Skip to content
Tallyfy

Import a document as a template

Turn any document into a working workflow

Got a messy SOP sitting in a Word doc? A checklist buried in a PDF? Paste the text into your AI and it builds a clean Tallyfy template - complete with steps, form fields, assignments, and descriptions - from whatever you give it.

No cleanup required. The AI reads the mess and figures out what goes where.

What you’ll accomplish

  • Convert disorganized process documents into structured Tallyfy templates
  • Extract steps, form fields, and role assignments from messy source text
  • Skip the manual template-building process entirely

Try it now

Setup: If you haven’t connected Claude to Tallyfy yet, follow the connect your AI to Tallyfy guide first - it takes about two minutes.

Prompt to try:

Here's our employee onboarding SOP from a Word doc - it's messy and outdated but has the core steps. Create a clean Tallyfy template from it with proper steps, form fields, and descriptions.


EMPLOYEE ONBOARDING PROCEDURE
Last updated: 2019


When new hire starts, HR should collect all paperwork (tax forms, ID copy, emergency contacts, bank details for payroll).


IT needs to set up their laptop, email account, and software access. Usually takes 2-3 days. They need to know what department and what software the person needs.


Manager does orientation meeting - covers team intro, role expectations, first week goals. Should document the goals somewhere.


Buddy system - assign someone from the team to help them for first 2 weeks. Buddy should check in daily.


After 30 days, manager does first review. Need to document: performance so far, any concerns, goals for next 60 days.


HR closes out onboarding after 90-day review is complete.

What happens: Claude reads the document and identifies six distinct steps - HR paperwork collection, IT setup, manager orientation, buddy assignment, 30-day review, and 90-day close-out. It calls get_all_templates to check whether a similar template already exists in your Tallyfy account. Then it creates the template and builds each step with a clear description drawn from your source text. For each step, it calls suggest_form_fields_for_step to recommend fields that match the step’s content (text fields for employee name, a dropdown for department, checkboxes for paperwork items). Finally, it calls add_assignees_to_step to assign roles based on who your document names - HR, IT, Manager - so the right people get notified when each step is active.

How it works behind the scenes

The AI works through your document in a logical sequence - and understanding each step helps you write better prompts.

Step 1: Check for duplicates

Before creating anything, the AI calls get_all_templates to scan your existing Tallyfy templates. If something called “Employee Onboarding” already exists, it’ll flag that and ask whether to update the existing template or create a new one. No accidental duplicates.

Step 2: Parse the document

The AI reads your text looking for natural boundaries - paragraph breaks, numbered items, phrases like “then,” “next,” or “after that.” Each distinct phase of your process becomes a candidate step. With the example document above, the AI identifies six steps even though the text isn’t formatted as a list: the HR paperwork phase, the IT setup phase, the orientation meeting, the buddy assignment, the 30-day review, and the 90-day close-out.

Step 3: Create the template and add steps

The AI creates the template in Tallyfy and calls add_step_to_template for each identified step. Then it calls edit_description_on_step for each one, writing clear instructions drawn from your source text. “HR should collect all paperwork (tax forms, ID copy, emergency contacts, bank details for payroll)” becomes a proper step description - specific, actionable, and stripped of the vague phrasing from the original doc.

Step 4: Suggest and add form fields

This is where the AI earns its keep. For each step, it calls suggest_form_fields_for_step to analyze what data needs to be captured. The IT setup step mentions needing to know “what department and what software” - that becomes a department dropdown and a multi-line text field for software requirements. The orientation step says “should document the goals somewhere” - that becomes a text field for first-week goals. Then add_form_field_to_step adds each suggested field to the right step.

Step 5: Assign roles

The AI reads role references in your document - “HR should,” “IT needs to,” “Manager does” - and calls add_assignees_to_step to set assignments. If your Tallyfy account has job titles or groups matching those roles, the AI uses those. If the names in the document don’t match anything in your organization, it’ll ask before assigning.

What the AI does with messy formatting

Disorganized documents aren’t a problem. The AI handles all of these without you needing to clean anything up first:

Prose paragraphs: Most old SOPs aren’t bulleted lists - they’re paragraphs of instructions. The AI reads them the same way a human would, picking out discrete actions and responsibilities.

Mixed formats: Some sections might be bulleted, others numbered, others just paragraphs. The AI handles inconsistency fine.

Implied information: Phrases like “collect all paperwork” imply a checklist even if one isn’t provided. The AI picks up on these and suggests appropriate form fields.

Vague assignees: “Someone from the team” for the buddy role is ambiguous. The AI will note the ambiguity and either ask you to clarify or suggest a group assignment so any team member can take it.

Outdated instructions: Your 2019 SOP might reference tools that no longer exist or processes that have changed. The AI builds the template from what’s written - it won’t silently update outdated content, but it won’t refuse to work with it either. You can ask it to modernize specific steps if you want.

Missing details: If a step in your document is thin on specifics, the AI writes what it can from context and may flag that the description could use more detail. It doesn’t fabricate instructions that aren’t in your source.

Tips for better results

Paste the full document text, not a summary. The AI works better with more context. If your Word doc is ten pages, paste all ten pages. Summaries lose the specific details that make form fields and descriptions useful.

Mention the step types you want. If some steps should be approvals rather than standard tasks - say, the 90-day review needs sign-off from HR leadership - include that in your prompt: “Make the 90-day review step an approval step.”

Call out parallel steps. If some steps should run at the same time rather than sequentially, say so: “The IT setup and paperwork collection happen simultaneously - they shouldn’t be blocked on each other.” Tallyfy supports parallel steps and the AI will configure that if you ask.

Include related forms or checklists. If your document references a separate checklist (a new hire paperwork checklist, a software access request form), include that text too. The AI uses it to build more accurate form fields.

Ask the AI to list what it found before building. If you’re not sure your document is structured clearly enough, add “Before building the template, list the steps you identified and ask me to confirm” to your prompt. You get a preview and a chance to correct any misreadings before anything is created in Tallyfy.

Specify your role names. If your document says “Manager” but your Tallyfy account uses a job title called “Department Head,” mention that: “When the document says Manager, use the ‘Department Head’ job title in Tallyfy.” This avoids a round of clarification questions mid-build.

Take it further

Templates

How templates work in Tallyfy and how to create and configure them for your team.

Form fields

How form fields work in Tallyfy and what types are available for collecting data in tasks.

BYO AI

Learn about the bring-your-own-AI integration and what it enables across your Tallyfy workflows.
Get in touch
About Tallyfy