From one message to a plan you can defend.

Oraqul is an AI agent that shows its working. It reads what you are working on, checks each assumption against real audience taste data, and ends with what to do first. Every claim comes with a receipt.

Agent traceExample

Reading the message: series, crime drama
Finding the comparable titles, Grey Tide and The Ferry Office
Looking up who else your audience loves
Ranking the UK, Canada and the US
Testing 9 assumptions in the plan
Writing the bottom line

Step 1: Describe it in one message.

Say what you are working on the way you would to a colleague. Oraqul works out the category and the job from what you write.

Text, files or both. Up to five files.
The hints are optional. Pick one, or just type.
The project is saved when you send.

What are you working on? Tell me the title and what you want to know.

Plan the release of Night Ferry, a Mexico City crime drama series, for the UK and the US. Comparable shows are Grey Tide and The Ferry Office.

Not sure how to put it? Start from:Make a planCheck my planFix my plan

Step 2: It states its assumptions and asks only when it must.

Oraqul says what it read and what it assumed, with an example of how to correct it. It asks a question only when a wrong assumption would waste the run: the title, the category or the job when they cannot be read, and, for a full plan, the budget band and the platforms, which decide what it may propose.

Type a correction any time. Only what it changes runs again.

Reading this as a new plan: series, crime drama, territories UK and US, plus Canada (assumed, you did not name it). Say “only the UK and US” to change it.

What is the budget band? I will only propose what it can carry.

Micro: under about $10kIndependent: up to about $100kMid-size: up to about $1MMajor: over $1MSkip, assume independent

Step 3: It tests every assumption against real data.

Each assumption in the plan becomes a question for the audience taste data. Code reads the values that come back and sets the verdict by a published rule. The model never sets a number.

Assumption in the planData lookupVerdict

Crime thriller fans are the corecall_8f3aHolds
The UK should open firstcall_2c71Breaks
Seoul for a second wavecall_b904Weak
Pre sales will cover the spendno callCould not test

Open in Mexico City and London, then Toronto. Move the UK to week three.

Medium confidenceTwo cities have thin data.

Do thisOpen in Mexico City and London in week one.View receiptAdd Toronto in week two.View receiptMove the UK to week three.View receipt

Step 4: It ends with what to do first.

Every answer opens with the bottom line, how sure Oraqul is in a word, and up to three steps. Each step links to its receipt. The full plan, every assumption and all the evidence sit one click down.

One plan for each place you open.

Home first, then the diaspora cities. Each territory gets its own call, its lead channels and its partners, and a call on something you chose keeps your choice and says how to handle it.

Night Ferry, territoriesExample

1. LagosHome marketLead on Instagram, TikTok, YouTubeOkayHigh confidence
2. LondonDiaspora cityLead on YouTube, Instagram, TikTokOkayMedium confidence
3. HoustonDiaspora cityLead on YouTube, Instagram, TikTokRiskyMedium confidence
4. AtlantaDiaspora cityLead on YouTube, Instagram, TikTokAvoidLower confidence

The craft around the data, written down.

Every category plans by a versioned set of launch rules, each with its source and date. You can read them, and each plan says which version it used.

Worked back from your date

Film plans run months. Music rollouts run six to eight weeks. Books send advance copies two to three months out. The calendar is written by code, and a beat already over is dropped.

T-26 to T-16Lock the key art and set up the sales and presale page.
T-12Release the first trailer and open advance screening sign-ups.
T-10 to T-6Book advance screenings with community groups in each territory.
T-8Open presales and read the first week against the target.
T-1Premiere in the home market; the other territories follow within days.

One signal to judge it on: advance screenings and presales. Your target, not a forecast.

Briefs and scripts, never ads

Oraqul writes what a team needs to make the creative. A person makes the final trailer, ad or image.

Lagos · InstagramHook one: open on the strongest image, in the first two seconds.Script, 25 seconds: the hook, who it is for, the proof, the ask and the date.Shot list: five shots, from the core moment to the title card.Key art brief: one image, the title, one line. Oraqul does not draw it.

Hooks, scripts, shot lists, key art briefs and caption packs, for one territory and channel, or for every territory when you ask for all of them.

Partners that fit the launch

Qloo finds the brands your audience likes in each category and city, and ranks them next to the city's own taste. Each partner shows its Qloo fit and whether the budget and who is launching can reach it. Price is always unknown. A flagship that the launch cannot reach is listed apart as a goal for later, and never written into the plan's moves unless you named it.

Tier A, FlagshipOnly for releases with scale.A stretch for independent, out for micro. Price: unknown.
Tier B, Category and integrationBanks and payments, food and drink, hospitality, telecom.A stretch for micro. Price: unknown.
Tier C, LocalCinemas, creators, diaspora businesses, community groups.Fits every budget. Price: unknown.

Then keep going, in the same project.

Ask for the next step

Check rival dates, draft a partner brief, test a what if or make the creative kit, from the chips under the message box.

Look at what you missed

Nearby markets and areas of your own cities appear under Worth a look, each with why, what the data does not show, a first test and what to measure. They stay beside your plan until you add one.

Run the campaign

The plan’s calendar becomes actions you edit and tick off, each with a guide to adapt. Set up a two-angle test, add your own figures, and Oraqul compares them, or says inconclusive.

Correct it in words

“Check my plan instead” runs a check. Naming new places plans again with them.

Bring your results

After it is out, add results by place or age band, with the period and where they came from. Oraqul finds what missed and writes a revised plan beside the old one. It says what happened, never why.

Check what changed

Ask again weeks later. Qloo and the web are asked again, and you see what moved, which pages say something different and which lines of your plan rest on them. The plan stays as it is.

