_id} with bounded exponential backoff.
Honor Retry-After on 429 responses.
Retry transient 503 responses only when no job_id was returned, and use caller-side safeguards to prevent duplicate submissions.
Do not blindly retry POST /invoke after a network timeout because the public contract does not currently include an idempotency key.
Store completed results in your own system if users need long-term access.

Composition ideas

Search then synthesize

Use search or knowledge endpoints to gather candidate material, curate useful findings into the seven section strings, then invoke this agent to produce non-technical front matter.

Interview or form intake

Collect user input through a structured form or interview flow, normalize responses into the seven section fields, then display overview, key_insights, and recommended_focus_areas in a review screen.

Report assembly

Run earlier analysis, evaluation, or research steps first, preserve detailed findings in the report body, and use this endpoint to generate the executive opening section.

Workflow memory

At a checkpoint or session end, map current workflow state into the seven sections, invoke the agent asynchronously, and store the completed summary as a compact handoff for the next step.

Application fit

Project intake recaps for team dashboards and review queues.
Research digest front matter after evidence has already been gathered and curated.
Design or planning briefs from context, stakeholder needs, constraints, and success criteria.
Curriculum module summaries for instructors or program leads.
Product discovery synthesis from interview themes and decision constraints.
Operational dashboard narratives for initiatives, incidents, programs, or campaign status.
Evaluation readouts from rubric results, assessment findings, and follow-up criteria.
Session recaps for AI assistants at the end of multi-step workflows.
Grant, coalition, or partner briefings for leadership and funders.

Back to agent catalog