ALSO
Field manual

The AI category guide

The AI-native company runs one leading pick per business function — not forty-seven dashboards that don't talk to each other. This guide maps each function to what "good" looks like and what to evaluate, so you can advise customers with conviction and route them to the right ALSO solution.

Find your constraintBrowse the AI catalogue
Marketplace›AI category guide
How to read this guide

Closed loops beat clever prompts

A toaster runs for a fixed time whether or not the bread burns. A thermostat measures, compares to a target, and adjusts until the gap is zero. Most AI deployments are toasters dressed as thermostats. For every function below, the same question decides whether a tool is real: does it measure its own output and improve, or does it drift silently until a customer notices?

Sales & GTM

Turn signals into pipeline without adding headcount

Read the guide

Customer Support

Resolve routine tickets at a fraction of the cost

Read the guide

Marketing

Personalise at account level, measure at revenue level

Read the guide

Finance

Close the books in days, not weeks

Read the guide

Legal

Treat every agreement as an asset, not a filed document

Read the guide

HR & Recruiting

Compress the hiring pipeline from weeks to days

Read the guide

Knowledge

One grounded answer instead of forty-seven dashboards

Read the guide

Meetings

Recover a day a week with structured notes that feed the loop

Read the guide

Eval & Closed Loop

The layer that separates an AI product from an AI demo

Read the guide

Sales & GTM

Turn signals into pipeline without adding headcount

What it is

The go-to-market function covers everything from identifying your total addressable market and enriching account data to running outreach, recording meetings, and coaching reps. AI-native GTM collapses what used to be four or five disconnected tools into one continuous revenue motion.

What "good" looks like

A closed loop: every outbound touch is measured against reply and conversion rates, and the system adjusts targeting and messaging automatically. Good GTM AI is proactive — it surfaces the next best action rather than waiting for a rep to ask — and it writes back to your CRM without manual data entry.

What to evaluate

  • Does it measure reply, meeting, and conversion rates — or just send?
  • How clean is the data enrichment and how many sources does it waterfall through?
  • Does it update the CRM automatically, or add another system to maintain?
  • Can it explain why an account or contact was prioritised?
Explore ALSO solutionsRun the AI Profit Diagnostic

Customer Support

Resolve routine tickets at a fraction of the cost

What it is

Support AI handles inbound customer conversations end-to-end — understanding intent, taking action across systems, and escalating only when a human is genuinely needed. The leading pattern is measured cost-per-resolution, not deflection theatre.

What "good" looks like

A thermostat, not a toaster. Good support AI tracks resolution rate and CSAT on every conversation, so quality is visible instead of degrading silently. Each unresolved or poorly-rated interaction becomes a labelled example that improves the next one.

What to evaluate

  • What is the measured resolution rate, and how is CSAT tracked per conversation?
  • Can it take real actions in your systems, or only answer questions?
  • How clean is the handoff when it escalates to a human?
  • Is pricing tied to resolved conversations, or to seats?
Explore ALSO solutionsRun the AI Profit Diagnostic

Marketing

Personalise at account level, measure at revenue level

What it is

Marketing AI spans brand-consistent content generation, account-level personalisation, and campaign production at scale. The winners tie creative output to attributed pipeline rather than to impressions.

What "good" looks like

Full-funnel attribution closes the loop: content and landing pages are personalised per segment or account, and performance feeds back into what gets produced next. Brand consistency is enforced so scale never means drift.

What to evaluate

  • Can output be attributed to pipeline and revenue, not just clicks?
  • How well does it hold your brand voice and guardrails at scale?
  • Does it personalise per account or segment, or just spin generic copy?
  • Does performance data feed back into the next round of production?
Explore ALSO solutionsRun the AI Profit Diagnostic

Finance

Close the books in days, not weeks

What it is

Finance AI covers spend management, the accounting and ERP layer, and treasury. The AI-native pattern automates reconciliation and logging so the month-end close shrinks dramatically and cash is put to work automatically.

