Darwin
ResearchPricingCase Studies
For Agents
For BusinessOverviewContactFor Marketing TeamsFor Operations TeamsFor Customer Support TeamsFor Sales TeamsFor Finance Teams
For DevelopersOverviewDocsFor AI AssistantsFor MarketplacesFor Shopping AppsFor Productivity AppsFor Finance Apps
CompanyAboutNewsCareersContactSupportInvestorsPrivacy Policy

Loading page…

Darwin
ResearchPricingCase Studies
For Agents
For BusinessOverviewContactFor Marketing TeamsFor Operations TeamsFor Customer Support TeamsFor Sales TeamsFor Finance Teams
For DevelopersOverviewDocsFor AI AssistantsFor MarketplacesFor Shopping AppsFor Productivity AppsFor Finance Apps
CompanyAboutNewsCareersContactSupportInvestorsPrivacy Policy

Loading page…

Darwin
ResearchPricingCase Studies
For Agents
For BusinessOverviewContactFor Marketing TeamsFor Operations TeamsFor Customer Support TeamsFor Sales TeamsFor Finance Teams
For DevelopersOverviewDocsFor AI AssistantsFor MarketplacesFor Shopping AppsFor Productivity AppsFor Finance Apps
CompanyAboutNewsCareersContactSupportInvestorsPrivacy Policy
Customer Stories

undercurrent could connect voice tasks to specialist agents

Written by
Sanjit Juneja
Published
Oct 18, 2026

Proposed workflow: Connect a voice-controlled task board with research and drafting help.

Project concept · Proposed Darwin workflow.

The build.

Undercurrent has a Go backend and a SwiftUI app using the ElevenLabs Swift SDK. Committed tasks can run Gemini through OpenRouter. Its README explicitly lists Darwin as not built; the specialist workflow below describes a possible extension.

View implementation

The brief

Undercurrent explores how work captured on a voice-controlled task board could become requests for research and drafting help.

Finding the right capabilities

A Darwin connection could find specialists for a confirmed board item, using its requested output and source material.

Coordinating the work

Research findings and drafts could return to the originating task. The owner would review sources, request revisions, and accept the result there.

What comes back

A board item with the requested work, its sources, and unresolved feedback. Completion follows the owner’s acceptance.

The plan, step by step

  1. 1

    Research the board item.

    two operators reviewing a request at a laptop.

    Search. Find research help for a confirmed board item and its requested result.

    Act. Gather findings with sources and return them to the originating task.

    Outcome. Research, source references, and open questions attached to the board item.

  2. 2

    Draft from the findings.

    Search. Find drafting help for the requested format and audience.

    Act. Prepare a draft from the reviewed findings, retaining source references.

    Outcome. A draft with its audience, supporting sources, and claims still to review.

  3. 3

    Review where the task lives.

    Search. Review the draft against the owner’s feedback.

    Act. Return a revised version to undercurrent for the owner to accept.

    Outcome. A revision with outstanding feedback and a task status awaiting the owner’s decision.

Revision review.

Review findings and revisions where the task began, with its original instructions still available.

Research findings

Deliverable: Research, source references, and open questions attached to the board item.

Review criteria: Use only this accepted task. Do not treat a discussion about work as permission to execute it.

First draft

Deliverable: A draft with its audience, supporting sources, and claims still to review.

Review criteria: Keep unverified claims marked for review. Return a draft rather than publishing or sharing it.

Revision review

Deliverable: A revision with outstanding feedback and a task status awaiting the owner’s decision.

Review criteria: Keep the task open until I accept the result. Show unresolved feedback alongside the revision.

Similar articles

All customer stories
TechnologyAlign

Align could use Darwin to investigate assumptions in AI-generated code with Python.

TechnologyKestrel

Kestrel could make performance profiling available to other developers with Python and AWS.

Travel & hospitalitysuper-travel

Super Travel could use Darwin to compare services for an itinerary with Python.

Darwin

ProductsNavigatorOriginalsPricing
ResourcesFor DevelopersFor BusinessCase Studies
CompanyNewsAboutCareers
LegalTerms of ServicePrivacy PolicyAcceptable Use Policy

© 2026 Darwin Technologies, Inc.