What AI projects look like in practice.

You need that opinion from two years ago, buried in a folder with a cryptic name.

Searches through precedents and your own opinions, answers quoted and linked to the source.

Every contract starts from scratch, hand-adapting clauses already written a hundred times.

Contract drafts generated from the agreed framework: standard clauses, exceptions highlighted, versions tracked.

Procedural deadlines kept on post-its and calendars that don’t talk to each other.

Automatic procedural reminders: deadlines, hearings and expiry dates summarised every morning per case file.

The same file translated three times for three clients, each with its own terminology.

Assisted translation with glossary terminology applied and a memory of phrases already used.

The final review happens at night: numbers, references and terms to recheck by hand.

Automatic final quality check: numbers, glossary terms and reference consistency verified line by line.

After every job, a long e-mail explaining choices and points to confirm.

End-of-job summary for the client: what was translated, terminology choices made, points to confirm.

Invoices and expenses arrive in every format, to be re-entered by hand one by one.

Automatic reading of invoices and accounting documents, with the entry already set up.

At month’s end you chase whoever still hasn’t delivered documents, between e-mail and calls.

Tax reminders per client listing the documents still missing, sent automatically mid-month.

The same ten client questions, rewritten from scratch every time.

Ready answers to clients’ recurring questions — always verified by the professional before sending.

Applications compiled by hand-copying survey data, check after check.

Cadastral filings compiled from survey data, with a consistency check before submission.

Understanding what actually changed between two project versions takes hours of comparison.

Automatic project-variant comparison: surface, height and use differences highlighted in one table.

The take-off built in spreadsheets, line by line, always from scratch.

Assisted estimating: quantity take-off starts from cadastral data and price-book items used before.

Dozens of CVs to read in a few days, all in different formats.

CVs read and summarised against requirements, compared directly to the open role.

Questions prepared in a hurry, CV in hand, between one interview and the next.

Interview questions generated from the CV and the role description, with points to probe already highlighted.

Contracts, checklists and materials rewritten for every new hire.

Personalised onboarding: checklists and materials prepared for the role and destination team.

Confirmations, rescheduling and reminders written by hand ten times a day.

Routine e-mails — confirmations, reminders, cancellations — written and ready to review in seconds.

Meetings end and the minutes stay unwritten for days.

Meeting minutes from the recording: points, decisions and actions with owner and deadline.

Flights, hotels and timings compared across ten open browser tabs.

Complete trip preparation: flights and hotels compared against company preferences.

E-mails in three languages always asking the same things: timings, parking, breakfast.

Pre-stay question replies drafted in the original language with real room and service information.

Reviews arrive in every language and go unanswered for weeks.

Review replies drafted in the original language, consistent in tone, with critical cases flagged to management.

Arrivals, departures and groups summarised by hand before nine.

Automatic daily report: arrivals, departures, special requests and groups, ready before the morning briefing.

Questions at all hours about check-in, breakfast and parking, even at midnight.

Recurring guest questions answered instantly, with the property’s real information.

Every guest briefed at the last moment on keys, gates and access.

Access instructions and self-service check-in sent automatically the evening before, with map and directions.

Cribs, allergies and timings scattered across chats, e-mails and memory.

Special requests (crib, allergies, timings) collected before arrival and summarised for whoever welcomes guests.

Group requests all arrive together in high season, all to be answered by hand.

Group requests turned into proposals with real availability for the requested dates.

Availability and prices recalculated and rewritten for every group.

Pre-filled group e-mails with real availability and seasonal prices, ready to personalise.

The per-channel, per-room balance rebuilt by hand from registers at season’s end.

End-of-season report: occupancy per channel and room type, compared with last year.

Calls and messages arrive in the middle of service: one badly noted reservation is enough to lose a table.

Bookings from messages and calls seated at the right table, conflicts flagged immediately.

Allergies and special occasions remembered from memory when seating groups.

Evening reservations summary generated in the afternoon: tables, allergies, special occasions, groups.

Quantities estimated by eye, between stock and expected bookings.

Supplier orders prepared from stock levels and expected bookings, confirmed with one tap.

Restaurant and shop clients ordering by phone or messages, at every hour of the season.

WhatsApp and e-mail orders turned into delivery notes and invoices, with updated availability.

Treatment logs and mandatory records filled in by hand in the evening, often from memory.

Treatment log filled in by dictating from the field: product, dose, plot — verified and ready for inspections.

CAP applications and regional subsidies, with different deadlines and documents for each measure.

Deadlines monitored and documentation gathered in advance; application drafts start from data already on the farm.

