Lokara

Lokara · on the Anukara simulation engine

Rehearse the decision before your customers see it.

A hundred simulated people, each with a life, a budget and friends, decide what to do with your offer over days. Every decision is recorded with its reason. You read the report before you ship.

Request a pilot See a real run
The First 100 · a referral launch in Columbus Day 10 of 10
A · No reward 47 signed up
B · 20 dollar referral reward 38 signed up
Drawn from the real record of run 2, 28 October 2026: the same 100 people in six circles, two offers, 10 days, $8.03 of model cost. Each dot is a person; a ring is someone telling a friend. The reward did not help.

The problem

Surveys tell you what people say. Live tests cost you real customers.

A price rise, a launch offer, a policy change: expensive to reverse, and every way of testing it first misses something.

Survey

Stated answers, collected in days.

Little on who downgrades, leaves or comes back by the third renewal. Nothing on what a friend said in between.

Live test

Real behaviour, real revenue at risk.

Months before the result can be read, and the half of your customers who saw the worse option remember it.

Lokara

The same people, under each option, over time.

Simulated people decide at each step under explicit rules. Every decision is recorded. Branch from any day and compare.

How it works

Four steps from a question to a record you can read.

  1. 01

    Build the people

    Describe them to the assistant, set every trait yourself, or drop in a file you already have. Each person gets a persona, a routine, a budget and friends.

    A simulated person's profile in Lokara, with their traits and persona
  2. 02

    Write the method

    Say what the company does, what each person can decide, which decisions last, and how word travels. Two offers become two options of one study.

    A method set up in Lokara with its options
  3. 03

    Run it

    The same hundred people live through each option. Watch the days pass and the map light up. Click any person on any day for their reasoning.

    The run screen in Lokara: a map of people lighting up as they decide
  4. 04

    Read the report

    The numbers first, then a written report with quotes and evidence, compiled from the record rather than from a model's opinion.

    The top of a Lokara report with the compiled outcome

Evidence

Real runs. Real records. No forecasts.

Everything below comes from a finished run's record. The people are simulated; the numbers are what they did.

The First 100 · 28 October 2026

Gather launches in Columbus with a referral offer. Does paying for referrals help?

100 people in six circles, 10 days, two offers, the same people under each.

OfferSigned upEver told a friendTimes told
A · No reward 47 48 80
B · 20 dollar referral reward 38 47 77

Per 100 people. Model cost of the run: $8.03. Word spread about as far either way; nine fewer signed up with the reward, which the report calls a direction rather than a settled difference.

The Price Move · October 2026

Folio raises its Full plan from $20 to $25. Who stays, who leaves, and what if we back out?

200 subscribers, three renewals, two options, plus one fork that returns to $20 after the first renewal.

PathReceipts, 3 renewalsPaying at the end
A · Full stays $20$11,000193 / 200
B · Full rises to $25$13,080195 / 200
Fork of B · back to $20 after renewal one$11,790198 / 200

Receipts inside the simulation, not revenue. The fork shares B's first renewal; it is a branch, not a third cohort.

A real run, replayed

Watch the hundred people decide, day by day.

This is the record of run 2 of the First 100, played back on this page: the same 100 people in six circles, two offers, 10 days. Nothing here is live. Every count, name and reason comes from the record.

Period 10 of 10 · Wed 11 Nov · each period is one day

Speed

Universe · run 2

First-100 referral launch · Gather launches in Columbus

Gather organises dinners, games and other group activities with the people you already meet. Time runs left to right and each row is a simulation, a node per day.

2 simulations · 2 done · $8.03 model cost

Simulations · time in periods 12345678910 Replay · period 10 A No reward Done · period 10 of 10 47% signed up at the end B 20 dollar referral reward Done · period 10 of 10 38% signed up at the end a fork that could be made a fork that could be made a fork that could be made Press coverage Press coverage Drop the reward 12345678910 Replay · period 10 A · No reward 47% at the end B · 20 dollar reward 38% at the end
  • A simulation, a node per finished period
  • Period in progress
  • Not reached yet
  • Planned fork
  • A fork that could be made

The method declared 2 branch points. Pick one to see where it falls and what the record observed.

ANo reward

Day 10 · Wed 11 Nov

Launch post: “Gather is live in Columbus. Plan your group's next night out in a minute, with everyone's calendar in one place.”

Where the 100 people stand at the end of the day

  • Sign up 0
  • Tell a friend 0
  • Wait 99
  • Ignore 1

47 of 100 signed up so far

Decisions that day · a few of the 100, as recorded

  1. Tyler MabryPickup soccer group

    Wait

    “Wait. Contacts are still a deal-breaker, and chat works while the soccer group has not moved.”

    Also weighed Ignore, Tell a friend, Sign up, Read the forum

  2. Ben HatcherParents' playgroup

    Wait

    “I will wait another day because privacy and contact-sharing concerns still outweigh Gather's possible scheduling benefit. The Parents' playgroup has not clearly moved to Gather, so waiting remains the sensible choice.”

