Workflow template for operations and process improvement

AI tools in use: survey summary

Before the first AI plan, the owner sees what people already use: which assistants and AI features, how often, for what work, who pays, and which answers mention customer or confidential information. Built by team from an anonymous survey export, with small teams held back.

View this template on GitHub

Put this on your planBuild this in a sprint

Add it to a workflow in the FullyPowered Planner, beside the teams and roles that will run it. Book a fit call

The outcome

The person writing the company's first AI plan starts from what is already happening, not from a guess. One summary shows, team by team, which assistants and AI features people use and how often, what they use them for in their own words, and which of those tools the company pays for, which people pay for themselves and which they use free. The seats the company buys sit beside the number of answers that say they use them.

The pieces of work people name most often become a ready list of candidates for the opportunity scorecard, so the plan begins where people have already found value. Answers that mention customer, personal or confidential information are set apart for the owner, by response id, with the company's own rule quoted above them. That section lists answers from teams too small to report one by one, because it is read only by the owner and by the person who runs the work and hands it over.

Nobody is named. A team with too few answers is counted only in the company total. The summary judges no one's use and decides nothing: the owner reads it and chooses what happens next.

Who starts the work, and when

The owner, the operations lead or the office manager runs it once before the first AI plan is written, as soon as the survey closes, and again every six months with a fresh survey. The person writing the plan reads the summary, and only the owner, and the person who runs the work and hands it over, read the section of set-apart answers.

What they supply

  • The response export from your survey tool, one row per answer, with these columns: response_id, team, role, tools_used, how_often, what_for, paid_by (company, self, free), data_types_entered and what_would_help. The survey is answered without names. If a column header does not match, the workflow template stops and asks which column is which.
  • The list of AI subscriptions the company pays for, from finance: each tool, its plan, the number of seats and the team they belong to.
  • The company's rule on what information may go into an assistant, if it has one.
  • The team list, with the number of people in each team.
  • The smallest group to report, which is five answers unless the company sets a larger number.

It works from the files a person supplies. It needs no connection to the survey tool, and it never asks for a password, a login code or a key.

What comes out

One summary in a fixed order: each team's tools and how often they are used, the company total, tools paid for by the company against tools people pay for themselves or use free, seats against answers (counts only), the pieces of work in people's own words, the answers set apart for the owner, the candidates for the opportunity scorecard, what people said would help, and the teams with too few answers to report. The counts are checked against each other so a missing row shows before anyone relies on the page.

Names that appear in people's own words are removed, and the role column is never printed. Every section is printed even when it is empty.

An example

Illustrative example with sample data. The practice, its teams, the tools, the answers and the dates below are all made up for this example. The practice is an illustrative 30-person architecture and interiors practice whose operations lead ran a short anonymous survey before the owner writes the first AI plan. The assistants are called Assistant A, B and C and the meeting notes feature Meeting notes D, so that no real product's name stands in for sample data. Nothing here describes a real organization or a measured result. First come the inputs the operations lead supplied, in full, then the summary the workflow template returned.

The inputs

1. Settings

Company: an illustrative architecture and interiors practice
Survey closed: Friday 13 November 2026
Summary goes to: the operations lead, for the owner
Smallest group reported: 5 answers

2. The response export from the survey tool

response_id,team,role,tools_used,how_often,what_for,paid_by,data_types_entered,what_would_help
R01,Design,Architect,Assistant A; Assistant B,daily; weekly,Drafting design narratives for planning applications; rewording client emails,company; free,internal documents; customer information,A shared set of instructions for the design narrative
R02,Design,Interior designer,Assistant A,weekly,Writing specification notes from my sketches list; rewording client emails,company,internal documents,Training on what we may put in
R03,Design,Architect,Assistant B,daily,Drafting design narratives; summarising planning policy documents,self,public information; internal documents,A company seat for Assistant B
R04,Design,Technician,none,,,,none,Someone to show me what it is good for
R05,Design,Architect,Assistant A; Meeting notes D,weekly; weekly,Notes from client meetings; drafting design narratives,company; company,customer information; confidential information,Clear rules on client drawings
R06,Design,Interior designer,Assistant A,monthly,Rewording client emails,company,customer information,Training on what we may put in
R07,Project delivery,Project manager,Assistant A,daily,Writing meeting minutes from my notes; drafting the weekly client progress report,company,customer information; internal documents,A template for the progress report
R08,Project delivery,Project manager,Assistant A; Assistant C,weekly; rarely,Drafting the weekly client progress report; checking contractor quotes against the scope,company; free,confidential information; internal documents,Know which assistant we should use
R09,Project delivery,Site coordinator,Assistant C,weekly,Writing up site meeting minutes for Sam,self,internal documents,A company seat
R10,Project delivery,Project manager,Meeting notes D,weekly,Notes from client meetings,company,customer information,A template for the progress report
R11,Project delivery,Project coordinator,none,,,,none,Know which assistant we should use
R12,Finance,Bookkeeper,Assistant B,weekly,Explaining spreadsheet formulas; drafting payment reminder emails,self,customer information; financial figures,A company seat for Assistant B
R13,Finance,Accounts assistant,Assistant C,monthly,Drafting payment reminder emails,free,customer information,Training on what we may put in
R14,Finance,Finance manager,none,,,,none,Know which assistant we should use
R15,Office,Office manager,Assistant B; Meeting notes D,daily; weekly,Drafting the staff newsletter; writing up the minutes of staff meetings,self; company,personal information; internal documents,A company seat for Assistant B
R16,Office,Receptionist,Assistant B,rarely,Rewording client emails,free,customer information,Someone to show me what it is good for

