Documentation Index
Fetch the complete documentation index at: https://docs.utari.ai/llms.txt
Use this file to discover all available pages before exploring further.
What Bottleneck Diagnosis helps you do
Bottleneck Diagnosis cuts through the noise. Instead of producing a broad audit of every possible issue, it helps JeremyAI find the one constraint that is actually limiting growth right now, then tells you what to fix first. Use it when you want JeremyAI to look across the full growth system, connect the numbers, and tell you where to put attention first. That can include ads, lead quality, offer, pricing, landing pages, booking, show rate, sales consults, follow-up, revenue, margins, retention, delivery capacity, and broader operations. Bottleneck Diagnosis can work through a messy business situation, compare multiple possible constraints, and still drive toward one clear operating answer. Instead of sending a one-line question, give JeremyAI the context and ask it to diagnose the constraint:I need a bottleneck diagnosis for this business.Here is the offer, timeframe, traffic source, spend, leads, booked calls, shows, buyers, revenue, follow-up process, and any relevant notes.Compare the likely constraints across ads, offer, booking, show rate, sales, follow-up, economics, and operations.Tell me the most likely bottleneck, the attack point for the next 7 to 14 days, what we should not do yet, and what data would change your diagnosis.\
The worker may calculate stage conversion rates, rank competing issues, ask focused follow-up questions, and go deeper when the diagnosis points to a specific business area. The outcome should still be simple: one attack point, the evidence behind it, and the next actions to take.
When to use Bottleneck Diagnosis
Use this worker when:- leads are coming in, but sales are not following
- ad costs are rising and you are not sure whether ads are the real problem
- booked calls are not showing up
- consults or demos are not closing
- revenue is up, but profit or payback is weak
- your team wants to scale spend, but you are not sure the business is ready
- you are about to make a big decision, such as scaling spend, rebuilding a funnel, changing an offer, or hiring for sales
- you have multiple possible issues and need to know where to put attention first
Example use cases
Bottleneck Diagnosis can be used for different operating questions, not just one funnel review. Common use cases include:- Before scaling spend: check whether the business is ready for more traffic or whether downstream conversion needs to be fixed first
- Show-rate or booking problems: diagnose whether leads are failing to book, failing to show, or failing to buy after the consult
- Sales consult leaks: compare lead quality, offer fit, objections, close rate, follow-up, and payment friction
- Revenue without profit: look at CAC, margin, payback, refunds, retention, or fulfillment capacity before deciding to scale
- Weekly operating review: run the same diagnosis after new numbers are available so the team can decide what to attack next
Get the strongest diagnosis
JeremyAI can reason broadly out of the box. Bottleneck Diagnosis is useful when you want that broad reasoning focused into a practical operating review. To get the most out of the worker, do not treat it like a quick chat question. Treat it like you are handing JeremyAI the current state of the business and asking it to make the next operating call. Bring four things when you can:- The journey: how someone goes from first touch to money collected
- The numbers: spend, leads, booked calls, shows, buyers, revenue, and conversion rates
- The economics: price, AOV, margin, CAC, ROAS, payback, refunds, retention, or LTV when available
- The decision: what you are trying to decide, such as whether to scale spend, fix follow-up, rebuild a page, change the offer, or inspect sales calls
- calculate the key stage rates
- compare the likely constraints
- rank the issues by probability, severity, and downstream risk
- pick the one attack point for the next 7 to 14 days
- say what you should not do yet
- state its confidence level
- explain what data would change the diagnosis
Use it in rounds
You do not need a perfect data room before starting. Use Bottleneck Diagnosis in rounds:- First pass: give the best numbers and context you have.
- Evidence pass: add exports, screenshots, CRM notes, call recordings, transcripts, or Meta CLI output if the worker needs more proof.
- Decision pass: ask what to fix first, what to avoid, and what would change the answer.
- Follow-up pass: come back after the first fix with updated numbers and ask for the next bottleneck.
