Recipes & installable blueprints

A complete hiring pipeline in two forms and one workflow. HR posts a role once; candidates apply through a public link. Every application is scored against that role's actual job description, and the outcome happens immediately, qualified candidates get an invite with the hiring manager's Calendly link and book their own slot, everyone else gets a polite decline. Either way, the hiring channel sees a one-line summary. Nobody is left waiting three weeks for an answer.

A signup form captures the customer. From there this recipe runs their entire first week (a 6-email onboarding plan, daily follow-through, and automatic escalation the moment something's overdue) so no customer falls through a CSM's inbox again.

End-to-end newsletter operations: a signup form collects subscribers, a scheduled workflow sends campaigns via Gmail, and unsubscribes update your list automatically. Once it's running, the only manual step left is writing the email.

A daily workflow scrapes every competitor's pricing page, asks TinyGPT to tell real pricing changes (price moves, plans added or removed, feature changes, trial or usage limits) apart from design and copy noise, then updates your tracker and emails the process owner. One alert per real change. Nothing else.

Submit a topic and this recipe runs the whole pipeline: six AI-suggested angles, a dedicated writer per platform, matching hero images, every draft saved to a table, then a single approval gate. Approve once and each draft goes to its own channel. Reject once and nothing posts.

Discover influencers, enrich their contact details, research each one with AI, and run a personalized multi-step outreach sequence, all from one TinyTable. Pick a niche at install and the recipe finds matching creators, enriches every row in one call (email, website, LinkedIn, X, Instagram, YouTube), and has TinyGPT research each creator and draft a personalized pitch. A daily workflow then sends the sequence (initial outreach, two follow-ups, a final touch) and stops the instant anyone replies.

Every day this recipe visits the competitor websites you've configured and runs a web search on each one, looking for announcements, articles, blog posts, and press releases. AI keeps what actually matters (launches, pricing moves, funding, partnerships, acquisitions) drops the hiring posts and campaign noise, and writes a short executive summary of what changed and why it matters to you.

Most content pipelines lose posts in the gaps, an outline nobody reviewed, a draft waiting on an editor who never saw it land. This recipe closes every gap: an idea becomes an outline, the outline gets a decision, the draft gets written and reviewed, and only an approved post reaches your publication. Every stage has a named owner and nothing advances until they act.

Add a brand you want to work with. The company gets enriched, AI writes a genuinely tailored first email from real data, and a daily sequence follows up until they reply. A reply stops the sequence immediately, gets categorized, and (if they're interested) lands in front of your partnerships team.

Campaign approvals rarely fail because people disagree. They fail because the request is buried four replies deep and the launch date arrives before the answer does. This recipe puts every proposal on one form, hands the approver a link they can act on in seconds, and writes the decision and who made it onto a row you can point at six months later when finance asks who signed off on the spend.

A teammate drops a rough idea. AI polishes it into a platform-ready caption with hashtags, the owner approves by email or Slack, the publish date is blocked on your shared Google Calendar, and a daily run publishes what's due to LinkedIn and recaps it to the team.

A shopper submits a support form. Before any human opens it, AI has read it, assigned a category and priority, and picked the owning team from your list. A real helpdesk ticket is opened, its link saved back to the row, the right Slack channel is notified (urgent ones flagged) and the customer gets an instant confirmation telling them who's handling it and when to expect a reply.

Post-purchase feedback usually dies in one of two places: nobody asks, or somebody asks and the replies pile up unread. This recipe closes both gaps. The moment an order is marked Completed, the customer gets a survey link. When they answer, AI reads the response, labels it, and pushes it straight to the team that owns it (Product, QA, or Support) while praise gets quietly set aside for your testimonials. Leadership gets the trend summary weekly without asking for it.

A shopper files a return with photos. The recipe pulls the real order from Shopify, and AI weighs the request against your return window and policy, then stops and hands it to a person. Nothing is auto-decided. Your reviewer sees the request, the photos, and the AI recommendation in one place, makes the call, and the customer is emailed the outcome either way. Approved refunds can be issued straight through Stripe.

A customer files a warranty claim with proof of purchase and photos of the issue. The recipe confirms the purchase in Shopify, and AI weighs the claim against your warranty window and coverage rules, then stops. Your warranty team is pinged in Slack, sees the evidence and the AI's suggested resolution in one place, and approves or rejects. The decision is written back to the sheet and emailed to the customer every time.

One form replaces the enrollment email thread. A student picks a course and submits; the recipe checks that course against your catalog's real capacity and either confirms the seat (incrementing the count and putting the schedule on the student's calendar) or waitlists them and tells the registrar the course is full. The right email goes out either way, without anyone in the office writing it.