Animal IDs, movements and treatments recorded in notebooks and loose sheets.

Digital barn register: births, treatments and movements noted by dictation, ready for authorities and audits.

Feed ordered at the last minute, between shrinking stock and rising prices.

Consumption estimated per head and costs split out: the supplier order leaves when it’s really needed.

Clients asking for origin and lot on every delivery; traceability rebuilt afterwards, by hand.

Every sale leaves with lot, date and origin already matched: traceability exists before anyone asks.

Blooms, treatments and queen cells noted per apiary, in different notebooks.

Diary dictated after every visit: apiary, super, colony state — comparable across seasons.

Direct sales across markets, shops and online, managed between chats and spreadsheets.

Orders from every channel gathered into one list with batches and weights, ready for market week.

Labels with lot and expiry are rewritten by hand after every extraction and jarring.

Labels generated from production data: lot, apiary and expiry, consistent on every jar.

Harvest and cellar notes in notebooks nobody ever rereads.

Digital cellar book: vineyard, batch, analysis and ageing — searchable by vintage and bottle.

Orders from private clients, restaurants and wine shops tracked between e-mail and calls.

Orders collected and turned into delivery notes and transport documents, with delivery reminders.

Tasting bookings via messages, with groups, intolerances and timings to coordinate.

Bookings managed with automatic reminders and a summary per visit: group, tastes, what to open.

Recipes and lots tracked on loose sheets; reconstructing what happened in a batch takes hours.

Batch sheet generated from the recipe: ingredients, temperatures, timings — comparable across brews.

Bars and restaurants ordering with a different message every time, between kegs and returnables.

Chat and e-mail orders turned into delivery notes with kegs and empties already counted.

Labels and declarations for every variant, rewritten at every new brew.

Labels and product sheets generated from the recipe: ABV, ingredients, allergens — consistent with the lot.

Interventions, feeding and stocking per pond noted on paper calendars.

Per-pond register dictated on the spot: operations, stocking, conditions — with season-by-season history.

Restaurateurs and private clients ordering fresh fish by phone, early in the morning.

Orders gathered into a catch-and-delivery plan: who receives what, when and at what time.

Lots and health documents rebuilt afterwards, when the request arrives.

Every lot is born with origin and document already matched: the file closes in minutes, not days.

Quotes written in the evening, from memory, after a day under the sinks.

Quote prepared by dictating the job, materials and times from the van, before you’re even back at the office.

Prices and times for every job type to be remembered for every customer.

Standard quotes generated from the job type, with materials and times already priced.

Customers call back only when the boiler stops: nobody keeps the calendar.

Scheduled maintenance reminders sent to customers at the right moment: boiler, shower heads, drains.

Reports compiled in the evening from the survey’s dirty sheet.

Inspection reports generated from the installation data, signed by the client on a tablet.

“When was the last inspection?” — the answer lives only in your head.

The date of the last inspection is available in seconds for every installation and customer.

Quotes prepared in the evening, line by line, price book open.

Quotes for new work generated from the dictated site survey, with up-to-date price-book items.

The client calls to ask when it’s ready; the shop stops to answer.

Repair status updates sent to the client by message, without stopping the front desk.

Explaining in writing what was done and what should be done soon, forgetting nothing.

End-of-job communications generated from the diagnosis: what was done, what to do soon, clear costs.

Nobody knows which cars are close to their service deadline.

Service-reminder campaigns planned on real licence plate and mileage.

The record is filled between sessions, always behind schedule.

Patient record compiled from end-of-session dictation, structured and readable.

Session notes written in a hurry, readable only by whoever wrote them.

Structured treatment notes generated at the end of each session and attached to the patient record.

Hygiene and checks remembered only when the patient calls in pain.

Personalised reminders per patient: hygiene, orthodontic checks, next due date — by SMS or e-mail.

The plan is updated at the end of the day, when the head is already tired.

Therapy plan updated at the end of each session: progress, adjustments, next goals.

Sheets prepared by copy-pasting from old documents, figure by figure.

Personalised exercise sheets generated from the plan, with the right images and progressions.

The right videos scattered between the phone gallery and cloud folders.

Videos and images of prescribed exercises pulled from the library and sent with the right instructions.

The owner tells the story, the record is filled in later, from memory.

Patient record updated from dictation, with scheduled booster reminders.

Post-operative instructions rewritten from scratch every time.

Discharge letters and post-op instructions generated from the surgical record, ready to review.

Parasite control and vaccines remembered only by whoever calls, not whoever should.

Parasite-control and periodic-check reminders sent at the right time per species and size.

Supplier prices change and margins are recalculated only at year’s end.

