What if our volume grows?

The agent handles it; nothing breaks. Model usage rises with volume and is billed at cost, so you see it in the monthly report. If volume stays well above the quoted band, we agree a new monthly fee at the next renewal.

Yes. Packaged Agents run month to month with 30 days' notice. The setup fee isn't refundable once the agent is live.

Who owns the prompts and configuration?

Your data, your business rules, your templates and the test examples built from your work are yours. If you leave, we hand them over in a documented export. Our shared agent framework and tooling stay ours.

How is our data handled?

Access is scoped to what the agent needs, read-only by default, with write access only where you approve it. We use model providers under terms that exclude training on your data, and every action is logged.

What if we want it customised beyond the package?

Small changes to rules, templates and routing are part of the monthly fee. New integrations, new workflows or a customer-facing agent are an Agent Build, priced separately in two fixed stages.

Do we need the pilot first?

No. The pilot is for teams that want proof on their own numbers before committing; it comes with a money-back guarantee, and the setup fee is waived if it works. If you already know the workflow costs you hours, go straight to setup.

Which AI models does it use?

Whichever does your task best at the lowest cost on your own test set — usually a mix from Anthropic, OpenAI or Google, with routine steps routed to a smaller, cheaper model. We switch when a better option passes the evals.

Packaged Agents · Run for you

Ready-made AI agents, set up and run for you.

Three agents for the work that eats the most hours in a growing company: sales follow-ups, invoices and documents, and the support queue. We fit one to your tools and rules, put it on your real work, and then keep it running — hosting, monitoring, evals, model upgrades and fixes, for one monthly fee.

See the three agents How pricing works

Setup: $1,500; Per agent. Waived after a successful pilot
Monthly: From $500; Per agent, set by volume
Live in: 1–2 weeks; From access to live work
Model usage: At cost; Passed through, shown in the report

The three agents

Pick the agent for the work that eats your week.

Each one is a tested pattern we fit to your tools, rules and tone. Same pricing for all three; you can run one, two or all of them.

Sales follow-up agent

Follow-ups drafted every morning, the CRM kept current and quiet deals flagged, inside HubSpot or Salesforce.

Today

Reps write follow-ups from memory after calls
Deal stages and next steps updated days late
Stalled deals noticed at the pipeline review

With the agent

A drafted follow-up for every open conversation, waiting each morning
Stage, next step and contacts proposed for one-click approval
Deals with no activity flagged to the owner in Slack or Teams

Example run

Illustrative

07:00Read 18 call notes and 36 email threads since yesterday
07:02Drafted 15 follow-ups in each rep's tone
07:023 drafts mention discounts — held for rep review
07:0324 stage and next-step updates proposed for approval
07:03Morning digest posted to each rep in Slack

Systems it touches

HubSpot or SalesforceGmail or OutlookCall notesSlack or Teams

We measure

Time to follow up
Drafts sent with light edits
CRM fields complete

Human checkpoint

Every email is a draft. A rep reviews and sends it.

Full detail on the Sales follow-up agent

Need something these three don't cover? Start with an AI Agent Pilot on your workflow, or talk to us about an Agent Build.

What the monthly fee covers

You get the work done. We keep the agent working.

An agent that worked in week one can drift when your tools, your data or the models change. The monthly fee pays for the work that stops that.

sales-followup · prod

CRM · read + draftInbox · draftsSlack · post

Always on

Hosting

The agent runs on infrastructure we manage, with scoped credentials to your tools. Nothing for your team to install or patch.

Runs, last 7 days

Every run

Monitoring & alerts

Every run traced and logged. Failures, slowdowns and odd outputs alert us, usually before your team notices.

tone_matches_reppass

no_price_promisespass

fields_from_sourcepass

Every change

Evals

A test set built from your own examples. Every prompt, rule or model change has to pass it before it reaches your work.

Model change proposed

Same eval score, lower cost per task

Switch after eval pass

When it pays

Model upgrades

When a newer or cheaper model does your task as well or better on your test set, we switch. If it doesn't, we don't.

New vendor format handled

Template wording updated

CRM field renamed

As needed

Fixes & small changes

When one of your tools changes its API or a new edge case shows up, we fix it. Small rule and template changes are part of the fee.

Monthly report

Volume

Accuracy

Model cost

Every month

Monthly report

Volume handled, accuracy, time saved against your baseline, model cost, and what we changed. One page, plain English.

Pricing

One setup fee, one monthly fee, models at cost.

No per-seat pricing and no markup on model usage. Two ways in: prove it with the pilot first, or start straight away.

Setup

$1,500per agent, once

Waived after a successful AI Agent Pilot on the same workflow.

Monthly

From $500per agent / month

Set by the volume the agent handles. Quoted on the fit call.

Model usage

At costbilled monthly

What the model providers charge for your agent's work, passed through.

The monthly fee covers

Hosting and scoped access to your tools
Monitoring, alerts and tracing
Evals on every change
Model upgrades when they pay
Fixes and small rule changes
A one-page monthly report

What sets the monthly price

How many emails, documents or tickets the agent handles
How many of your tools it connects to
Whether you want more than one team or region on it

If volume moves up or down for a few months, we re-quote at the next renewal, not mid-month.