Admissions committees don't need a faster decision, they need less time spent getting to the decision. Applicants submit their transcripts and statement of purpose through one form. An AI pre-screen returns a fit score, a recommendation, and a short summary, all written back onto the application row. Your officer opens a secure Slack review with the summary and the documents in one place and clicks Accept or Reject. The matching email is sent immediately. The AI has no vote.

Students tell you the truth about a course only when they're certain it can't be traced back to them. This recipe collects no identifying field at any point, so there's nothing to trace. AI reads each submission, flags anything abusive or off-topic before staff ever see it, and tags sentiment and themes onto the row. When a rating is low or the sentiment turns negative, the department head is alerted in seconds: with the course, the instructor, and the comments, and nothing about who wrote them.

The gap between admitted and enrolled is where institutions lose students, and it's almost always administrative. This recipe closes it in one pass. The student uploads their photo ID, transcripts, and health form once. Their record is created, a checklist card appears for student services, orientation is scheduled and the invite sent, the branded welcome pack lands in their inbox, and their program advisor is pinged, all before anyone in the office has opened an email.

Contracts get stuck in two places: waiting on legal, and waiting on whoever legal is waiting on. This recipe makes both stages one tap. A requester submits the contract with its key terms and the draft file. AI rates the risk (auto-renewal, uncapped liability, indemnification, exclusivity, missing terms) so neither reviewer is reading cold. Legal clears it, management signs off, and the requester is emailed at every stage, including which stage sent it back.

Most expense claims are routine, and most manager time spent on them is wasted. This recipe uses TinyVision to actually read the uploaded receipt (merchant, date, amount, category) then has AI cross-check that against what the employee typed and flag anything out of policy. Claims under their department's limit with no flag are approved automatically. Everything else goes to a manager with the summary already written. The employee always hears back.

Document collection is a chasing problem, not a filing problem. A client uploads their tax forms, invoices, and bank statements through one secure form. AI compares what arrived against your required-document checklist for that purpose and returns complete or incomplete with a specific list of what's missing. Complete packages ping your accounting team and confirm receipt to the client. Incomplete ones email the client the exact list to re-upload, no one on your side writing that email.

Approvers shouldn't be the ones spotting that an invoice was already paid last month. A submitter uploads the vendor invoice and types its key fields. AI validates them (missing data, amounts over your threshold, line items that don't reconcile, possible duplicates already on file) and flags what it finds rather than guessing. The approver gets the details, the flags, and a link to the file, and taps once. Approved invoices notify accounting and can be booked straight to QuickBooks.

A real approval chain, not a form that emails somebody. An employee submits what they want to buy, from whom, for how much, and why. Their manager decides first. If the amount hits your finance threshold, a second approval routes to finance, genuinely sequential, not a copied-in email. Once it clears, procurement is told to raise the PO and the requester gets the green light. Rejections come back with the reason so they can revise instead of re-asking.

Vendor onboarding fails on the same thing every time: an application arrives missing one document, and someone has to notice, then write the email, then wait. This recipe notices instantly. A prospective supplier submits their company details plus tax form, insurance certificate, and banking document. AI confirms all three are actually attached and the details look complete. Complete applications go to finance for a one-click approval; incomplete ones get chased automatically. Approved vendors are activated and welcomed.

Employees answer honestly only when anonymity is structural rather than promised. This survey never asks for a name, an email, or any identifying field; rows carry a timestamp and nothing else. On your schedule, an AI Feedback Analyst reads the whole batch and returns headline metrics, recurring themes, and a prioritized set of improvement areas. And it checks batch size first: below your minimum, no digest is generated at all, so a five-person team can't be reverse-engineered from a summary.

Leave lives in email right up until the morning two lecturers from the same department are away and nobody saw it coming. This recipe gives faculty one form, sends the department head a request they can decide in a tap, emails the decision straight back, and leaves you with a full term of leave history on a single sheet instead of scattered across four inboxes.

Employees stop referring when their last referral vanished without a word. This recipe fixes the specific thing that kills the program: the referrer is notified automatically at every stage, from submission to hire. Along the way, AI writes a neutral fit summary so recruiting can triage faster (advisory only, never a score or a decision) and a Hired update marks the bonus eligible and celebrates it in the recruiting channel, which is the part that makes the next referral happen.

Onboarding fails in the gap between offer accepted and day one, and it fails the same way every time, somebody forgets a step. This recipe turns that whole stretch into one submission. HR enters the new hire once; their task card is created, their welcome email goes out with a document-upload link, and the team is told who's starting and when. When the documents come back, a second workflow matches them to the right record and marks them received. The tracker sheet always shows exactly where each hire stands.

