Which AI models do you use?

Whichever fits the job and your rules: models from OpenAI, Anthropic, Google or open-weight models you host yourself. We test two or three against your own eval set and show you accuracy, speed and cost side by side before we pick.

What if the AI gets it wrong?

It will, sometimes — so we design for it. Agents only act on their own where a mistake is cheap and reversible. Anything that moves money, deletes records or contacts a customer for the first time waits for a person. Every run is logged, and every failure goes into the eval set so it doesn't repeat.

Who owns what you build?

You do. Code, prompts, workflows, eval sets and docs are handed over in your accounts and repositories. No platform lock-in and no licence fee to keep it running.

Fixed price per package, agreed before we start. Model and hosting costs are billed at cost and estimated in the audit — usually a few hundred dollars a month for a small team. The audit fee is credited to your first build.

How long until something is live?

Four weeks for a first agent in one channel is typical: one week of audit, two of prototyping on your real data, one to deploy with hand-off and logging switched on. Bigger systems take 6–12 weeks.

Who maintains it after launch?

Every build includes 30–60 days of fixes. After that, your team runs it with our runbooks, or we keep improving it on the Scale Retainer. Either way you get a weekly accuracy report.

Does it work with our tools?

Probably. We connect to most CRMs, help desks, accounting tools and inboxes through their APIs, or through n8n, Make or Zapier when that's simpler for your team to maintain. If a tool has no API, we tell you in the audit rather than build something fragile.

Booking Q1 2027 projects

Free automation audit

142 automations shipped

61,000 hours handed back

4.2 weeks avg. to launch

No hype.
Working
agents.

Q: What happens to our data?

It stays in the tools you already use. Agents read what they need at run time; run logs are kept for 30 days for debugging, then deleted. We use model providers' business terms that exclude training on your data, and we sign a DPA before we touch anything. (Demo commitments — edit to match yours.)

We build AI agents and automations for support, sales and ops teams — scoped, priced up front, and wired so a person signs off anything risky.

Book an automation audit

Try the agent demo

Support agent · Fennick & Oak

Online · answers in seconds

AI demo (scripted)

Agent: Hi! I'm the Fennick & Oak support agent. I can track orders, change addresses and start returns.

Scripted demo — nothing you type or click leaves this page.

Support agent · Fennick & Oak

Online · answers in seconds

AI demo (scripted)

Agent: Hi! I'm the Fennick & Oak support agent. I can track orders, change addresses and start returns.

Scripted demo — nothing you type or click leaves this page.

4.2 wks — avg. time to first live agent

142 — automations shipped

61,000 h — hours handed back per year

100 % — of money-moving actions human-approved

Demo figures for the template. Results vary.

AI Agents
RAG
Voice AI
CRM Sync
Lead Routing
Document AI
Evals
Python
n8n
Make
Zapier
REST APIs
Webhooks
Human-in-the-loop
AI Agents
RAG
Voice AI
CRM Sync
Lead Routing
Document AI
Evals
Python
n8n
Make
Zapier
REST APIs
Webhooks
Human-in-the-loop

What we build

Three offers. No catalog of forty.

Most teams need one of these first. We tell you which in the audit — and which one to skip.

Offer 01/03

See what’s included

Support and sales agents that answer, qualify and book — and know when to stop and hand over.

Support agent on web chat, email or WhatsApp

Lead qualification, routing and booking

Voice agent for appointments

Human hand-off with the full context

Eval set and a weekly accuracy report

from

$6,500

one-off build

AI Agents

Offer 02/03

See what’s included

The busywork between your tools — CRM updates, invoices, reports, onboarding — done the same way every time.

CRM sync and lead routing

Invoice intake and PO matching

Weekly reports, assembled and sent

Customer onboarding sequences

Approval steps wherever money moves

from

$3,800

per workflow

Workflow Automation

Offer 03/03

See what’s included

Your documents, tickets and wiki made searchable, and copilots your team uses inside the tools they already have.

Knowledge base over docs, tickets and wiki (RAG)

Internal copilot in Slack or Teams

Document processing for PDFs, forms and contracts

Runs in your own cloud if you need it

Access rules that match your org chart

from

$18,000

scoped build

Custom AI Systems

Selected work

Numbers before adjectives.

Every case study shows the baseline, the result and how long it took. All clients and figures here are illustrative.

Fennick & Oak

E-commerce

64%

chats closed without hand-off

64%

64% of support chats closed without a hand-off

AI Agents

5 weeks

Read case

Larkstone Realty

Real estate

4 min

lead response time

3 h 10 min

4 min

Every new lead answered in under five minutes

Workflow Automation

4 weeks

Read case

Penhallow & Reyes

Accounting

210

documents filed per person per day

210

3.5× more documents filed per person per day

Custom AI Systems

7 weeks

Read case

Illustrative results from fictional clients. Results vary.

All case studies

Packages

Fixed scope. Fixed price. Then we build.

Prices are demo values in USD. Every build ends with a handover: code, prompts, eval sets and docs are yours.

Fixed scope. Written down before we start. Changes are quoted, never assumed.

Weekly demos. A working build every Friday, on your real data.

Handover docs. Runbooks, prompts and eval sets your team can own.