Questions people ask.

Anything else, open the examples and click a receipt.

See the examples

Is Oraqul an AI agent or a dashboard?
An agent. You tell it what you are working on in plain words. It plans its own steps, looks things up in the taste data, tests what it finds, reviews its own draft against your request, and comes back with a plan. You keep talking to it in the same chat.
Where does the data come from?
Three places. Taste and place are measured in Qloo, cultural data on what real audiences like across film and TV, music, books, podcasts, games, brands and places. Dated market facts, such as rivals and release windows, are sourced from web research and link to their page. Every number and verdict is computed by code. The model writes the plan, and launch craft is labelled as such.
Can the AI make up a number?
No. Scores, verdicts and confidence are computed by code from the data returned in the same run. The model only writes the words around them, and an answer that names a title or place outside that data is rejected and rerun. Read the method.
What happens after the launch?
Add your results as totals by city or age band, with the period they cover and where they came from. Oraqul sets each place against where the plan put it, in code, and a fix writes a revised plan around who showed up. The old plan is kept beside the new one. Your figures are your own evidence, kept apart from the taste data: they show what happened, and Oraqul never states why as fact.
Can it tell me what has changed since my plan was made?
Yes, when you ask. It asks Qloo and the web again with your plan’s own brief and sets the answers beside what the plan saved: findings that moved, pages that now say something different, each with the day it was read, and the lines of your plan that rest on them. Something is called changed only when the plan had an earlier finding or fact to compare. Your plan is untouched until you ask for a revision.
Does it find markets I had not thought of?
Yes, and it keeps them beside your plan. Nearby cities that rank above your weakest, and the strongest area of each of your cities, appear under Worth a look with why, what the data does not show, a first test and what to measure. Nothing joins your plan until you add it: ask “What if we add Accra?”. If the evidence is thin, it says so rather than filling a list.
Can it help me run the campaign?
It turns the plan’s content calendar into actions in the project, each with a guide from the creative kit that you adapt. Edit an action, tick it off, and set up a two-angle test for any of them. Add your own figures and Oraqul compares them, saying inconclusive when they are thin or close. A revised plan gets its own set, and nothing is posted or scheduled for you.
How does it use the places I name?
Every city is read, not just listed. Oraqul asks Qloo what the audience for your comparable titles looks like there, ranks all your cities in one call so their scores compare, and gives each a verdict and its own plan. For film and music plans it also reads neighbourhoods near the strongest cells, as leads to confirm on the ground. Nearby cities that rank above your weakest appear under Worth a look. A place with too little data to rank is marked as such, with lower confidence, and your own cities are never dropped.
What is a receipt?
The record behind one claim: what was asked, what came back, and the rule that turned it into a verdict. In a plan, a claim about taste carries a small number that opens its receipt, and the Evidence tab lists them all.
Does it ask about my budget?
Once, if your brief does not say. It asks for a band (micro, independent, mid-size or major), or you can skip it and it plans for an independent release and says so. The band, and who is launching if your brief says (a solo author, a small team, an established company or a major studio), decide which partners and moves it may propose. A flagship brand is not part of the plan for a solo launch, a small team or a micro budget. It is listed apart as a goal for later, and never written into the plan’s moves unless you named it.
Can I set a target?
Yes. Every plan is judged on one signal for its category, such as advance screenings and presales for a film or wishlists for a game. Give your own number, or where Oraqul has a benchmark of its own it shows it, marked as a target to aim for. A target is a goal you set, never a prediction.
What is a plan for each market?
One plan for each place you open, called a territory: the home market first, then the diaspora cities, such as Lagos, then London, Houston and Atlanta. Each has its own call (okay, risky or avoid), lead channels, and the partners Qloo ranked there. WhatsApp is a sharing layer and never a lead channel. Your own cities are never dropped.
Can it check my release date against rivals?
Yes, with the rival releases and dates you give it, for example “Rivals: Harbor Lights on 2027-03-12”. The data has no release calendar, so the dates are yours; Oraqul measures how much of your audience each rival shares.
Does it make trailers or ads?
No. It writes the scripts, shot lists, key art briefs, caption packs and hook options a team needs to make them, for one territory and channel, or for every territory at once when you ask for all of them. A person makes the final creative.
Does it forecast sales or streams?
No. It measures audience taste, who likes what and where, and gives a direction, not a forecast. Sales, streams and ticket numbers are yours to bring, and when you add them Oraqul compares them with what the plan expected.
What if my title is not in Qloo yet?
That is expected. The data holds taste, not unreleased titles, so Oraqul does not look for yours. It describes the launch by what surrounds it: comparable titles, the people involved, themes and the audience you want, and asks the data about those. A clear match is evidence for the plan. A weak one is only read as a warning when it is compared with other titles in the same call.
What kinds of launch does it cover?
Film and series releases, music releases and tours, book launches, podcast seasons, game launches and single live shows or short runs. Each category has its own checks, its own calendar worked back from your date, and its own signal.
Does it work outside the US and UK?
Yes. It tests titles from any market and ranks cities like Mexico City, Seoul and São Paulo beside London, Chicago and Toronto. Where the data for a place is thin, it says so plainly and lowers its confidence.
What happens to my data?
Only public names, places and age bands go to Qloo, the taste data provider. Uploaded results are kept as totals, with personal columns dropped. Your work is visible only to your team. What data goes where.
Can it contact partners for me?
It drafts partner briefs and next steps for you to send. It never sends anything itself.

Start anywhere.
Leave with a plan you can revise.

Create account See the examples