Offer letters carry the most risk and the least process. This recipe adds both gates. A hiring manager submits the candidate, role, and compensation; AI drafts the letter body from those facts and nothing else. It never invents a salary, a benefit, an equity number, or a date. HR reviews and can edit before approving; a rejection there stops the flow immediately. Approved offers go to management for a second sign-off, and only when both land does the candidate receive anything.

Review cycles do not collapse because managers refuse to write reviews. They collapse under the admin around them: chasing self-assessments, remembering who owes peer input, and stitching three documents together per employee. Upload your employee list once and this recipe runs the whole cycle: inviting each stage in order, summarising every submission as it lands, and keeping one row per person that shows exactly who is outstanding.

Mandatory training is only a problem at audit time, which is exactly too late to fix it. This recipe keeps the register live. HR assigns a course with a due date; the employee gets the link and a one-click completion form. When they complete it and upload their certificate, the assignment closes and the reminders stop on their own. A daily sweep finds everything still open, escalates anything past due to L&D, and keeps nudging until the register is clean.

Phone tag is the most expensive thing in a clinic's day, and no-shows are the second. This recipe removes both. A patient picks a provider, date, and slot on one form. The appointment lands on the provider's Google Calendar with the patient invited, and a confirmation goes out by email and SMS within seconds. Then a daily job finds tomorrow's appointments and texts every patient a reminder, which is the single most reliable thing you can do to cut no-shows.

Healthcare onboarding carries compliance weight that generic HR templates miss. This recipe collects the hire's credential documents on day one, then has AI build a checklist specific to their role and department, the mandatory items every hire needs plus the ones only their role needs. That checklist becomes a real Trello card your team ticks off, orientation is booked for their start date, the hire is welcomed, and HR is told it's all underway.

Service-quality feedback in a clinic usually arrives too late to act on, or arrives as a public review. This recipe asks first. When staff mark an appointment Completed, the patient gets a short post-visit survey. AI classifies the sentiment and sorts comments into doctor experience, staff service, facilities, waiting time, billing, or suggestions. Low scores reach the clinic manager in seconds, with the department and the comments, and no patient identity. Leadership gets the trend report weekly. This is service-quality feedback for staff, not clinical triage and not an emergency channel.

Patients leave a consultation holding a piece of paper they can't read. This recipe turns that photo into something they can. TinyVision reads the prescription image directly and pulls the medication names, dosages, frequencies, and durations. AI then explains them in plain language and adds general education per medication: what it's for, how it's usually taken, common precautions and side effects. It only restates what was extracted. It never diagnoses, never changes a dose, and never contradicts the prescription. Every email carries a prominent not-medical-advice disclaimer, and an unreadable photo is sent back for a retake rather than guessed at.

The recipe watches your shared support inbox. Every new email is read by AI, sorted into one of six categories, scored for priority and sentiment, and logged as a ticket in TinyTables. The questions you answer constantly get a draft reply pulled from your own knowledge base; everything else routes to the right team with the context attached. Your dashboard updates itself as tickets open, move, and close.

Customers report bugs through a form. AI reads the description and the attachments, categorizes the issue, estimates its severity, links it to anything similar already reported, and writes a summary for engineering. It's assigned to the right team automatically, with that summary attached.

Feedback comes in through a form, or gets imported from the places you already collect it. AI classifies each response, reads the sentiment, groups the recurring themes, and routes it to the team that can actually act: Product, Engineering, Support, or your marketing testimonial pile. At the end of every week it writes the trend report for you.

Customers submit feature requests through a form. AI groups the duplicates, scores each idea by how many customers actually asked for it, and hands your product manager a ranked report every week. When a request moves to In Progress or Released, every customer who asked for it hears about it, without anyone drafting an email.

Two things go wrong with launch checklists: the template is either too heavy for a minor release or too thin for a major one, and the go/no-go happens in someone's head. This recipe fixes both. AI generates a checklist scaled to the launch tier and scoped to the channels you actually selected, grouped by workstream with suggested days-before-launch. Then a daily sweep watches every in-progress launch and, once it enters the readiness window, demands an explicit human go or hold.

A prospect fills in the demo request form. They get a Calendly link straight away, sales gets notified, and the moment they book, TinyCommand matches the booking back to their original request and updates everything itself. If they go quiet, the reminders go out on their own.

Automatically log every form submission to a Google Sheet. Receive data via webhook, format it into a row, and append it to your spreadsheet in real time.

Receive leads via webhook, score them with TinyGPT against your ideal customer profile, then route hot leads to Slack and create contacts in your CRM. Cold leads are silently logged.

Receive incoming leads via webhook, format the lead details, and instantly post a notification to your Slack channel. Never miss a new lead again.

AI-powered support ticket classification and routing. Incoming tickets are analyzed by TinyGPT to determine category and priority, then routed to the appropriate Slack channel, urgent tickets go to #support-urgent, everything else to #support-general.