Package

Scope

Timeline · Support

Price

PKG-01

Automation Audit

Find the three automations worth building

Scope

Two workshops, a process map, time-and-cost baseline, ROI model and a build plan

Timeline · Support

1 week

Q&A for 14 days

Price

$1,500

credited to your first build

Book

PKG-02

Starter Agent

One agent, one channel, one job

Scope

1 agent, up to 2 integrations, eval set of 100 real cases, human hand-off

Timeline · Support

3–4 weeks

30 days of fixes

Price

$6,500

Estimate ROI

PKG-02

Starter Agent

One agent, one channel, one job

Scope

1 agent, up to 2 integrations, eval set of 100 real cases, human hand-off

Timeline · Support

3–4 weeks

30 days of fixes

Price

$6,500

Estimate ROI

PKG-03

Growth Automation

An agent plus the workflows around it

Most chosen

Scope

1 agent + 3 workflows, CRM sync, approval steps, live dashboard

Timeline · Support

5–6 weeks

60 days + monthly review

Price

$14,000

Estimate ROI

PKG-03

Growth Automation

An agent plus the workflows around it

Most chosen

Scope

1 agent + 3 workflows, CRM sync, approval steps, live dashboard

Timeline · Support

5–6 weeks

60 days + monthly review

Price

$14,000

Estimate ROI

PKG-04

Scale Retainer

Ship and tune something every month

Scope

2 build sprints a month, monitoring, eval reviews, new automations from the backlog

Timeline · Support

Monthly, 3-month minimum

Same-day responses

Price

$4,800

/mo

Estimate ROI

PKG-04

Scale Retainer

Ship and tune something every month

Scope

2 build sprints a month, monitoring, eval reviews, new automations from the backlog

Timeline · Support

Monthly, 3-month minimum

Same-day responses

Price

$4,800

/mo

Estimate ROI

PKG-05

Enterprise

Custom AI systems in your own cloud

Scope

RAG and copilots, security review, SSO, VPC or on-prem deployment

Timeline · Support

10+ weeks

SLA and named engineer

Price

from

$40,000

scoped in the audit

Scope it

PKG-05

Enterprise

Custom AI systems in your own cloud

Scope

RAG and copilots, security review, SSO, VPC or on-prem deployment

Timeline · Support

10+ weeks

SLA and named engineer

Price

from

$40,000

scoped in the audit

Scope it

Demo prices. Running costs (model usage, hosting) are billed at cost and estimated in the audit.

Illustrative estimate

What could it hand back?

Pick a task your team repeats. Move the sliders. See hours, value and payback.

Where it pays off

Six departments. One rule: a person signs off.

Typical first automations by team, with the time they usually hand back. Demo figures — your audit measures yours.

agents in production

92%

support chats closed without hand-off (median)

<2 min

human pickup when an agent hands off

client datasets used for model training

Demo figures. Results vary.

Support

Answer order and account questions

Tracks orders, changes addresses, starts returns — and hands the rest to a person with the full chat.

~18 h / week

· demo

Sales

Qualify inbound leads in minutes

Asks the questions your reps would, scores the lead, books the call and updates the CRM.

~11 h / week

· demo

Ops

Turn emails and forms into tickets

Reads the request, fills the ticket, picks the owner and chases missing details.

~9 h / week

· demo

Finance

Read invoices, match POs, queue approvals

Extracts the fields, matches the purchase order and waits for a person above your threshold.

~14 h / week

· demo

Answer policy questions, run onboarding

Answers from your handbook with sources, and ticks off the first-week checklist.

~6 h / week

· demo

Marketing

Draft the weekly numbers report

Pulls ad, web and CRM numbers into one summary for a person to check and send.

~5 h / week

· demo

One workflow, end to end

Invoice intake for a finance team. Hover or focus a step to read what it does.

Invoice intake · demo run

A new message lands in the invoices inbox. The run starts from the attachment, not the email text.
The model reads vendor, totals, tax and due date, then matches the invoice to an open purchase order. Low confidence goes straight to a person.
A plain rule, not AI: anything over the threshold, from a new vendor or without a PO match needs approval.
The approver sees the extracted fields next to the PDF and approves, edits or rejects. Nothing is posted without this step.
The bill is created in the accounting tool, the PO is closed and the requester is told. Every step is logged.

How it works

Four steps. First agent live in week four.

Week 1

01 Audit

We map the work, measure it and pick the first automation with the clearest payback.

Weeks 2–3

02 Prototype

A working agent on your real data, tested against 100+ past cases and demoed every week.

Week 4

03 Deploy

Live in one channel, with hand-off, logging and approval steps switched on from day one.

Ongoing

04 Monitor & improve

Weekly accuracy reports, reviews of every failure and the next automation from the backlog.

Questions

Skeptical? Good.

It stays in the tools you already use. Agents read what they need at run time; run logs are kept for 30 days for debugging, then deleted. We use model providers' business terms that exclude training on your data, and we sign a DPA before we touch anything. (Demo commitments — edit to match yours.)

What happens to our data?

It stays in the tools you already use. Agents read what they need at run time; run logs are kept for 30 days for debugging, then deleted. We use model providers' business terms that exclude training on your data, and we sign a DPA before we touch anything. (Demo commitments — edit to match yours.)

Ask the demo agent →

Ready when you are

Your first agent.
Four weeks.

Start with a free 30-minute automation audit. You leave with a shortlist, a cost estimate and no obligation.

Book an audit

Try the demo

No hype.Workingagents.

64% of support chats closed without a hand-off

Every new lead answered in under five minutes

3.5× more documents filed per person per day

Fixed scope. Fixed price. Then we build.

Six departments. One rule: a person signs off.

Answer order and account questions

Qualify inbound leads in minutes

Turn emails and forms into tickets

Read invoices, match POs, queue approvals

Answer policy questions, run onboarding

Draft the weekly numbers report

One workflow, end to end

01 Audit

02 Prototype

03 Deploy

04 Monitor & improve

No hype.
Working
agents.