TaskHound gives managers a single operating picture across agents, projects, and teams: live visibility, governed delivery, measurable productivity, and a bill that stays predictable. No shadow terminals. No unreviewed pushes. No surprise AI invoices from us.
$19.95/user/mo · Bring-your-own agents · Human-gated promotion to origin · Cancel anytime
Oversight
One live board
Every agent session across projects — who, what, where, status.
Cost model
Seats + Containers
We never bill LLM usage. Your AI spend stays with your providers.
Risk control
Human push gate
Agents never hold origin credentials. Promotion is a leadership decision.
Adoption path
Team → org
Start one squad; expand seats, Containers, and enterprise support as you scale.
When every engineer runs agents on a laptop, leadership has no operating picture. TaskHound is the AI Operating System that puts agent work on a shared board: project, branch, agent vendor, elapsed time, last action, and pending permissions — filterable by person, project, or status.
checkout-v2 · Claude Code
Mara · payments · 34m · running tests
auth-refresh · Codex
Priya · platform · needs approval
infra-tf · Gemini
Jon · devops · queued (slots full)
demo live · checkout-v2--acme.th.works
Pipeline · awaiting promote to origin
Presence, not surveillance — agent work visible the way CI shows builds.
Leadership cares about cycle time, rework avoided, and capacity reclaimed. TaskHound compounds value by making sessions steerable, learnings shared, and delivery gated — so AI hours turn into shipped outcomes instead of fragmented experiments.
Illustrative model for planning conversations. Adjust assumptions to match your org; it does not include provider LLM spend (billed by your agent vendors, not TaskHound).
Est. monthly platform cost
$228.50
seats + Container tier
Est. monthly value reclaimed
$16,453
hours × loaded rate × 4.33 wks
Illustrative monthly net & payback
+$16,225 net · < 1 week payback
Model: seats × $19.95 + selected Container tier. Value uses your hour and rate inputs. Real results depend on adoption, agent quality, and process maturity.
Steering mid-session, forking worktrees, and auto-distilled knowledge mean less rework and less “what did the AI already try?” archaeology. Managers get session history and board activity as evidence — not anecdotes from private chats.
Your TaskHound bill is seats and Containers. Linking Claude Code, Codex, Gemini, or Grok is free and unlimited; concurrent slots size capacity. No prepaid balances, no LLM markup, no usage surprises from us — finance can model it on a spreadsheet.
Parallel agents in isolated Containers act like an expandable bench for spikes — still gated by human promotion, still visible on the board, still on the same knowledge base.
Full seat and tier table: team pricing. Need a formal business case or multi-team quote? Request an enterprise conversation.
TaskHound turns agent work into durable operational records: who ran what, what changed, what was approved, what learned, and what demoed. That is the difference between innovation theater and an accountable AI program.
Every transcript is retained per project — tool calls, diffs, steering notes. Audit a decision months later without reconstructing Slack threads.
Organize work across projects with the team board; use Orbit-style constellation views for a fleet-level map of agents and initiatives.
Auto-distilled learnings, pinned conventions, and plain markdown in the Container — institutional memory that survives turnover.
Build status, live th.works demos, and promote-to-origin decisions form a clear delivery timeline for stakeholders.
Permission prompts and human push gates create an explicit control surface — who authorized the risky step is on the record.
Owner, Admin, and Member roles; audit events for privileged actions; seat-level agent credentials never shared between users.
We never resell or meter LLM usage. Engineers link the agent subscriptions you already budget for. TaskHound is seats at $19.95/user/mo plus Container tiers sized by concurrent agents — one Stripe subscription, proration on seat changes, cancel anytime.
Agents run in dedicated, isolated TaskHound Containers — not on engineer laptops and not multi-tenant with other customers. Pre-session snapshots, one-click restore, secrets encrypted at rest, and a hard rule: agents only push to the hound remote.
Roll out without asking the org to standardize on one model vendor. Bring Claude Code, Codex, Gemini, and Grok side by side on the same projects and knowledge base. Developers keep their preferred tools; leadership gains one desk to govern them.
Composite scenarios based on TaskHound’s actual capabilities (live board, governed pipeline, shared knowledge, seat-based pricing). Use them as templates for your internal business case — then replace with your telemetry once live.
Agents had been running on personal machines with no shared history. After TaskHound, every session landed on one board; knowledge pins stopped re-solving the same auth bugs each sprint.
Pipeline deploys on th.works gave each branch a shareable URL. Managers stopped chasing “works on my machine” screenshares; promote-to-origin stayed human-gated.
CFO rejected usage-based “AI platforms.” Seat + Container pricing fitted the annual plan; BYO agents kept LLM spend on existing vendor contracts with no double markup.
Representative leadership voices reflecting the product’s design goals. Talk to us for reference conversations on enterprise deals.
I do not need another chatbot. I need to know which agents are touching which repos, whether anyone approved the push, and what it costs next quarter. TaskHound is the operating layer — the agents stay ours.
The ROI conversation finally got simple: seats and Containers on one invoice, LLM spend on the contracts we already have. Our board cares about predictability as much as velocity.
Security signed off because agents never hold origin credentials and every session is restorable. Adoption worked because developers did not have to abandon Claude or Codex.
Standard team plans start at $19.95 per user per month with a Scout Container included. Scale concurrent agents with Pack, Kennel, or Alpha — upgrades resize live; extra slots are $4.95/mo. Full comparison and FAQ live on the pricing page.
Stripe subscription · proration on seat changes · cancel anytime
Rolling out across business units, need a security questionnaire package, volume planning, or a structured pilot with success metrics? Write us — the people who build TaskHound read enterprise mail.
The live team board shows every running session — who, which agent, which project and branch, elapsed time, last action, and pending permissions. Admins can pause or stop any session. Full transcripts remain searchable history per project. Mobile and desktop apps push when agents need input.
No. TaskHound never bills for LLM usage. Teams bring their own Claude Code, Codex, Gemini, or Grok subscriptions. Your TaskHound bill is seats ($19.95/user/mo) plus Container tiers — a single Stripe subscription you can cancel anytime.
Agents can only push to the hound remote inside your TaskHound Container. They never hold origin credentials. A human must promote a reviewed branch to origin. Pre-session snapshots and one-click restore provide an operational undo.
Yes. Start with seats and a Scout Container (included), upgrade Container tiers as concurrent agent demand grows, and add seats with automatic proration. Enterprise programs are available for multi-team rollouts, security review support, and procurement.
TaskHound is the AI Operating System your managers can stand behind: visible fleets, governed delivery, compounding knowledge, and a cost model finance understands.
Also exploring for other roles? Main site · For Solo Developers · For Development Teams