A pipeline that builds itself around your sales goal.

Five things: the pipeline a search hands back, the system that produces it, the source territory it navigates, the scout itself, and the workspace around it. Every material claim carries the evidence behind it.

See how the scout works See a sample run

From brief to checked pipeline — in one place.

The search hands over a pipeline, not a spreadsheet. Every lead arrives with the evidence behind it, the signal that makes now the moment, and a confidence number you can argue with. Learning from the outcomes you mark is planned, not built: the figures below show what it is for.

Your pipeline, last 90 days, 9 searchesConfirmed outcomeIllustrative

Leads found	Qualifying	Contacted	Replied	Meetings	Won	Expanding
1,284	612	410	126	48	11	4
across 9 searches	48% of previous	67% of previous	31% of previous	38% of previous	23% of previous	36% of previous

Leads, sorted by fit

Fit is measured against your criteria — not a vendor's generic score. Click a lead.

Showing Nordpump Industri AB.

Nordpump Industri ABThe role to reach: Hållbarhetschef

96%

Fit, 4 sources

Why this lead

In CSRD scope from FY2026Register and filing0.98 confidence
Head of sustainability owns CSRD reportingInferred. From your playbook0.90 confidence
No reporting tool visibleInferred. Absence recorded on 4 sources0.71 confidence
Not an existing customerYour exclusion list1.00 confidence

What we could not stand behind

Works with a reporting consultancyNo dated source
Direct line not publishedThe switchboard is the route

Signals

Regulatory obligation, strongFY2026
Hiring ESG analyst, strongReason to call.11 days ago
News mention, moderate5 weeks ago

Competitor view

No footprint found. 3 sources checked, including two competitor case-study pages.

Illustrative leads at invented companies. A public scout shows the role to reach, never the person.

Planned: analytics built on your outcomes, not a benchmark

Illustrative demo data — not a Phiuto performance claim.

Conversations and meetings per weekIllustrative

04812RepliesMeetingsWeek 1Week 12

Illustrative. Meetings 8× in 12 weeksResearch time −11 h/week vs. your own baseline

Cost per qualified leadThe number that decides whether this is software or services.Illustrative

Run 1€18.40
Run 5€11.20
Run 9€7.60

Falls because the companies are already resolved, the sources are already understood, and your model already knows what a good lead looks like.

What's working for you, with sample sizes

Illustrative

Recent sustainability hiren = 6338%
Sustainability lead as first contactn = 14229%
No competitor footprint visiblen = 8824%
First report due this yearn = 5121%
CFO as first contactn = 979%

Planned: when a pattern is strong enough, Phiuto will propose it as a change to your search — “your won deals skew to companies that hired into the function in the last six months; shall I weight that?” You will confirm it or dismiss it. Nothing will change behind your back.

One lead, as the planned product would follow it

Illustrative

Day 0
Opener draftedFact-based, evidence-linked, rewritten by you. Phiuto has no send button.
Day 0
Re-checked before contactRole still current, company still in scope — verified the morning you reach out.
Day 4
Follow-up proposedDrafted only because a new signal landed: they published the report.
Day 9
RepliedMarked in the pipeline — the first real outcome data on this segment.
Day 12
MeetingBuying group updated: CFO added as sponsor from a public appointment notice.
Day 54
WonWritten back to your model: this profile, this signal, this persona, this approach.
Day 90
Expansion signalThey opened a second site. Same engine, now working for customer success.

The figures here are illustrative. Measuring them on your own leads, outcomes and market is planned, not built. A run that produces no reusable knowledge is a run we ran badly.

A qualified lead is not one API call.

Between “this is what I sell” and a company worth calling sits a coordinated system of specialised components — each one a real engineering problem. The agents are not the product; the system that drives them toward a sales goal, learns your business along the way, and can prove how it got there, is.

Simulated trace. Illustrative figures.