3. The subscription list from finance

tool,plan,seats,team
Assistant A,Team plan,8,Design
Assistant A,Team plan,4,Project delivery
Meeting notes D,Business plan,6,company-wide

4. The team list

team,headcount
Design,14
Project delivery,9
Finance,4
Office,3

5. The company's rule on information

Client drawings, contracts, fee figures and client emails go only into
Assistant A on the company's account. Nobody's personal details go into
any assistant.

The output

AI TOOLS IN USE: SURVEY SUMMARY, an illustrative architecture and interiors practice, survey closed Friday 13 November 2026

  • For: Operations lead, for the owner. Prepared from: the response export (16 rows), the subscription list (3 lines), the team list (4 teams), the company's rule on information (supplied). Smallest group reported: 5 answers.
  • Answers: 16 of 30 people. Teams reported: 2. Teams held back: 2. The survey was answered without names. Every figure is a count of answers, not of everyone in a team.
  • Check: 16 rows = 16 read + 0 unreadable.

1. By team

Design (6 answers of 14 people)

  • In use: Assistant A: 4 (daily 1, weekly 2, monthly 1, rarely 0). Assistant B: 2 (daily 1, weekly 1, monthly 0, rarely 0). Meeting notes D: 1 (daily 0, weekly 1, monthly 0, rarely 0). No tool: 1.
  • Paid by: company 5, self 1, free 1 (7 tool mentions).

Project delivery (5 answers of 9 people)

  • In use: Assistant A: 2 (daily 1, weekly 1, monthly 0, rarely 0). Assistant C: 2 (daily 0, weekly 1, monthly 0, rarely 1). Meeting notes D: 1 (daily 0, weekly 1, monthly 0, rarely 0). No tool: 1.
  • Paid by: company 3, self 1, free 1 (5 tool mentions).

Teams held back: Finance, Office, counted in the company total only (section 9).

2. Company total (16 answers of 30 people)

  • In use: Assistant A: 6 (daily 2, weekly 3, monthly 1, rarely 0). Assistant B: 5 (daily 2, weekly 2, monthly 0, rarely 1). Assistant C: 3 (daily 0, weekly 1, monthly 1, rarely 1). Meeting notes D: 3 (daily 0, weekly 3, monthly 0, rarely 0). No tool: 3. Answers naming at least one tool: 13.
  • Paid by: company 9, self 4, free 4 (17 tool mentions).

3. Paid for by the company, by people themselves, or free

  • Assistant A: company 6, self 0, free 0. On the subscription list: yes.
  • Assistant B: company 0, self 3, free 2. On the subscription list: no.
  • Assistant C: company 0, self 1, free 2. On the subscription list: no.
  • Meeting notes D: company 3, self 0, free 0. On the subscription list: yes.
  • Said to be paid by the company but not on the subscription list: none.
  • On the subscription list but named in no answer: none.

4. Seats against answers (counts only)

  • Assistant A, Team plan, Design: 8 seats; 4 answers in that team say they use it on the company's account.
  • Assistant A, Team plan, Project delivery: 4 seats; 2 answers in that team say they use it on the company's account.
  • Meeting notes D, Business plan, company-wide: 6 seats; 3 answers across the company say they use it on the company's account.
  • 16 of 30 people answered, so a seat with no answer against it may belong to someone who did not answer.