Before you start
You can start with a manual prompt. The worker does not need every integration connected to be useful. In the community worker view, use Integrations when you want Bottleneck Diagnosis to reason from connected data sources, and use Triggers only when you want a recurring diagnosis after the same data is updated. You do not need to change the worker’s internal setup to use it. For a stronger diagnosis, bring the clearest numbers you have:- the timeframe you want diagnosed, such as last 7 days or last 30 days
- the offer, price point, and business model
- traffic source and spend
- leads, booked calls, shows, buyers, revenue, and close rate
- AOV, gross margin, CAC, ROAS, payback, or repeat purchase data when available
- follow-up speed and channels
- screenshots, exports, call notes, or CRM data if the worker cannot access the source directly
Start a diagnosis
- Open JeremyAI.
- Go to Workers.
- Select Bottleneck Diagnosis.
- If you want the worker to use connected data, confirm the relevant integrations are connected or have exports, screenshots, or key numbers ready.
- Click Start Chat.
- Describe the business and paste the numbers you want diagnosed.
- Ask the worker to identify the biggest bottleneck and what to fix first.
Start with a strong diagnostic prompt
A strong prompt gives JeremyAI the business model, the timeframe, the numbers, and the decision you need to make. Use this structure:I need a bottleneck diagnosis for [business / offer].Timeframe:[Last 7 days / last 30 days / this launch / this campaign]Business model and offer:[What you sell, price point, target customer, delivery model]Goal or decision:[What you are trying to decide right now]Numbers:- Spend:- Traffic source:- Leads:- Booked calls / demos / checkouts:- Shows / completed appointments:- Buyers:- Revenue:- AOV:- Gross margin:- CAC / ROAS / payback:Follow-up and sales context:[Speed-to-lead, reminder flow, sales process, common objections, close rate notes]Data attached or connected:[Screenshots, exports, CRM, booking tool, payment data, call notes, Meta CLI output]Please diagnose the most likely bottleneck, rank the other possible constraints, pick the one attack point for the next 7 to 14 days, tell me what not to do yet, and explain what data would change your diagnosis.\
Shorter prompts can still work, especially when you are early. Good diagnostic prompts can also sound like:
- “Where is the constraint right now, and what should we fix before increasing spend?”
- “Is this an ads problem, a sales problem, a follow-up problem, or an offer problem?”
- “What is the one attack point for the next 7 to 14 days?”
- “What would make you change this diagnosis?”
Example prompt
Use this format when you have basic funnel numbers:I own a high-ticket med spa.Our main offer is a $2,500 body contouring and skin tightening package.We spend about $500/day on Meta ads. Last week we got 90 leads, 42 booked consults, 18 showed up, and 3 bought.Our team usually follows up with leads within 12-24 hours.I'm not sure if the problem is the ads, the offer, the booking process, the show rate, the sales consult, or the follow-up.Please diagnose the biggest bottleneck and tell me what to fix first.\
The worker should respond with a direct diagnosis, not a generic checklist:
The bottleneck is not lead volume. You are getting 42 booked consults from 90 leads, which is strong enough to diagnose downstream.The biggest leak is show rate: only 18 of 42 booked consults showed up.Do not scale spend yet. Attack confirmation, reminder, and speed-to-lead for the next 7 days before changing the ads.\
What the worker should give you
A useful diagnosis should include:- the likely bottleneck
- the math or evidence behind it
- a ranking of likely constraints by probability, severity, and downstream risk
- the one attack point to focus on first
- the next 3 to 5 actions
- what would change the diagnosis
- what data would increase confidence
- “What are you assuming that could be wrong?”
- “Which number most changes your confidence?”
- “What would make this an ads problem instead of a follow-up problem?”
- “What should we avoid doing until this bottleneck is fixed?”
- “Turn the attack point into a 7-day action plan.”
