Skip to content
Nolli
Request a demo

Platform · Decide

Know where AI belongs, and what it’s worth.

Every step of every workflow, sorted into delete, plain code, agent or human. Every opportunity ranked by value, priced per task, with a build, buy or skip call, before you spend.

Request a demoFor AI transformation leaders →
NolliOpportunities Sales
Updated today

Why it ranks here

Weekly revenue review prep

Same steps and the same sources every week. Only the recommendation needs a person.

How oftenEvery Monday
Time spent14 h a week
People involved6 people
Judgment neededLow on 4 of 5 steps

Every step, sorted

  1. Copy totals into slides by handDelete
  2. Pull closed-won dealsPlain code
  3. Collect account updatesAgent
  4. Reconcile conflicting figuresAgent
  5. Write the recommendationHuman

Nolli recommendsAgent prepares · you decide

Four buckets

Every step goes in exactly one bucket.

Delete it, run it as plain code, give it to an agent, or keep it with a person. Sorting the steps before choosing tools is how forward-deployed engineers avoid automating the wrong thing.

Delete
The step shouldn’t exist once the work is redesigned.

Waits, re-keying, chasing and redoing work. Remove it before you automate anything.

  • Copy totals into slides
  • Re-export with the right filter
  • Wait for a reply
Plain code
The step follows fixed rules and needs no judgment.

If this, then that. Cheaper, faster and more predictable than a model.

  • Pull last week’s deals
  • Send the deck at 11:00
  • Create the record
Agent
The step needs judgment, and you can check the result.

Messy inputs, many variations, and enough past cases to test against.

  • Read account emails
  • Reconcile conflicting figures
  • Classify an invoice line
Human
A mistake would be costly, public or hard to undo.

Approvals, negotiation, signatures and payments. An agent can prepare the work.

  • Approve the discount
  • Sign the contract
  • Submit the payment

Guide: The four buckets: when to use AI agents

Build, buy or forego

A clear call for every opportunity, with a price.

Every opportunity gets a call backed by evidence, and a value in hours and dollars, before anyone builds or buys.

Build, buy, or forego.

Every opportunity gets a clear call: build an agent, buy a tool, or keep it human.

  • BuildTicket triageSupport · $166k a year
  • BuyInvoice matchingFinance · $95k a year
  • ForegoNDA redlinesLegal · keep it human

Priced in hours and dollars.

See the hours each opportunity frees up every month, and what they’re worth a year.

$236ka year, from 3 workflows worth building

  • Ticket triage210 h a month
  • Weekly revenue review prep56 h a month
  • Deal desk summaries32 h a month

People’s time at a $66/hour loaded rate

Model choice

The right model for each task.

For each task, Nolli finds the cheapest model that passes your checks, whether frontier, small or open-weight, and shows the full cost per task, including people’s time fixing mistakes.

  • Cost per correct result, not per token
  • Small models for routine cases, frontier for hard ones
  • Re-checked as models and prices change
Guide: How to compare AI model cost per task
Route a support ticket3,000 tokens in, 200 out · a wrong route takes 5 minutes to fix
Frontier
97.4% right$0.183
Small
97.1% right$0.164
Summarize a one-hour meeting12,000 tokens in, 800 out · an edit takes 3 minutes
Frontier
98.3% right$0.216
Small
97.9% right$0.085
Reconcile conflicting revenue figures50,000 tokens in, 5,000 out · a wrong reconciliation takes 45 minutes to fix
Frontier
98.2% right$1.64
Small
88.4% right$5.82
  • Model cost
  • Fixing mistakes
Cost per task at Anthropic list prices. Blue is the model; orange is people fixing mistakes. Task values are illustrative. Pricing source.

AI spend

See what you spend on AI, and what it buys.

Seats nobody uses, tools that do the same job, and calls that could run on a cheaper model. Nolli shows where AI spend goes and where to save, next to the value each change creates.

  • Unused and overlapping AI seats
  • Model cost per task, per workflow
  • Savings found before renewals, not after

Questions leaders ask

Good to know

Something we haven’t answered? Email the founders.

How does Nolli decide where AI belongs?

It breaks each workflow into steps and sorts every step into one of four buckets: delete it if it shouldn’t exist, plain code if it follows fixed rules, an agent if it needs judgment you can check, and a person if a mistake would be costly or hard to undo.

How do you put a value on an opportunity?

From the work itself: how often it happens, how long it takes, how many people touch it and what their time costs, set against the cost per task of running it with AI.

Can Nolli tell us whether to build or buy?

Yes. Every opportunity gets a call, build an agent, buy a tool or keep it human, with the evidence behind it, before you spend.

Which AI models do you use?

For each task, the cheapest model that passes your own checks, whether frontier, small or open-weight. Nolli shows the cost per task for each.

Keep going

Go deeper

Next · ProveProve it on your own work.Agents run in shadow mode first, scored against what your people actually did. Then the same workflow is measured by people, people with AI and agents, on time, quality and cost.Explore ProveGuideWhen to use an AI agent: sort every step into four bucketsDelete, plain code, agent or human: a decision tree for every step, with real data and a worked example.GuideHow to choose an AI model by cost per taskMeasure cost per task, not price per token, and find where a cheaper model does the same job.

Start with one team

See it on your own work.

Nolli maps one function, ranks its opportunities and proves the best ones, before you scale.

Request a demoEmail the founders