5. Pieces of work, in people's own words

  • Meeting minutes and notes: 5 answers (R05, R07, R09, R10, R15). Teams: Design, Project delivery; also teams held back. Their words: "Notes from client meetings"; "Writing meeting minutes from my notes"; "Writing up site meeting minutes for [name removed]"; "writing up the minutes of staff meetings".
  • Rewording client emails: 4 answers (R01, R02, R06, R16). Teams: Design; also teams held back. Their words: "rewording client emails".
  • Drafting design narratives: 3 answers (R01, R03, R05). Teams: Design. Their words: "Drafting design narratives for planning applications"; "Drafting design narratives".
  • Drafting payment reminder emails: 2 answers (R12, R13). Teams: teams held back only. Their words: "drafting payment reminder emails".
  • Drafting the weekly client progress report: 2 answers (R07, R08). Teams: Project delivery. Their words: "drafting the weekly client progress report".
  • Checking contractor quotes against the scope: 1 answer (R08). Teams: Project delivery. Their words: "checking contractor quotes against the scope".
  • Drafting the staff newsletter: 1 answer (R15). Teams: teams held back only. Their words: "Drafting the staff newsletter".
  • Explaining spreadsheet formulas: 1 answer (R12). Teams: teams held back only. Their words: "Explaining spreadsheet formulas".
  • Summarising planning policy documents: 1 answer (R03). Teams: Design. Their words: "summarising planning policy documents".
  • Writing specification notes: 1 answer (R02). Teams: Design. Their words: "Writing specification notes from my sketches list".

6. Set apart for the owner: customer, personal or confidential information

For the owner only; remove before sharing.

The company's rule, as supplied: "Client drawings, contracts, fee figures and client emails go only into Assistant A on the company's account. Nobody's personal details go into any assistant."

  • R01, Design. Assistant A (company); Assistant B (free). Ticked: internal documents, customer information. Their words: "Drafting design narratives for planning applications; rewording client emails".
  • R02, Design. Assistant A (company). Ticked: internal documents. Not ticked; the answer mentions "client emails". Their words: "Writing specification notes from my sketches list; rewording client emails".
  • R05, Design. Assistant A (company); Meeting notes D (company). Ticked: customer information, confidential information. Their words: "Notes from client meetings; drafting design narratives".
  • R06, Design. Assistant A (company). Ticked: customer information. Their words: "Rewording client emails".
  • R07, Project delivery. Assistant A (company). Ticked: customer information, internal documents. Their words: "Writing meeting minutes from my notes; drafting the weekly client progress report".
  • R08, Project delivery. Assistant A (company); Assistant C (free). Ticked: confidential information, internal documents. Their words: "Drafting the weekly client progress report; checking contractor quotes against the scope".
  • R10, Project delivery. Meeting notes D (company). Ticked: customer information. Their words: "Notes from client meetings".
  • R12, team held back. Assistant B (self). Ticked: customer information, financial figures. Their words: "Explaining spreadsheet formulas; drafting payment reminder emails".
  • R13, team held back. Assistant C (free). Ticked: customer information. Their words: "Drafting payment reminder emails".
  • R15, team held back. Assistant B (self); Meeting notes D (company). Ticked: personal information, internal documents. Their words: "Drafting the staff newsletter; writing up the minutes of staff meetings".
  • R16, team held back. Assistant B (free). Ticked: customer information. Their words: "Rewording client emails".

11 answers set apart.

7. Candidates for the opportunity scorecard

  1. Meeting minutes and notes: named in 5 answers. Teams: Design, Project delivery; also teams held back.
  2. Rewording client emails: named in 4 answers. Teams: Design; also teams held back.
  3. Drafting design narratives: named in 3 answers. Teams: Design.
  4. Drafting payment reminder emails: named in 2 answers. Teams: teams held back only.
  5. Drafting the weekly client progress report: named in 2 answers. Teams: Project delivery.

The owner scores and chooses.

8. What people said would help

  • A company seat: 4 answers. Teams: Design, Project delivery; also teams held back.
  • Clear rules and training on what may go in: 4 answers. Teams: Design; also teams held back.
  • Know which assistant to use: 3 answers. Teams: Project delivery; also teams held back.
  • Shared instructions or templates for a piece of work: 3 answers. Teams: Design, Project delivery.
  • Someone to show what it is good for: 2 answers. Teams: Design; also teams held back.

9. Teams with too few answers to report

  • Finance (4 people): fewer than 5 answers. Counted in the company total only.
  • Office (3 people): fewer than 5 answers. Counted in the company total only.
  • Answers with no team: 0.

10. Data gaps and questions

  • none

Draft for the operations lead, for the owner. Nobody is named, no use is judged, and nothing has been sent, bought, cancelled or changed.