Connect useful data sources
Connect the smallest set of source-of-truth systems needed to diagnose the business. You do not need to connect every app. Do not connect apps just because they are available. Connect or provide the systems that own the numbers you want JeremyAI to reason from. Good data sources include:- paid traffic data: Meta CLI output
- website and funnel analytics: Google Analytics, Microsoft Clarity, PostHog, or Amplitude
- CRM and pipeline data: HubSpot, Salesforce, Pipedrive, Close, Attio, Zoho, or Dynamics 365
- booking data: Calendly, Cal, or Google Calendar
- revenue and payment data: Stripe, Shopify, Square, Zoho Books, or Zoho Invoice
- call and meeting data: Gong, Fireflies, Recall.ai, Zoom, or Google Meet
- follow-up and nurture data: Gmail, Outlook, Klaviyo, Mailchimp, ActiveCampaign, Brevo, SendGrid, MailerLite, or WhatsApp
- manual tracking and source docs: Google Sheets, Airtable, Google Drive, Google Docs, or Notion
Meta CLI data
For Meta ad data, use Meta CLI output for now. Useful Meta ad data includes:- spend
- impressions
- clicks
- leads
- CPM
- CPC
- CPL
- campaign or ad set names when relevant
- creative performance notes
How the worker stays focused
Bottleneck Diagnosis is already configured to keep the conversation focused on finding the constraint, ranking the likely causes, and choosing the first fix. You do not need to configure the worker before using it. Start with the business context, add the numbers or connected data sources you want it to use, and ask for the next operating call.Triggers
Triggers let Bottleneck Diagnosis run automatically. They do not change the worker itself or connect new data sources. A trigger needs its own schedule and prompt that tells the worker exactly what to do when it runs. Use a trigger only when the task is repeatable. For one-off diagnosis, use Start Chat instead. You can create a trigger from the worker’s Triggers tab or from the main Triggers area in Utari. For a scheduled trigger:- Click Create new.
- Choose Scheduled Trigger.
- Select Bottleneck Diagnosis as the worker.
- Give the trigger a clear name and optional description.
- Set the schedule and timezone.
- Choose the execution method.
- Write the trigger instructions.
- Create the scheduled task.
- the timeframe to review
- which connected sources or documents to use
- which numbers to calculate
- what output format you want
- where the result should be saved or sent, if applicable
- what the worker should do if a source is missing
Run a bottleneck diagnosis for the last 7 complete days.Use the connected CRM, booking, payment, analytics, spreadsheet, and Meta CLI output if available.Calculate the key stage conversion rates. Identify the most likely bottleneck, rank likely constraints by probability, severity, and downstream risk, and give one attack point for the next 7 to 14 days.If a source is missing, say what is missing, give the best provisional diagnosis from the available data, and list the data that would make the diagnosis stronger.\
Before turning on a trigger, run the same instruction manually in chat. After the trigger is active, check the first few runs and pause it if the output is incomplete, too broad, or running at the wrong time.
Event-based and webhook triggers should only be used when the event source or webhook is already configured for your workspace.
If the answer is not useful yet
You usually do not need support to improve a weak diagnosis. Give the worker better context and ask it to be more decisive. If the worker gives you several things to fix, ask it to pick one and explain why that one comes first. Try prompts like:Pick one attack point. Do not give me a list. Explain why this comes first.\
What are you assuming here, and what data would change your diagnosis?\
Use the numbers I gave you and rank the possible bottlenecks by probability, severity, and downstream risk.\
Use this Meta CLI output for the ad data.\
If the worker cannot access a source directly, paste the key numbers, upload an export, or share a screenshot with sensitive information hidden.
When to contact support
Contact support if:- Bottleneck Diagnosis is not visible after refreshing, confirming you are in the correct workspace, trying an incognito or private browser window, and clearing browser cache and cookies
- a connected integration is not available to the worker after reconnecting it
- a trigger is enabled and configured correctly but does not run
- you cannot tell whether the worker should have access to a specific data source