    Also weighed Sign up, Tell a friend, Ignore

  3. Alyssa BurkhartBook club

    Wait

    “I’ll wait for now. The scheduling part could help, but unclear pricing and contact handling are still a dealbreaker, fr.”

    Also weighed Sign up, Tell a friend, Ignore, Read the forum, Like a post

  4. Dylan PierceWeekend running club

    Ignore

    “Gather has not shown enough practical value to justify a permanent commitment, and its price and privacy hurdles are still deal-breakers for me. I’m going to ignore it rather than keep circling the same track.”

    Also weighed Wait, Sign up, Tell a friend

B20 dollar referral reward

Day 10 · Wed 11 Nov

Launch post: “Gather is live in Columbus. Plan your group's next night out in a minute. Invite a friend and you both get 20 dollars when they join.”

Where the 100 people stand at the end of the day

  • Sign up 1
  • Tell a friend 0
  • Wait 97
  • Ignore 2

38 of 100 signed up so far

Decisions that day · a few of the 100, as recorded

  1. Angela SuttonWeekend running club

    Wait

    “I would wait today because Gather's notification and contact practices remain unclear, while our established group chat is serviceable and the running club's continued waiting positions do not justify another recommendation.”

    Also weighed Tell a friend, Ignore, Read the forum, Like a post, Reply on the forum

  2. Kaylee BrandtParents' playgroup

    Wait

    “I mean, the useful part is still there, but the group is still cautious and nothing has eased my contacts-and-kids privacy line, so I’ll wait rather than recruit again.”

    Also weighed Tell a friend, Sign up, Ignore, Read the forum

  3. Jeff RoushChurch supper circle

    Ignore

    “I’m going to ignore Gather because the $20 referral still does not justify sharing contacts or changing a routine that already works well enough, and there has been no convincing practical benefit.”

    Also weighed Wait, Sign up, Tell a friend, Read the forum

  4. Travis WeaverThursday board-game night

    Ignore

    “Nope. The $20 reward does not fix the app and contact-sharing problem, and our routine works fine without Gather.”

    Also weighed Wait, Sign up, Tell a friend, Read the forum

The record

Every decision has a reason you can read.

Click a person on any day and see what they knew, who they heard from, what they decided and why. Then follow the thread to the people they told.

  1. 01 What they knew
  2. 02 Who they heard from
  3. 03 What they decided, and why
  4. 04 Who they told next

Readings, day by day.

Where everyone stood after each day and which decisions were taken. Click a person on any day for their own reasons, then follow the people they told.

From the First 100 record under no reward, where 47 of 100 people signed up.
Open the full screenshot of the readings
Readings of the First 100 in Lokara: an area chart of where people stood after each of ten days, and a bar chart of the decisions taken each day.

A report written from the record.

The report says what the result means for the decision, how sure it is and what the record cannot settle. Its numbers and quotes carry marks you can click back to the decisions they came from.

The written report of the same run.
Open the full screenshot of the written report
The written report of the First 100 in Lokara: a paragraph on what the result means for the decision, with numbered evidence marks after each claim and quotes from two of the people.

Forks

What if the local paper runs a story after day four?

The method declared a branch point, Press coverage: “At the end of day 4, if fewer than 20% have signed up, the local paper runs a story on the launch.” In run 2 the record observed 5% at the end of day 4, so it was met. A fork from there keeps the same people and days 1 to 4, and runs the rest again with the story landed. The replay above marks where it falls.

Days 1 to 4: same people, same history Press coverage, after day 4 Original: no reward, as run Fork: the paper's story from day 5 branch after day 4 Day 12345678910 Days 1 to 4: same people, same history Press coverage Day 1Day 2Day 3Day 4Day 5Day 6Day 7Day 8Day 9Day 10 Original: no reward Fork: the paper's story from day 5
A drawing, not a result: this fork of the First 100 has not been run, so it shows no outcome. Its condition was met in run 2, and the replay above marks where it falls. The Price Move above includes a fork that has been run.
  1. 01

    Same people, same history

    The days before the fork are shared, not run again.

  2. 02

    One change

    From the chosen day one thing changes, here the paper's story, and the remaining days run again.

  3. 03

    Compare with the parent

    Read the branch against the run it left.

The engine

Anukara

anukara.dev

Every Lokara study runs on Anukara, our own simulation engine. It gives each person a memory and a reason, keeps every decision on record, saves checkpoints to branch from, and meters every run. Built for Lokara first, and open to other applications through its SDK.

For builders, labs and investors. The same engine, through the same SDK.

Open Anukara
The Anukara site: a simulation engine for worlds of people who decide, with the recording of a real run

Today the people decide with a frontier language model. Our own model, built to act more like people, replaces it next.

Founder

Vighnesh Bheed

Saarbrücken, Germany

Ten years of engineering. Neural networks and digital twins for experiment design, computer-vision models for defect detection on industrial machines, and an agent-verification platform. Building Anukara and Lokara since February 2026.

Contact

Bring one decision.

A price rise, a launch offer, a policy change. We rehearse it on a hundred people and hand you the record and the report.

Request a pilot