Run 1Your first search

Run 2Same seller, with reuse (planned, not built)

Orchestrator
Owns the sales goal, the budget and the stopping rules. The only component that can decide to stop.
Goal accepted, public sources only
Goal accepted, public sources only
Seller model
Reads your website into a seller model and testable lead criteria. Every assumption is surfaced, never guessed.
6 criteria compiled from your website
Seller model reused
Source intelligence
What each source holds, how it’s accessed, on what legal basis, what it costs, when it breaks.
34 candidate sources
Skipped
Strategy
Scores search paths on yield, fit quality, freshness, cost, access and compliance.
Rung A: register and websites
Re-plan to rung B: filings, job posts
Starts at rung B, from memory
Acquisition & extraction
Registers, company websites, press, filings and job posts, turned into structured claims.
412 companies, 63 tested
+7 companies, 22 signals
473 companies, 96 tested
Company resolution
One company is one record (org.nr), with the evidence attached.
397 companies resolved
84% of companies reused
Verification
Cross-checks the company fit and the signals against independent sources. Circular corroboration is detected.
34 corroborated
Corroborated
Corroborated
Goal evaluator
Sorts every lead into qualifying, company_only, disqualified or undecidable, and says “unattainable within constraints” out loud.
34 of 50, not satisfiedCriterion unmet, re-plan.
41 of 50, deliver and gap report
48 of 50, satisfied first attempt
Memory
Every run writes back: the seller model, strategies, failures, costs, source behaviour, resolved companies. Learning from the outcomes you mark is planned.
+6 sources, +397 companies, seller model v3
Reuse up

Complete, memory updated.

Complete. Fewer steps.

Non-negotiables, enforced in code

Provenance is part of the record, not a log.
Confidence is computed from corroboration — never asserted by a model.
Licence terms and personal-data limits are checked in code, at acquisition and at delivery.
Budgets are hard ceilings — a run stops at its limit, never past it.
Every run keeps its evidence trail — auditable end to end.
“This company could not be tested from public pages, because Y” is an output type, not an exception.

The models are rented. The seller models, the source intelligence and the company graph are the part we are building to keep. Reusing them from one run to the next is planned, not built.

Where the companies are, as a territory our agents navigate.

Every source we've ever touched — registers, company websites, press, filings, job boards — is a node in a graph, connected by what links to what. Agents traverse it in parallel: reaching a node probes it, a page that names the right role marks it a hit and opens its neighbours, a dead end is recorded so no agent wastes time there again. A simulated traversal is below.

Source intelligence graph

Illustrative. A simulated traversal.

Agent
Hit
Probed
Dead end
Unvisited
Path taken

Agents: 6
Sources probed: 120
Hits: 46
Dead ends mapped: 15

The graph is meant to record which sources carry the evidence for which kind of company, what they cost, where they fail. It holds companies and sources, never people: that is the part built to last. Carrying it from one run to the next is planned, not built.

A scout that shows its work.

The scout turns a seller's brief into a checked shortlist: discovery, verification, the evidence behind each lead, and the reason to reach out. Below is the shape of one result; the live scout produces it from your own website.

See how the scout works See a sample run

What you get back

Nordpump Industri AB: the role to reach is the head of sustainability. The company falls inside CSRD scope from its 2026 financial year, and eleven days ago it advertised for an ESG analyst. One claim was cut.

Illustrative lead. Invented company. A public scout names the role to reach, never the person.
1From your playbook. The role that owns CSRD reporting at a company like this.
2Checked. Public company register and the annual filing.
3Reason to call. A job post, eleven days old.
4Can't verify. No dated source for it. Cut before delivery.

Illustrative. Blue is checked against a public source, the highlighter is the reason to call, and red is cut because nothing backs it.

The workspace around the scout.

Lead Engine is where the scout is heading: account intelligence, buying committees, territory whitespace, sequences, pipeline and exports prepared around evidence. Today it is a preview on fictional data, not a live feature.

Open the workspace preview All examples

The scout opens soon as a public beta.

Anyone will be able to try it, and it will be a beta. Runs can be slow, and coverage is uneven: public company registers are searched in Sweden, Norway and Finland today and job ads in Sweden. Elsewhere the scout works from company websites and the open web.

Open to everyone, soon
No account and no login will be needed. You will paste your company's website and press the button.
Every source visible
You watch it search, page by page, and keep the list of everything it read. A claim it cannot back with a source is cut, not padded.
Companies and roles
Public scouts return the companies that fit and the role to reach at each. Never a named person. A search built with the founder can go further.

See how the scout worksMessage the founder on LinkedIn to hear when it opens.

Tell us what you sell and who you sell to.

We are early, and will start with a small number of teams — B2B sales, GTM and revenue teams, agencies, founder-led sales.

See how the scout works Talk to the founderPrefer to talk first? Message the founder on LinkedIn. We build the first search with you. That search can go further than a public scout.