Skip to main content

We implement AI and automate businesses

Businessautopilot

Affordable digital transformation that grows your profit.

AI employees do concrete work on your real company data every day, and your team stops carrying the routine by hand.

Message on Telegram See the pipelines
01 Marketing 02 Sales 03 Operations 04 Finance 05 Management 06 Other
Live channel · workshop
96 action items extracted from meetings in a month
380 sales calls reviewed in two weeks
6,000 product cards written in three weeks instead of three years

01 · Pipeline board

Six pipelines on autopilot

Each pipeline is bought separately and takes its place in one system. Signal, action, result. Three pipelines for the team, two for the owner, one built around your task.

For the team

01 · Marketing on autopilot

Leads come from search, social media and the dormant customer base without a copywriting department.

Signal:
company knowledge, product specs, buyer queries, old customer base
Action:
product cards, articles and posts for human review, base ranking and AI outbound calls
Result:
search and your own base become lead channels
How the catalogue was filled in three weeks instead of three years
6,000 product cards in three weeks instead of three years
×20 lower cost per card than a copywriter
1 day per new product: description on arrival

For the team

02 · Sales on autopilot

No lead is lost, every call is scored by the end of the day.

Signal:
leads, calls, messages, funnel
Action:
qualification at any hour, every call scored against the company checklist
Result:
a report per employee the same day
How 380 calls got reviewed instead of a handful
380 calls reviewed in two weeks
1 min per call review instead of fifteen
24/7 lead intake and qualification

For the team

03 · Operations on autopilot

Living procedures, routine operations done by the system, a newcomer productive in three weeks.

Signal:
knowledge, procedures, decision history, work systems
Action:
role-based answers to the team, orders and documents handled by digital employees
Result:
knowledge stops living in heads, operations stop being done by hand
How an order went down to 2 minutes
2 min per standard order instead of twelve
3 wks to onboard a newcomer instead of three months
5,000 groups of duplicate customer records found

For the owner

04 · Finance on autopilot

The money already sitting in your data is visible every morning.

Signal:
tills, accounting system, exports, sales history
Action:
daily digest per location, alerts on drops, search for losses in write-offs and discounts
Result:
money decisions are made on numbers, not gut feeling
How six locations came into one window
₽762M of turnover verified over five years
6 sites in one window every morning
₽300K of monthly write-offs brought under control

For the owner

05 · Management on autopilot

Every meeting moves the annual goal, the gap is visible the next day.

Signal:
meetings, chats, tasks, annual and quarterly goals
Action:
action items into the tracker, plan versus actual by goal, a hypothesis bank from idea to measurement
Result:
the owner sees the gap the next day, not next quarter
How 96 action items got owners and deadlines
96 action items from meetings in a month
34 growth hypotheses in a quarter
1 day until the gap is visible, instead of a quarter

On request

06 · Your pipeline on autopilot

Your process is not on the list? We build a pipeline around it using the same method.

Signal:
a process that does not fit the standard list
Action:
the task is scoped in the intro session, the pipeline is designed, a pilot runs on your data
Result:
the same principle: signal, action, measurable result, on your own process
Discuss your pipeline at the intro session
6-8 wks to the first pilot, same as the other pipelines
Free intro session, before we estimate anything

02 · Results log

Point A. Solution. Point B.

6,000 01 Marketing on autopilotEquipment e-commerce store, several thousand SKUs product cards written in three weeks instead of three years, cost per card twenty times lower Expand Collapse

Expanded · 01 Marketing on autopilot

6,000

product cards written in three weeks instead of three years, cost per card twenty times lower

Equipment e-commerce store, several thousand SKUs

Before

Half of the 6,000 products had an empty description or a single supplier line. One writer closed 20 cards a week, search traffic was flat.

What we did

The system takes product specs, company knowledge and buyer queries. It writes descriptions, headings, search markup, articles and posts. A human reviews.

After

6,000 cards written in three weeks. New products are processed the day they arrive. Search traffic stopped being a cost line and became a channel.

For the owner

The catalogue finally looks like a store, and a lead from search costs less than a paid one.

380 02 Sales on autopilotLegal services for individuals, three departments on the phone calls reviewed in two weeks instead of a handful, system score 80 versus 79 from the manager Expand Collapse

Expanded · 02 Sales on autopilot

380

calls reviewed in two weeks instead of a handful, system score 80 versus 79 from the manager

Legal services for individuals, three departments on the phone

Before

About 500 leads a month, 2,500 calls a week. Quality control listened to a handful and skipped short calls, although half of the meaningful ones were exactly there.

What we did

A voice assistant takes leads in the evening and on weekends. Every call across three departments is scored against company rules, the funnel is reconciled daily.

After

380 calls reviewed in two weeks. The system score matched the manager score. A review takes a minute instead of fifteen.

For the owner

Managers work with ready reports and listen only to disputed calls. Conversations with sales reps run on numbers.