Average selling price and margin per product summarised weekly, without spreadsheets.

Stock estimated by eye between shelves and register.

Supplier price-list comparison with a reorder proposal based on real turnover.

The same questions about OTC products rewritten in writing every time.

Written answers to chronic patients about how to take their medication — always verified by the pharmacist.

The treatment history lives in notebooks and messages.

Client card with treatment history, products used and notes for the next visit.

“What do you recommend?” — the answer depends on what was done months ago.

Follow-up treatment suggestions generated from the client’s history and treatment cycle.

Whoever stops coming simply disappears, without even a message.

Win-back of inactive clients: whoever hasn’t returned in months gets a message with an appointment already proposed.

Orders placed by gut feeling, between what sells and what sits on the shelf.

Supplier orders prepared from real sales, anomalies already flagged.

Every new item means labels and descriptions to write by hand.

Labels and product sheets generated from the supplier sheet, including colour and size variants.

Before the sales season you guess what to order by looking at the shelf.

Analysis of what to order before the sales season, based on real stock and sell-through.

The same property described three times, once per portal and language.

Multilingual listing descriptions written once and kept consistent across portals and social.

Listing questions sit unanswered for hours while you’re on a viewing.

Requests from listings get an immediate reply, together with the first questions to qualify the contact.

Nobody knows which listings work and which channels bring contacts.

Interest report per property: how many contacts, from which channels, which recurring questions.

Data arrives from five different platforms, each with its own export.

Monthly report based on data from all platforms, accompanied by a clear commentary on the results.

Every month starts from scratch: what we published, what is worth telling.

Monthly newsletter drafts generated from the content published that month, in the company’s tone.

Watching competitors means ten newsletters to read and forget.

Competitor digest: campaigns, product news and prices, every Monday morning.

Supplier offers arrive in different, non-comparable formats.

Supplier offers compared on matched line items, differences highlighted row by row.

Price hikes are noticed only once the invoice is issued.

Automatic alerts when a supplier’s price crosses the threshold defined by the company.

Orders, delivery notes and invoices to reconcile by hand, line by line.

Delivery notes and supplier invoices reconciled with the order, differences already highlighted.

The CRM gets updated when there’s time — which means never.

CRM updated from the sales team’s e-mails, with deals and stages already realigned.

Rebuilding the picture of at-risk deals takes half a day of reading.

Weekly summary per salesperson: stalled opportunities, deals at risk and suggested actions.

Why do we lose? The answers are in the e-mails, but nobody rereads them.

Analysis of lost deals with recurring reasons, ready for the team meeting.

The same tickets open and resolve, every week, from scratch.

First-level ticket replies prepared using the internal knowledge base; complex cases are forwarded to the specialists.

March’s incident solution lives in the head of whoever solved it.

Runbooks generated from resolved incidents: next time the solution is documented and searchable.

Patches, certificates and access: a mental checklist nobody sees.

Periodic security digest: patches, access, certificate expiry — explained to management.

Every shoot leaves hundreds of shots to thin out by hand.

First photo cull done by AI to your criteria, refined in minutes.

Gallery, watermark and deadlines managed by hand for every client.

Client photos and videos gathered in one gallery with watermark and delivery deadlines.

One change bounces between chats, calendar and reminders.

Shoot postponements and rescheduling handled automatically against the real calendar.

Briefs arrive as voice notes and messages to decipher before starting.

Briefs rewritten as clear points, with the right questions to ask the client before starting.

Deliverables and deadlines ending up on post-its around the monitor.

The brief is turned into a work plan with deliverables, formats, deadlines and starting references.

“What changed since rev 2?” is a question nobody can answer right away.

Version comparison with an immediate summary of the changes made to the layout between rev 2 and rev 4.

The timeline lives in a giant sheet no supplier ever looks at.

Event timeline generated from the plan: who does what, where and when, in a format that can be shared with suppliers.

Ten suppliers’ proposals to collect, compare and chase by e-mail.

Supplier quotes and confirmations are collected and compared, with payment deadlines always under control.

Every cost line updated by hand, hoping to stay in budget.

Cost summary per line, with alerts when a spend exceeds its allocated budget.

Last-minute cancellations and gaps in the calendar nobody fills.

Reminders the day before and a waiting list notified for every gap: the calendar defends itself.

Colour, highlights and last visit’s products? The answer lives in whoever did it.

Client card updated at every appointment: services, formulas and preferences ready before the client sits down.

Clients come back when the colour has grown out, not when it should be done.

Follow-ups scheduled on history: whoever is late for a touch-up gets the message at the right moment.