Where it fits

  • Operations and process improvement, where the person who runs the business plans which work AI should prepare.
  • Professional services, where people pick up assistants on their own to draft client emails, notes and reports long before there is a plan.
  • Law firms, where the firm administrator needs to know which assistants touch client matters before the partners agree what goes where.
  • Marketing and advertising, where every team already has a favourite AI tool and the agency pays for some of them.
  • Established businesses, where AI use has grown quietly, one person at a time.
  • Growth & scaling, when new hires bring their own AI tools and the company wants one plan.

It is the look before step one, plan it, in how organizations plan their use of AI.

It is a different piece of work from software subscription and vendor spend review, which looks at what every tool costs and what renews, and from culture health check, which reads whether the company's values show up in daily work. This one looks at what people do with AI today. Workflow fit check then tests one candidate piece of work before anything is built.

What has to change for your company

Write the survey once, with the columns above, and keep it anonymous: no name or email question. Decide the smallest group you will report; five is the default, and a larger number protects people in small teams further. Bring the team list up to date on the day the survey closes. Write your rule on what may go into an assistant in a sentence or two, if you have one. Decide who reads the summary and who reads the set-apart answers; remove that section before the summary goes any further. It runs in ChatGPT, Claude, Copilot or Gemini, from the exports you give it.

One risk to weigh before you share the summary beyond the owner: in the section of pieces of work, a piece named only by a team too small to report, such as a task only the finance team does, can point to that team even though the team is not named. Remove or generalize those quoted words before the summary goes further, or share only the counts. The summary gives no legal or security opinion and recommends no product. Whether a use fits your rule, and what follows, is for the owner and the person who answers for the company's obligations.

What has been checked

FullyPowered AI wrote this workflow template. It was editorially reviewed on 19 November 2026. It was tried on its example on 19 November 2026 in one run, published with it. Edits after that run, to how ties are ordered, how repeated phrasings are counted, how small teams are held back, who reads the set-apart section, the order of tools, how the company's rule is quoted and how first names are treated, have not been run again. It has not yet been tested in a client environment, and it has not yet been tested as a packaged version in ChatGPT, Claude, Copilot or Gemini.

Testing in your environment means running it on a past survey whose answers the operations lead already knows, with cases written in advance: a team with three answers, a column header that does not match, an answer that mentions client emails without ticking customer information, a name typed into an answer. The owner checks every count against the export before anyone relies on the summary. Pilot acceptance-test preparation sets those cases up.

Who decides

People decide

Your people keep

  • What happens next, chosen by the owner
  • Whether a use fits the company's rule, and what follows
  • Which candidates go on the opportunity scorecard
By design

It does not

  • Names nobody
  • Judges no one's use
  • Gives no legal or security opinion
  • Recommends no product
  • Needs no connection to the survey tool

Questions people ask

Who is this for?

The owner, the operations lead or the office manager, before the company's first AI plan is written and again every six months with a fresh survey.

What do I need to supply?

The response export from your survey tool, the list of AI subscriptions the company pays for, the team list, and your rule on which information goes into an assistant, if you have one.

Does it identify who answered?

No. The survey is answered without names, names in people's own words are removed, the role column is never printed, and a team with too few answers is counted only in the company total.

Does it say whether someone broke our rule?

No. Answers that mention customer, personal or confidential information are set apart for the owner with the rule quoted above them. The summary judges no one's use and gives no legal or security opinion.

What has been checked?

It was editorially reviewed and tried once on its example, with the run published alongside it. Edits made after that run have not been run again, and it has not yet been tested in a client environment.

The full workflow template

Everything above is the outline. The complete workflow template is the working version: the roles, the inputs, the procedure, the exceptions and stop conditions, the output format, the reviewer checklist, the acceptance tests, the instructions for an AI assistant and what to customize for your company. It comes with a download and the link to its folder in FullyPowered’s public repository.

View the full template

Free with an account. Sign in by email: no password and no card.

Next step

Put this workflow template to work for your team: on your plan, or built with you in a sprint.

  1. 01

    The full workflow template

    Free with an account. Sign in by email: no password and no card.

  2. 02

    FullyPowered Planner

    Add it to a workflow in the FullyPowered Planner, beside the teams and roles that will run it.

  3. 03

    Build this in a sprint

    In a sprint, FullyPowered adapts this workflow template to your rules, proves it on your own work and puts it in the hands of the roles that own it.

These workflow templates are examples and a starting point, not a complete list. We can build any workflow for your company, including one that runs several steps, in a Workflow Sprint or a scoped implementation.

Rather talk it through first? Book a fit callDiscuss adapting this

FullyPowered Planner. A card to start, everything unlocked, nothing charged for seven days.Security & your informationPlanner termsRefunds

See more: AI workflow examples for law firms