2 min 03 Operations on autopilotManufacturing and wholesale, about 2,000 customers per standard order instead of twelve, a new employee productive in three weeks Expand Collapse

Expanded · 03 Operations on autopilot

2 min

per standard order instead of twelve, a new employee productive in three weeks

Manufacturing and wholesale, about 2,000 customers

Before

Knowledge lived in three systems, folders and heads. Procedures were outdated. A new manager took three months to get up to speed, a standard order took twelve minutes.

What we did

Knowledge and procedures consolidated on the client server with role-based access. Virtual consultants answer, virtual employees act in the work systems.

After

An order is placed in two minutes. 5,000 groups of duplicates found, an honest list of paying customers of the year assembled for the first time. A newcomer asks the system, not a busy colleague.

For the owner

The company stopped keeping knowledge in people and started keeping it in the system.

₽762M 04 Finance on autopilotBakery chain, six locations of turnover verified over five years, a digest across six locations arrives every morning Expand Collapse

Expanded · 04 Finance on autopilot

₽762M

of turnover verified over five years, a digest across six locations arrives every morning

Bakery chain, six locations

Before

The owner assembled the digest by hand from till exports, production was planned by eye, write-offs reached 300 thousand rubles a month.

What we did

Read-only connection to the accounting system, a daily digest across all locations in Telegram: revenue, average ticket, write-offs, alerts on drops. 132 months of sales history pulled and verified.

After

The very first analysis showed: the chain grew on average ticket while the flow of receipts was falling. The priority shifted from "bake more" to keeping traffic. Next stage: production by sales forecast.

For the owner

Sees all locations in one window every morning and decides on numbers where it is burning, not on a call from the manager.

96 05 Management on autopilotManufacturing company, about 200 people action items and 84 decisions extracted from meetings in the first month, each with an owner, a deadline and a quote Expand Collapse

Expanded · 05 Management on autopilot

96

action items and 84 decisions extracted from meetings in the first month, each with an owner, a deadline and a quote

Manufacturing company, about 200 people

Before

Goals were checked once a quarter, by year end the gap was about 20 percent. Metrics were compiled by hand, managers reported tasks done by word of mouth.

What we did

The system recognises every meeting, extracts decisions and action items, creates tasks in the tracker and reconciles actuals against annual and quarterly goals.

After

The gap is visible the day after the meeting, not three months later. No lost agreements. In the same pipeline, 34 hypotheses were collected in a quarter, 9 tested, 3 gave a measurable effect.

For the owner

Stopped asking "what about the tasks": reads the log and comes to the coordination meeting with specific questions.

Discuss a similar process

03 · Launch sequence

How to start

The first step is free: in an hour we find the area where a pipeline pays back fastest.

Book an intro session

  1. Step 1

    Intro session

    We look at the business, I find the spots where AI brings money fastest.

    one hour, free

  2. Step 2

    Scouting

    A test on your real materials, often before the contract.

    a few days

  3. Step 3

    Stage 1 turnkey

    One or two pipelines on live data, acceptance criteria in the contract.

    6-8 weeks

  4. Step 4

    Growth

    Next pipelines, team training and support.

    per stage estimate

Data stays inside the company. Connections to systems are read-only wherever possible.

Acceptance agreed upfront. Criteria in the contract, part of the payment on confirmed result.

The product stays with the company. The team learns to run the system on its own.

04 · Frequently asked

Questions before the first conversation

01

Why is the intro session free?

In an hour it is clear whether there is an area in the company where a pipeline pays back. If there is none, I say so plainly and we part without an invoice. If there is, the next step is scouting on your materials.

02

Our business is special, AI will not fit.

Pipelines are built on company data: tills, CRM, messages, meetings. We have worked with bakeries, lawyers, manufacturing, e-commerce, wholesale. They share one thing: routine that rests on people.

03

How is this different from chatting with an AI model?

A chat answers a question, a pipeline does the work: it is connected to your systems, runs every day without reminders and reports in numbers. A human reviews instead of doing.

04

What happens to our data?

Data stays inside the company, connections to systems are read-only wherever possible. Sensitive processes run on models on your own server. Client names are never published.

05

How much does it cost?

The intro session is free. Scouting and stage 1 are estimated for the specific pipeline. Acceptance criteria are fixed in the contract before the start, part of the payment is tied to the confirmed result.

06

How much of our team time will it take?

A few hours over 6-8 weeks: access, answers to questions, acceptance. After that the system runs on its own and the team learns to work with it.

Founder of AI Potential

05 · AI Potential · who is behind this

Mikhail Zelenin. 20+ years in IT: from developer to AI systems architect. I live and run my business in the same kind of system.

What I sell to companies I first built for myself. I connect to tills, CRM, spreadsheets and messages, assemble a pipeline in 6-8 weeks and hand it over against acceptance criteria. No fake charts: only numbers that the client team verifies.

06 · Next step

Let us find the work a digital employee will take over first.

Message on Telegram