Understanding what really changed between two project versions takes whole afternoons.

Differences between versions highlighted automatically: drawings, reports and quantities always aligned.

Specifications written by reusing old documents, clause by clause.

Draft specification generated from the project and its requirements: to review and tune, not to write from scratch.

Bills of quantities rebuilt in spreadsheets at every revision.

Quantities updated as the project changes, with cost lines consistent between drawings and estimate.

Notices, attendance and minutes to write up for every building, every month.

Minutes generated from the meeting recording, with resolutions and outcomes already structured.

Building arrears chased between registered letters and phone calls.

Reminders graded by due date and amount, with prompts to whoever handles the case.

Maintenance jobs awarded over the phone, without comparing quotes.

Quotes collected and compared per job, with each building’s history one click away.

Price lists, variants and promotions changing every week, to be communicated customer by customer.

Price-list updates generated from the catalogue and personalised per customer, ready to send.

Orders taken in shops on scraps of paper and typed up at the hotel in the evening.

Orders dictated after each visit and turned into order forms with codes, quantities and deliveries.

Summaries for principals rebuilt from memory on Fridays.

Activity and pipeline reports updated continuously: trips, visits, outcomes.

Plans written by hand for every client and recalculated at every step forward.

Plans generated from goals, availability and progress — refined with the trainer’s experience.

Between sessions the client is alone: doubts and setbacks surface late.

Scheduled messages and guided replies between sessions, with critical cases flagged to the trainer.

Expired packages and vanished clients: you notice only if they come back.

Package expiry and inactivity monitored: the right message leaves before the client disappears.

Reports and visit notes written after clinic, at the end of the day.

Draft reports from visit dictation, to review and sign: no evening transcription.

Check-ups and follow-up tests remembered only for those who come back.

Reminders for clinical deadlines and recalls, without extra load on reception.

Certificates, prescriptions and insurance paperwork multiplying the bureaucracy.

Forms and certificates pre-filled from patient data already in the system, to verify and sign.

Every contract change means recalculating payslips and overtime by hand, client by client.

Runs updated from contract changes, with consistency checks before sending.

Filings, payments and certifications: fixed deadlines to chase for dozens of clients.

A deadline calendar per client with automatic warnings: nothing slips through.

Circulars, payslips and certifications distributed between e-mail and calls.

Documents and communications delivered with unified tracking: who received what, when.

Every piece is different: quotes written from scratch after each survey.

Quote generated from client measurements and requirements, to be refined with the carpenter’s experience.

Panels cut by eye: offcuts are paid for with every wrong piece.

Cutting plans computed from available panels: fewer scraps and less wasted material.

Laminates, hardware and finishes ordered late, between workshop and site.

Stock monitored and orders ready when material drops below threshold.

CMRs and waybills filled in the yard, between one load and the next.

Documents generated from the order and the trip: one dictation and the rest is ready.

Loads, routes and delivery windows coordinated by phone.

Trips planned taking weight, schedules and backhaul loads into account; the final decisions stay with the driver.

Trips invoiced weeks later, reconstructing what was said and what wasn’t.

Every closed trip becomes an invoice or credit note, without Sunday accounting sessions.

Cakes and celebrations ordered by phone between one batch and the next.

Orders from chat and phone gathered into one production diary with dates and set-ups.

Quantities decided by eye: pastries left over, or bread gone by ten.

Production planned on sales and pre-orders, with reminders for proving and baking.

Ingredient labels and allergen sheets recalculated at every new recipe or batch.

Labels and allergen sheets generated from the recipe, updated at every change.

Surveys documented with photos and notes, transcribed in the evening.

Report generated from on-site dictation, with photos attached and the claim already linked.

Repair quotes and cost comparisons collected between e-mails and calls.

Damage assessed by comparing multiple quotes, with discrepancies highlighted.

Claim documents scattered across e-mail, fax and folders.

A file per claim, always complete and up to date, ready for the insurer.

Exercises adapted to each student, prepared in the evening for the next day.

Worksheets and tests generated from the syllabus and each student’s gaps, to review and explain.

Homework corrected one by one, between lessons.

A first automated pass with gaps classified: the judgement stays with the teacher.

Updates to parents written at month’s end, from memory.

Progress reports generated from lesson and test history.

Shifts and sites coordinated between calls and messages to supervisors.

Weekly work plans generated from contracts and availability, with changes replanned.

Quality checks reported verbally, not documented: disputes stay a matter of words.

Photo checklists per site, with outcomes comparable over time.

Multi-service contracts renewed without rereading what’s included.

Deadlines and clauses monitored: the renewal proposal starts from each site’s actual usage.