The most widely used form for collecting sales inquiries, demo requests, and business leads. Capture name, email, phone, company, interest, and message, then connect it to your CRM or notifications later.

One of the most popular templates for gathering satisfaction ratings, product feedback, and improvement ideas. Star rating, likelihood to return, focus areas, and open comments.

A clean waitlist and newsletter signup form. Collects email, name, and content interests, then automatically adds the subscriber to your mailing list in Mailchimp or Brevo.

Measure customer loyalty with a classic Net Promoter Score survey. Ask the 0–10 recommend question, capture the reason, and collect improvement ideas.

GPT or Claude reads each new ticket and tags it with intent (bug, billing, feature-request, account). Routes to the right group automatically.

After a Calendly meeting ends, AI drafts a follow-up email referencing what was discussed (from Zoom transcript). Saves as Gmail draft for human review.

GPT or Claude reads every email and posts a 3-bullet summary to a Slack channel. Stop reading every thread.

When you add or update a row in Airtable, push the email + merge fields to Mailchimp. Respects consent and avoids duplicates.

Combine Calendly + HubSpot + AI to send the meeting host a one-page brief in Slack 15 minutes before each call.

Every Calendly booking creates a Zoom meeting. After the call, the recording is auto-saved to Google Drive with an AI summary as the file description.

When someone joins your Discord server, send a personalized welcome DM, add them to your Mailchimp audience (with consent), and notify mods of high-value joins.

When you cut a release in GitHub, list all PRs since the last release. AI rewrites them as user-facing bullets. Publish to your docs site + email subscribers.

Keep your eng tracker in lockstep with your repo. Every PR opens a Linear issue and pings the right Slack channel.

A real-time bridge from your inbox to your team channel — filtered to only the messages that matter.

When a file is added to a Drive folder, create a Notion page with title, file URL, and AI-generated summary. Useful for content libraries.

When a HubSpot deal closes won, create a row in Airtable's Customers table + sub-rows in Onboarding Tasks. CSM works the project from Airtable.

When support tags an Intercom conversation as 'bug', create a Linear issue with full chat history + customer tier. Notify reporter when fix ships.

When a Jira sprint closes, AI summarizes completed work, flagged issues, and team highlights. Emails to stakeholders + posts to Slack.

New leads get enriched with company data, then routed to the right CRM list, sales rep, or onboarding flow based on size.

When a ticket gets the 'needs-engineering' tag, create a Linear issue with full context. Notify the customer (via Zendesk) when fix ships.

When a customer is created or updated in Shopify, push to Mailchimp with the right tag (first-time buyer, repeat, VIP). Drives lifecycle campaigns.

When a Zoom recording finishes, Whisper transcribes, Claude summarizes with sections (decisions, action items), and the result lands in Notion.

Run analytics in Airtable, present in Notion. Records stay in sync without copy-paste.

When a page in your Notion Changelog database flips to Published status, post the content to a target Slack channel with formatting preserved.

Form submission with a brief triggers: AI outline → AI draft → AI title variants → save to Notion for human review. Cuts first-draft time from 4 hours to 4 minutes.

Compute weekly MRR from Stripe. When you cross a milestone ($10K, $100K, $1M), post a celebratory message to your team Slack.

When a Zoom recording is ready, post the link + AI summary in the same Slack channel that scheduled the meeting. Async-team workflow.

During launch day, every comment, upvote, and ranking change posts to a Slack channel so the team can engage in real time.

When a PR opens, fetch the diff. AI reviews for bugs, security issues, and clarity. Posts findings as PR comments. Humans still own merge decisions.

When a Salesforce Opportunity hits Closed Won, create an Asana project from a template with CSM assigned. Mirror project progress back to SFDC.

When inventory for a tracked SKU drops below reorder threshold, alert Slack + create a Trello card on the reorder board.

Run your shop with a celebratory channel. Every Shopify order posts to Discord with the customer, total, and product list.

Every morning at 9am, an AI composes a 3-sentence summary of yesterday's signups, sales, and tickets — posted to a team channel.

When a Stripe charge fails, AI drafts a recovery email referencing the customer's history. Sends via Gmail with the payment-update link. Logs to ops dashboard.

Replace Stripe's default receipt with a branded Gmail message that includes a personal note and your team's signature.

When a form is submitted, append a row in Airtable with each answer mapped to a column. AI-enrich free-text answers (sentiment, theme).

Stop pasting form responses. Every Typeform submission upserts a HubSpot contact tagged with the form name.

Every Friday, aggregate the week's shipped GitHub releases, AI-rewrite as user-facing, and send via Mailchimp to your changelog subscribers.

When you publish a new YouTube video, AI drafts platform-specific social posts (LinkedIn, X, Discord) and queues them for human review + send.