How we work

One partner. Four steps. Ninety days to a habit.

Most AI projects stall between the demo and daily use. We own the whole journey: finding the right first problem, building it into your tools, training your team, and staying on so it keeps working. Here is exactly what that looks like.

Book a discovery callSee the first 90 daysNot sure you are ready? Take the two-minute check

The four steps

What actually happens, and what we need from you.

No step is a black box. For each one, here is what we do, what your side commits, and what you walk away with.

Step 01

Assess

Week 1

We find the two or three places where AI pays off first, and agree on one to start with.

AI strategy & readiness

What happens

  • Half-day working session with the people who do the work, not just the people who manage it.
  • We map how a task actually flows today: tools, handoffs, waiting, rework.
  • Every candidate is scored on hours back, effort to adopt, and who would own it.

From your side

  • One owner who can say yes.
  • Two or three hours from the team that does the work.

You get

  • A ranked adoption plan with expected outcomes.
  • A scoped, priced first workflow.

Step 02

Build

Weeks 2 to 4

We ship the first workflow into the tools you already use. Working software, reviewed by your team as it takes shape.

Custom AI agents and automation

What happens

  • Design the agent or automation around your real data and edge cases.
  • Wire it into your CRM, inbox, WhatsApp, sheets, whatever you already run.
  • Short review loops: you see it working in week two, not at the end.

From your side

  • Access to the tools involved.
  • Fifteen minutes of feedback, twice a week.

You get

  • A live workflow with monitoring on.
  • A written note of what it does and does not do.

Step 03

Train

Weeks 5 to 6

Your people learn to use, trust, and steer it. Hands on, on their own queue, with a playbook in their words.

AI training & enablement

What happens

  • One session per role, sixty to ninety minutes, on live work.
  • We watch where people hesitate and fix the workflow, not just the training.
  • A named champion from your team co-writes the playbook.

From your side

  • A champion per team.
  • Ninety minutes per role.

You get

  • A team that runs it without calling us.
  • A playbook that survives staff changes.

Step 04

Support

Ongoing, cancel any month

We stay on to keep it working as models and your business change, and to decide with you what comes next.

Ongoing AI support

What happens

  • Monitoring with a human who reads the alerts.
  • Prompt and model tuning when the underlying models update.
  • A thirty-minute monthly review: usage, hours back, extend or park.

From your side

  • Thirty minutes a month.
  • A shared channel your team actually uses.

You get

  • A number every month.
  • The next win queued before the current one goes stale.

Who does what

Three people, clear jobs.

Adoption fails when nobody on the client side owns it. So we ask for two named people before we start, and we bring the rest.

The owner

Founder, MD or department head

Picks the first problem, unblocks access to tools and data, and shows up to the monthly review. About two hours a month.

The champion

One person on the team who does the work today

Sits in the build sessions, tests every draft on real cases, and co-writes the playbook. The person the team asks first.

Arthat

A senior lead plus the builders

Runs the assessment, builds inside your tools, runs the training days, and stays on for monitoring and the monthly review.

The first 90 days

Week by week, so there is nothing to guess.

A typical first engagement. Timelines stretch for bigger builds, but the shape stays the same.

  1. Week 1

    Kickoff and assessment

    Working session with the team. Ranked plan. One workflow chosen and scoped.

  2. Week 2

    First build in progress

    You see a working draft inside your tools. Feedback twice a week.

  3. Weeks 3 to 4

    First workflow live

    Monitoring on. A written note of what it does. Your team starts using it for real.

  4. Weeks 5 to 6

    Training sessions

    One per role, on live work. Playbook written with your champion.

  5. Week 8

    Second workflow scoped

    Usage is a habit. We pick the next win from the plan and start again, faster.

  6. Day 90

    Review and decide

    Hours back, what is used, what to extend or park. Continue with support, or run it yourselves.

How we work with you

Clear scope. Honest pricing. No lock-in.

Adoption is a relationship, not a transaction. We keep the commercial side simple so the work can stay the focus.

Scoped, not open-ended.

Every engagement starts with a defined first problem, a timeline, and a number. You know what you are buying before we start.

Priced on outcomes.

We quote against the result you want, whether that is hours saved, faster response times, or more qualified leads. Not hours billed.

Your team owns it.

Everything we build lives in your stack and your people are trained to run it. Ongoing support is there when you want it, never a lock-in.

After the build

Built to be used, not just built.

Software nobody uses is the most expensive kind. Here is what your team walks away with, beyond the working AI itself.

Training

Role-specific sessions

Session 2 of 3Hands-on

Sales team · Working the qualified queue

90 min · your real leads, not samples

    Review the agent's draft replies together
    When to override, when to trust it
    Ten leads live, side by side with the rep

Short, hands-on, and built around each role's actual work. Nobody sits through a generic AI lecture.

Playbook

Written in your team's words

Playbook · p. 4Written with Priya, ops lead

What to do when a lead comes in

    01Nothing. The agent replies within five minutes.
    02Check the draft in your inbox if it is flagged amber.
    03Green means booked. Just turn up to the call.

When in doubt: reply “hold” and the agent pauses that lead.

A plain-language guide your people help write, so it reads like them and survives staff changes.

Support

A shared channel, with us in it

#acme-x-arthat
R

Rahul · 10:12

The agent tagged a lead as “hot” but it looks like a student. Can we tune that?

Arthat AI

Arthat · 10:19

Yes. Added a rule for .edu domains and re-scored today’s leads. Want to see the diff?

Resolved in 7 minutes

Questions go to a channel we watch, not a ticket queue. Small fixes happen the same day.

Review

A monthly review with numbers

Monthly review · March30 min with your team
38
hrs back
3
workflows live
92%
used weekly
DecisionExtend support agent to WhatsApp next.
ParkedInvoicing, until the new ERP lands.

Every month we look at what is used, what is not, and decide together what comes next.

Names and numbers above are illustrative. Ask what it would look like for your team

FAQ

Good questions. Straight answers.

The things people ask on a first call, answered before you book one. No jargon, no pitch.

Still wondering about something?

Ask us directly. A 30-minute call, or a message if you would rather write. We answer everything, including “is AI even right for us?”

Book a discovery callor send a message

It means we stay with you for the whole journey, not just the build. We help you decide where AI pays off first, build it into the tools you already use, train your team to run it, and support it as your business and the models change. One partner from the first idea to daily habit.

Ready to make AI part of how your business works?

Book a 30-minute discovery call. We’ll listen, tell you where AI pays off first, and be honest if it doesn’t.

See what we’ve built