Skip to content
Talbot TechBuilding for the web since 1996 Work with Sean

The AI org chart / Who does the work

Launching a web-based startup doesn’t mean hiring humans anymore.

AI agents develop the software, design the experience, plan campaigns, and handle operations across my companies. Here’s the work they’re assigned, the tools behind it, and where I direct and review the results.

Explore the team Estimate the work ↗How to read the chart ↗

Snapshot 9 companies

Plexus benchmarks / Planning scenario

Estimated portfolio savings.

Totals cover the full role-card plan, including roles outside the current filters. These are illustrative savings, not measured company results. Annualized savings repeat the same monthly plan for 12 months.

Full portfolio · all assigned roles

Estimated monthly savings
$3,960.28
Annualized estimated savings
$47,523.36

40 role assignments · pilots excluded

Edit a scenario

01 / Companies → Functions → Roles

The team behind the companies.

Meet the companies ↗
Reset filters

1 of 9 companies shown 4 role assignments + 1 pilot

Lines show work and human accountability, not legal ownership or executive appointments.

Card amounts are monthly planning scenarios, not measured company savings. Function hours and shared budgets are allocated across roles once; pilot work is excluded. Card amounts are rounded individually, so combined calculations can differ by a few cents. Edit the assumptions ↗

  • Human accountability

    Sean Talbot

    Sets direction, owns the assignments, and reviews the results.

    Human
    • Public-interest projects

      Fervent Wishes

      Media, public records, and civic software.

      4 assigned roles · 1 pilot

      Company total · all functions

      Estimated monthly savings
      $454.46
      Annualized estimated savings
      $5,453.52

      4 role assignments · pilots excluded

      Development03

      CodexSupervised agent
      Newsroom & AI integration engineer

      United We Stand · Newsroom

      Develops the publishing system, source aggregation, and transcription and translation interfaces.

      Human review Sean Talbot

      Illustrative / month

      Human benchmark
      $153.84
      AI + review budget
      $20.67
      Modeled savings
      $133.17
      Scenario assumptions

      Planning hours are shared equally among roles within this function. Tool and review budgets are allocated by those hours across the portfolio.

      Workload
      2.35 hours / month
      AI / tool budget
      $6.69 / month
      Human review
      0.21 hours × $65.38 / hour = $13.98 / month
      Benchmark basis
      Employee wages
      Employer burden
      0.00%
      Edit this estimate
      Human oversight & verification

      Sean sets the brief, reviews the work, and approves publication or release.

      Latest evidenced work:

      Active Laravel publishing, reader, and local language-model integration work verified in workspace records.

      ClaudeSupervised agent
      Media intelligence engineer

      Atlas

      Builds transcript search, clip-selection tools, captioned rendering, and public evidence pages.

      Human review Sean Talbot

      Illustrative / month

      Human benchmark
      $153.84
      AI + review budget
      $20.67
      Modeled savings
      $133.17
      Scenario assumptions

      Planning hours are shared equally among roles within this function. Tool and review budgets are allocated by those hours across the portfolio.

      Workload
      2.35 hours / month
      AI / tool budget
      $6.69 / month
      Human review
      0.21 hours × $65.38 / hour = $13.98 / month
      Benchmark basis
      Employee wages
      Employer burden
      0.00%
      Edit this estimate
      Human oversight & verification

      Sean sets the brief, reviews the work, and approves publication or release.

      Latest evidenced work:

      Delivered Claude lane-worker tasks and media tooling commits verified in source and session records.

      CodexSupervised agent
      Media tooling engineer & QA

      Fervent media tools

      Builds audio review and editing features, import tools, and export workflows.

      Human review Sean Talbot

      Illustrative / month

      Human benchmark
      $153.84
      AI + review budget
      $20.67
      Modeled savings
      $133.17
      Scenario assumptions

      Planning hours are shared equally among roles within this function. Tool and review budgets are allocated by those hours across the portfolio.

      Workload
      2.35 hours / month
      AI / tool budget
      $6.69 / month
      Human review
      0.21 hours × $65.38 / hour = $13.98 / month
      Benchmark basis
      Employee wages
      Employer burden
      0.00%
      Edit this estimate
      Human oversight & verification

      Sean sets the brief, reviews the work, and approves publication or release.

      Latest evidenced work:

      Song Filter implementation, waveform editing, import, and WAV export work verified in workspace records.

      Design01

      CodexSupervised agent
      Newsroom interface designer

      Newsroom

      Designs the reader, caption overlay, and review interfaces around AI-generated text.

      Human review Sean Talbot

      Illustrative / month

      Human benchmark
      $66.67
      AI + review budget
      $11.72
      Modeled savings
      $54.95
      Scenario assumptions

      Planning hours are shared equally among roles within this function. Tool and review budgets are allocated by those hours across the portfolio.

      Workload
      1.33 hours / month
      AI / tool budget
      $3.79 / month
      Human review
      0.12 hours × $65.38 / hour = $7.93 / month
      Benchmark basis
      Employee wages
      Employer burden
      0.00%
      Edit this estimate
      Human oversight & verification

      Sean sets the brief, reviews the work, and approves publication or release.

      Latest evidenced work:

      Current reader and caption overlay implementation verified in the Newsroom workspace.

      Research & strategy01

      Local language modelsPilot
      Ukrainian language processing

      Newsroom

      Tests Ukrainian transcription, transliteration, and translation for newsroom work.

      Accountable Sean Talbot

      Pilot scenario

      Excluded from booked and modeled savings.

      Explore a pilot scenario
      Human oversight & verification

      Sean directs the pilot and reviews its output before use.

      Latest evidenced work:

      Local model experiments verified in workspace records; live broadcast automation remains a pilot.

02 / What a role means here

Real assignments.
Human direction.

Each entry is a role assignment supported by company or workspace evidence. The snapshot includes agents doing current project work under my supervision, as well as agents running scheduled tasks.

One platform can fill several roles across different companies and functions. The totals count those assignments, not separate employees or unique AI agents. Pilots keep their own label.

Verification notes show the latest evidenced work; this is an assignment map, not a live activity feed.

An unverified branch means we haven’t confirmed an assignment for this snapshot. The chart makes no company executive appointment claims.

Scheduled agent

An agent with a recurring operational assignment.

Supervised agent

An agent doing development, design, marketing, research, or other project work with human direction and review.

Pilot

An experimental agent capability still being tested.

Public company context: Talbot Tech portfolio and Parallax. Expand an assignment’s verification note for its scope and human oversight.

03 / The system behind the work

Building Parallax
inside these companies.

My companies are the first test for Parallax: software for priorities, budgets, campaigns, and company decisions. This chart shows the development, design, marketing, operations, and research work happening across the portfolio.

Parallax’s executive personas describe product responsibilities. The assignments above name the agents doing the work.

Explore Parallax