What "good" looks like

Reconciliation accuracy is measured and high, and exceptions are surfaced for review rather than buried. A good finance stack turns a ten-to-fifteen-day close into a three-day close and keeps a continuous, auditable trail.

What to evaluate

  • What is the measured reconciliation accuracy, and how are exceptions handled?
  • How much of the month-end close is genuinely automated?
  • Does it keep a clean audit trail regulators and auditors will accept?
  • How does it connect to your existing banking and ERP systems?
Explore ALSO solutionsRun the AI Profit Diagnostic

HR & Recruiting

Compress the hiring pipeline from weeks to days

What it is

This function spans applicant tracking, AI-first sourcing and matching, and the operational HR backbone — payroll, IT, and people data. AI-native hiring shortens time-to-hire while keeping the process structured and fair.

What "good" looks like

Structured, measurable hiring: funnel metrics are tracked stage by stage, and sourcing quality is evaluated against outcomes rather than volume. The system reduces manual screening without hiding how candidates were ranked.

What to evaluate

  • Is the process structured and measurable stage by stage?
  • How transparent and auditable is candidate ranking and matching?
  • Does it integrate with your HRIS, payroll, and IT backbone?
  • How does it guard against bias and meet compliance requirements?
Explore ALSO solutionsRun the AI Profit Diagnostic

Knowledge

One grounded answer instead of forty-seven dashboards

What it is

Knowledge AI is enterprise search and grounded question-answering across your fragmented systems — CRM, tickets, docs, wikis, and drives. It sits at the same layer as network access: infrastructure everyone relies on.

What "good" looks like

Answers cite company-specific sources rather than hallucinating context, and permissions are respected so people only see what they should. A good knowledge layer connects to where your work already lives instead of forcing another silo.

What to evaluate

  • Does it cite sources and respect existing permissions?
  • How many of your real systems does it connect to out of the box?
  • How does it handle stale or conflicting information?
  • What are the security and data-residency guarantees?
Explore ALSO solutionsRun the AI Profit Diagnostic

Meetings

Recover a day a week with structured notes that feed the loop

What it is

Meeting intelligence captures conversations and produces structured, shareable notes and action items that flow into your CRM and knowledge layer — turning ephemeral discussion into durable, searchable context.

What "good" looks like

Notes are structured and actionable, not raw transcripts, and they feed the broader system rather than sitting in a silo. Privacy is handled deliberately — capture and storage choices are explicit, not an afterthought.

What to evaluate

  • Does it produce structured notes and actions, or just a transcript?
  • Do outputs flow into your CRM and knowledge tools automatically?
  • How is audio captured and stored — and who controls it?
  • How accurate is it across accents, languages, and crosstalk?
Explore ALSO solutionsRun the AI Profit Diagnostic

Eval & Closed Loop

The layer that separates an AI product from an AI demo

What it is

Evaluation is the measurement and quality layer under every other AI deployment: tracing, automated scoring, human review, and quality gates in your delivery pipeline. It is what makes the closed loop actually close.

What "good" looks like

Every production output can be scored, and every bad one becomes a test case that prevents the next failure. Quality gates sit in the delivery pipeline so regressions are caught before customers see them — output, measure, error signal, adjust.

What to evaluate

  • Can it trace and score real production outputs, not just test sets?
  • Does it turn failures into reusable, labelled test cases?
  • Does it gate quality in your delivery pipeline?
  • Can it combine automated scoring with human review?
Explore ALSO solutionsRun the AI Profit Diagnostic

Not sure where to start?

The AI Profit Diagnostic pinpoints the single constraint holding a customer back — so you lead with the function that moves the needle first, then build out the rest of the stack from the ALSO catalogue.

Find your constraint
ALSO
© 2026 ALSO Holding AG. All rights reserved.
Privacy PolicyTerms of ServiceContact