AI forecasting: revenue and margin in real time

Revenue, margin and utilization forecasts from your real data. Manual version included.

The limits of traditional tools

Classic tools for forecasting are full of forms, screens and configurations. Consultants lose hours on admin instead of working with clients.

Dozens of screens and tabs to learn
Initial setup that takes weeks
No AI that understands consulting context
Data scattered across many tools
Reports always days behind

One AI, one conversation

hice.ai handles forecasting like a chat: type what you want, AI acts on your real data. For those who prefer the classic approach, the traditional manual interface is always available.

«Show me the status of all open projects»
«Update X with what I'm telling you»
«Generate the monthly client report»
«Alert me if anything goes over budget»

Try it in chat

Examples you can ask:

Static demo of hice's AI chat.

What you get

AI chat in plain English

Talk naturally, AI understands jargon, abbreviations and context.

Real actions on data

AI doesn't just answer: it updates, creates, modifies, always with audit trail.

Total memory

AI remembers every conversation, decision and historical data point.

Native integrations

Calendar, mail, billing, databases: all connected.

Configurable automations

Triggers on events, conditions, deadlines: AI acts or alerts.

Conversational dashboards

Ask any chart in chat: no more complex report builders.

Multi-tenant

Per-client data isolation, granular permissions.

Full audit trail

Every AI action is tracked, reversible, exportable.

Smart notifications

Only notifications that matter, grouped by AI.

EU data & GDPR

EU hosting, encryption at-rest and in-transit, daily backups.

Fast onboarding

No weeks of training: a chat is a chat.

Traditional interface

For those who prefer forms, tables and classic dashboards: always available alongside AI.

What a consulting firm can genuinely forecast

Not everything in a consulting business is forecastable to the same degree, and treating it as if it were is what makes forecasts lose credibility. Committed backlog and recurring engagements can be projected with reasonable confidence. New logo revenue in a quarter cannot, and no model improves that much.

The productive approach is to forecast each layer separately and present them as separate layers: signed work, extensions with a history of renewing, weighted pipeline, and everything else. A leadership team that sees the four layers argues about the right one instead of about the total.

Signed backlog projected month by month
Extensions forecast from historical renewal rates
Weighted pipeline shown separately, never merged in
Explicit statement of what is not being forecast

Why margin forecasts drift and how to catch it

Revenue forecasts get attention; margin forecasts get built once and then quietly diverge. The drift comes from a small number of recurring causes: seniority mix heavier than sold, non-billable time growing, rate increases not applied, and overruns absorbed rather than negotiated.

Each of those has a leading indicator available weeks before it shows in the accounts. Tracking the four as a small set — rather than waiting for the month-end margin figure — turns a forecast miss into a series of small corrections, which is the only form in which margin is actually recoverable.

Planned versus actual seniority mix, by project
Non-billable share trend by team
Rate increases due and not yet applied
Overrun hours absorbed without a change request

Presenting a forecast people will use

A forecast that arrives as a single number invites a negotiation about the number. One that arrives as a range with the assumptions written down invites a conversation about the assumptions, which is the conversation that changes behaviour.

The practical format is short: the range, the three assumptions that move it most, and what would have to be true for the low end or the high end to happen. Keeping the same three assumptions visible month after month also makes the forecast improve, because the misses become traceable to a specific belief that turned out wrong.

A range with a stated confidence, not a single figure
The three assumptions that drive most of the variance
What would have to change to hit each end of the range
Comparison of last month's forecast against what happened

Consulting firm, 80 consultants

An 80-consultant firm managed forecasting with legacy tools: hours lost on admin every week. With hice.ai AI absorbs most repetitive work. Result: -50% admin time, +30% delivery capacity, managers more client-focused.

hice.ai vs traditional forecasting tools

Feature	hice.ai	Traditional tools
Interface	AI chat + classic	Only forms and menus
Native AI	Yes, acting agent	No / add-on
Languages	Multi (IT/EN/FR/ES)	Often EN only
Setup	Days	Weeks / months
Real-time data	Yes	Delayed reports
Real-time KPIs	Yes: utilization, margin, bench	Delayed reports
Learning curve	One chat	Long training