90 AI Technologies
SV

Meet Nitti · Launching in Sweden

The best politicalconsultant you’venever heard of.

Brief an AI agent like a colleague. Nitti researches the issues that matter to your organisation, connects the evidence and delivers according to your instructions.

See what Nitti can do

Surpassing human abilities in most areas

What would you put Nitti to work on?

A month of reading, overnight.Sources included.

Brief Nitti on the report you need and it writes it. Every claim leads back to the paragraph it came from, and the report says nothing its sources don’t.

8m 04s · 1.9M tokens · Done. 143 responses, about 1,900 pages, the length of the Millennium trilogy: read, analysed and written up as a report you can digest, with a clear index and a source behind every claim.

Consultation analysis

The wind permitting reform: who is for it, who is against, and why

143 consultation responses, read against Norrsund Vind’s position · Fictional reform, illustrative analysis

In short

Almost everyone wants faster permits. The argument is about who gets to say no to a wind farm, and who pays for connecting it to the grid.

Of the 143 responses, eighty-three support the reform as a whole and forty-one reject it. Nineteen support the aim but object to one part, most often the municipal veto or the new grid-connection fee.

For Norrsund Vind the picture is good on timing and mixed on money. The fixed time limits have broad backing. The connection fee does not.

Support for each part of the reform
Fixed time limits
118
Municipal veto, given early
62
Grid-connection fee
39

What the reform proposes

Three things. A wind permit must be decided within a fixed time limit. A municipality keeps its say, but must give it early rather than at any stage. And new projects pay a fee towards their grid connection.

Who supports it

The Green Party backs the reform and wants the municipal veto replaced with a right to local compensation.

The Energy Producers’ Association welcomes the fixed time limits, which it calls the largest single uncertainty in a wind project today.

The Association of Northern Municipalities supports the reform on one condition: that the municipality hosting a wind farm gets a share of what it earns.

Who is against it

The Sweden Democrats reject the reform and want the municipal veto kept and strengthened.

The Landowners’ Federation supports its members’ right to host turbines, but cannot accept the proposed cap on lease payments and rejects the reform as drafted.

The Nature Protection Alliance opposes the fixed time limits. Its argument is that an environmental assessment takes the time it takes, and a deadline will cut it short.

Where the disagreement really is

Hardly anyone argues against faster decisions. The disagreement is about two things: whether a municipality can stop a project late in the process, and who pays for the grid.

On the veto, the responses split along party lines. On the fee, they split by geography: respondents in the north say it would stop projects there, respondents in the south say the grid has to be paid for by someone.

What this means for Norrsund Vind

The company’s main ask, predictable timing, has support from across the table, including from some who reject the reform as a whole.

The connection fee is the risk. It is the one part that the company’s natural allies also oppose, which makes it the point to press on.

Suggested next steps

Line up with the Energy Producers’ Association on the fee. Offer the Association of Northern Municipalities the revenue-sharing model it asked for. And put the fixed time limits, which almost everyone wants, at the front of every conversation.

Click a claim, or its number, to see the paragraph it came from.

Illustrative example · Norrsund Vind is an invented Swedish wind power company that develops and runs onshore wind farms in the north. The documents, the findings and the organisations that answer are invented with it; the parties named are real, the positions given to them here are not.

Brief it like a colleague. It keeps working.

Write to Nitti the way you write to a colleague: what to watch, how much detail you want and when to interrupt you. It works while you don’t, and it reaches you the way a colleague would: an email for the week, a text when something can’t wait, a call* when it really matters. You decide which is used for what, and who else should hear.

* Phone calls are in early testing.

Illustrative example · Norrsund Vind is an invented Swedish wind power company that develops and runs onshore wind farms in the north. The documents, the findings and the organisations that answer are invented with it; the parties named are real, the positions given to them here are not.

The position. The history. The context.

Connect statements over time with the roles a person held when they made them. Ask a question once, or make a meeting brief part of an ongoing assignment.

“Hi Nitti! Brief me before our meeting. How has this person’s position on wind power changed, and which public offices have they held?”
  1. 2014–2018

    Municipal councillor

    Voted against a wind farm in the home municipality.

  2. 2018–2022

    Member of the Riksdag

    Proposed revenue sharing for host municipalities.

  3. 2022–2026

    Committee member

    Backs fixed permitting times, on the condition of local compensation.

Illustrative career and positions; no real person is represented.

Thousands of motions. A clear place to start.

When motions arrive in the Riksdag, Nitti reads every one of them against your brief, not just the ones filed under your subject. A proposal that touches your issue from somewhere else is caught too. See the proposals that support your goals, the ones that challenge them, and what deserves a closer look.

10m 12s · 13.6M tokens · Done. 4,218 motions read against the brief and placed. Most do not concern Norrsund Vind and sit to the far left. Six marked as examples: hover or tap one. In the product every dot opens the motion behind it.

The motions, through Norrsund Vind’s lens

Same position ↑Opposite position ↓← Not relevantRelevant →

Agent assessments against the brief, not judgements about people. The parties and the committees are real; the motions, the members named and the positions given to them are invented for the example.

2 minute read

A historic leap.

Information does not become useful merely because it is collected. It must be understood in relation to other information, judged and, when necessary, acted upon.

Each day, the public record gains thousands of new pages. No reader could reach its end, it’s too vast and ever growing. For most of history, this was an unavoidable limit. We built libraries and institutions to make the record available. They helped us find what we knew to look for but no human could read and analyse the whole.

Now, for the first time, another kind of reader exists. Machine intelligence can move through an archive at a scale no person or team could ever match. It can retain what it finds, trace connections across time and continue reading as the record grows.

However intelligence alone is not enough. It needs a system around it: a way to gather the data, give it structure, preserve its sources and turn what it finds into work people can trust.

We built that system.

Our agents can move through the expanding public record, connecting documents, people and events across time. They return with reports, summaries and conclusions, each claim traced to the paragraph from which it came.

The archive has no end. For the first time, neither does the reader.

How it works

The path from information to action.

  1. Step 1: The system fetches

    Fetches

    The system fetches every new document in the public record, from motions and consultations to transcripts of debates, the moment they are published.

  2. Step 2: The system structures

    Structures

    The system builds a mirror of the political discourse, giving structure to millions of documents, decisions, speeches and more.

  3. Step 3: The system enriches

    Enriches

    The system enriches every item with metadata: categories, the people and organisations mentioned, geographical data, and technical metadata such as reference ids for billions of paragraphs.

  4. Step 4: The system makes it available

    Makes available

    The system makes the enriched data available to agents and humans alike: searchable, filterable and ready to be read and analysed by our agents and algorithms.

  5. Step 5: You filter and sort

    Filter and sort

    You narrow the scope for an agent with filters, categories and metadata, before you give the system and the agents their instructions for the work ahead.

  6. Step 6: You instruct the agents

    Instruct

    You instruct the agents: what you want them to look for, and how to act when they find something relevant.

  7. Step 7: The agents monitor

    Monitor

    The agents monitor and read every new document, gathering what they find over time and delivering daily or weekly reports and summaries. Tell them what is critical and they will do the reasoning, decide when to skip the schedule and bring the information to you instantly, by mail, SMS or a phone call (yes, they can speak).

  8. Step 8: The agents deliver

    Deliver

    The agents deliver reports, analyses, compilations, summaries, visualisations and updates, each claim traced to its paragraph.