Route 1 · Prove it first

AI Agent Pilot, then the agent

$3,000 fixed · 14 days, money-back guarantee. If it works, the agent keeps running and the $1,500 setup is waived.

About the pilot

Route 2 · Start straight away

Go straight to the agent

$1,500 setup, then from $500 a month. Right if you already know the workflow costs you hours and want it running.

Compared with the alternatives

Build it, buy a tool, or have it run for you.

All three are reasonable. Which one is right depends on who you have and how closely your process matches an off-the-shelf product.

Compare	Packaged Agent	Build it yourself	A general AI tool (SaaS)
Time to live	About 1–2 weeks	Weeks to months, depending on your team	Fast to try; your team sets it up
Where it works	Inside the CRM, inbox or help desk you already use	Wherever you build it	Often in its own app, or as an add-on
Fitted to your process	Your rules, templates, tone and examples	Completely — you write it	Within what the product allows
Who keeps it working	We do: monitoring, fixes, upgrades	Your engineers	The vendor runs the product; you run your setup
Testing on your own work	Evals from your examples on every change	If you build them	Varies by product
Cost shape	Setup + monthly + model usage at cost	Engineering time + hosting + model usage	Usually per seat or per usage
Best when	You want it working in your tools without hiring for it	You have engineers with time and want to own every line	Your process matches the product closely

How setup works

About two weeks from access to a running agent.

Most setups take one to two weeks, depending on how quickly access is ready. Shown here in working days. After a successful pilot, the agent is already live and simply moves onto the monthly plan.

D10

Kickoff & access

Read-only access; 20–50 real examples collected

D1–2

Configure

Your rules, templates, tone and routing

D3–4

Test on your examples

Evals built from your past work; must pass

D4–5

Live with review

Real work, every output reviewed, tuned daily

D6–9

Hand over to monthly run

Thresholds agreed, baseline set for the report

D10

Checkpoints

Access connected

Passes your examples

Monthly run starts

D1–2Kickoff & accessRead-only access; 20–50 real examples collected
D3–4ConfigureYour rules, templates, tone and routing
D4–5Test on your examplesEvals built from your past work; must pass
D6–9Live with reviewReal work, every output reviewed, tuned daily
D10Hand over to monthly runThresholds agreed, baseline set for the report

Who does what

What we need from you, and what we run for you.

You provide

One owner for each agent, about an hour a week during setup
Access to the tools the agent works in
20–50 real examples of the work, good and bad
People to review outputs in the first week live
A short monthly check-in on the report

We provide

Setup, connectors, rules and the review queue
Hosting, monitoring and alerts
Evals on every change and model upgrades
Fixes and small rule or template changes
The monthly report and check-in

Not included

New workflows or integrations (that's an Agent Build)
Customer-facing agents or apps
Model usage (billed separately, at cost)
Data clean-up or migration projects
Licences for your own tools

Is this you?

Built for teams that want the work done, not another tool to run.

A good fit if…

You have 5–250 people and no spare engineers for AI
Your sales, AP or support work lives in mainstream tools with APIs
The work is repetitive and the rules can be written down
You want someone accountable for keeping it working

Not the right start if…

Your workflow is unusual enough that no package fits — start with an Agent Pilot or Agent Build
You need a customer-facing agent that acts without a person in the loop
Your tools have no API or export
You need a regulated-data approval (HIPAA, PCI) before anyone touches the data

Where it sits

Prove it, run it, then build what's next.

Most teams start with a pilot, keep the agent running here, and move bigger or custom work to an Agent Build. How we build every system

FAQ

Questions we hear on fit calls

What if our volume grows?: The agent handles it; nothing breaks. Model usage rises with volume and is billed at cost, so you see it in the monthly report. If volume stays well above the quoted band, we agree a new monthly fee at the next renewal.
Can we cancel?: Yes. Packaged Agents run month to month with 30 days' notice. The setup fee isn't refundable once the agent is live.
Who owns the prompts and configuration?: Your data, your business rules, your templates and the test examples built from your work are yours. If you leave, we hand them over in a documented export. Our shared agent framework and tooling stay ours.
How is our data handled?: Access is scoped to what the agent needs, read-only by default, with write access only where you approve it. We use model providers under terms that exclude training on your data, and every action is logged.
What if we want it customised beyond the package?: Small changes to rules, templates and routing are part of the monthly fee. New integrations, new workflows or a customer-facing agent are an Agent Build, priced separately in two fixed stages.
Do we need the pilot first?: No. The pilot is for teams that want proof on their own numbers before committing; it comes with a money-back guarantee, and the setup fee is waived if it works. If you already know the workflow costs you hours, go straight to setup.
Which AI models does it use?: Whichever does your task best at the lowest cost on your own test set — usually a mix from Anthropic, OpenAI or Google, with routine steps routed to a smaller, cheaper model. We switch when a better option passes the evals.

$1,500 setup · from $500/mo

Have the agent do the work, and have us keep it working.

A free 45-minute fit call: we pick the agent, check your tools and quote the monthly fee from your volume.

Book the fit call